Georgia State UniversityScholarWorks @ Georgia State UniversitySociology Dissertations Department of SociologySpring 5-5-2012Self-Concepts of Homeless People in an UrbanSetting: Processes and Consequences of theStigmatized IdentityJosie L. ParkerGeorgia State UniversityFollow this and additional works at: https://scholarworks.gsu.edu/sociology_dissThis Dissertation is brought to you for free and open access by the Department of Sociology at ScholarWorks @ Georgia State University. It has beenaccepted for inclusion in Sociology Dissertations by an authorized administrator of ScholarWorks @ Georgia State University. For more information,please contact scholarworks@gsu.edu.Recommended CitationParker, Josie L., "Self-Concepts of Homeless People in an Urban Setting: Processes and Consequences of the Stigmatized Identity."Dissertation, Georgia State University, 2012.https://scholarworks.gsu.edu/sociology_diss/65 SELF-CONCEPTS OF HOMELESS PEOPLE IN AN URBAN SETTING: PROCESSES AND CONSEQUENCES OF THE STIGMATIZED IDENTITY by JOSIE LEIGH PARKER Under the Direction of Donald C. Reitzes ABSTRACT This study investigates social psychological strategies homeless persons use to develop and maintain the self while homeless. To understand this topic, I apply the identity theory of Stryker, self-esteem of Rosenberg, self-efficacy of Gecas and Schwalbe, and homeless identity meanings and behaviors of Burke. Additionally, I examine what is needed to no longer be homeless. In all, 326 surveys were collected at six different homeless service agencies such as shelters and meal sites. The data analysis includes descriptive statistics and multivariate regression. The results only partially support identity theory in that interactive commitment (increased number of homeless friends) predicts salience (frequently invoking the homeless identity across different situations) which predicts increased length of time in role. However, affective commitment and centrality of the homeless identity have no effect. This study does confirm Snow and Anderson’s findings that homeless persons on the streets for a shorter period of time will distance themselves from the homeless identity, while those on the streets longer will embrace the homeless identity. As opposed to previous research, I find that the majority of homeless respondents do not have low self-esteem or self-efficacy. Instead it is certain factors such as being homeless longer and more often, accepting the homeless identity, viewing the homeless identity as most important, little to no family support and having a high school diploma (or less) that result in homeless persons having low self-esteem or self-efficacy. With homeless identity meanings, people thinking negatively about themselves is the result of having more homeless friends, being homeless longer and more often, possessing low self-esteem and low self-efficacy. Placing great importance on homeless identity behaviors such as helping other homeless people and staying sober influence these outcomes: thinking positively about the self, stronger ties with other homeless people, more homeless friends and invoking the homeless identity more often in different situations. For homeless people to obtain housing, two factors, income and social support systems, are most important. Of all the control variables, sleeping on the streets and multiple disabilities demonstrate the greatest impact for almost all of the independent variables. The implications of these findings are discussed. INDEX WORDS: Homelessness, Urban, Poverty, Social psychology, Stigmatized identity, Identity salience, Identity centrality, Identity commitment, Identity meanings, Identity behaviors, Self-esteem, Self-efficacy, Disabilities SELF-CONCEPTS OF HOMELESS PEOPLE IN AN URBAN SETTING: PROCESSES AND CONSEQUENCES OF THE STIGMATIZED IDENTITY by JOSIE LEIGH PARKER A Dissertation Submitted in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy in the College of Arts and Sciences Georgia State University 2012 Copyright by Josie Leigh Parker 2012 SELF-CONCEPTS OF HOMELESS PEOPLE IN AN URBAN SETTING: PROCESSES AND CONSEQUENCES OF THE STIGMATIZED IDENTITY by JOSIE LEIGH PARKER Committee Chair: Donald C. Reitzes Committee: Charles Jaret Ralph LaRossa Electronic Version Approved: Office of Graduate Studies College of Arts and Sciences Georgia State University May 2012 iv DEDICATION To my mother and fatherv ACKNOWLEDGMENTS The completion of my dissertation has been a transformative experience. Through this learning process, I have come to know myself and the world around me better. It has been an undertaking that I could not have completed without the help and support of some very important people in my life. Thanks first to Gary Batchelor, my significant other. He has stood by my side during the stressful times and the exciting times. With his help, I have let go of my past to look forward to my future. My family has always encouraged me to pursue my dreams. I am thankful to my mother for all that she sacrificed so that I could succeed in this life. She and my father are deeply missed. While working on my dissertation, I have appreciated the support that my siblings, Tony, Sallie and Cindy, and their families have provided me. I wish to express my gratitude to my dissertation chair, Donald Reitzes, and my committee members, Charles Jaret and Ralph LaRossa. Don has been a wonderful mentor, instrumental in my development as a sociologist and social psychologist, researcher and writer. Thanks also to both Charlie and Ralph for all their time and effort in providing their expertise and guidance. During this process, I am grateful to Tim Crimmins at the GSU Center for Neighborhood and Metropolitan Studies, William Matson at Pathways and Mindy Stombler in the sociology department at Georgia State University for the opportunities they gave me to practice in my field. I am also thankful to all my professors at Georgia State University from whom I gained the knowledge and skill set needed to become a professional sociologist. vi Of course, I could not have survived this whole graduate school process without my fellow graduate students. They became my friends and confidants as we surpassed and enjoyed all the stages and challenges of our doctoral program. Finally, the dream of obtaining my doctorate was set many years ago in Tampa and I am grateful for all the support and encouragement that I received from my friends to pursue this tremendous accomplishment.vii TABLE OF CONTENTS ACKNOWLEDGEMENTS v LIST OF TABLES viii CHAPTER 1 INTRODUCTION 1 CHAPTER 2 CONDITIONS OF HOMELESSNESS 9 CHAPTER 3 STRATEGIES OF THE HOMELESS 20 CHAPTER 4 SELF-CONCEPT OF HOMELESS PERSONS 30 CHAPTER 5 METHODOLOGY 49 CHAPTER 6 CHARACTERISTICS OF THE HOMELESS SAMPLE 64 CHAPTER 7 RESULTS FOR IDENTITY THEORY, ACCEPTING 110 THE HOMELESS IDENTITY AND EXITING HOMELESSNESS CHAPTER 8 RESULTS FOR SELF-ESTEEM AND SELF-EFFICACY 135 CHAPTER 9 RESULTS FOR HOMELESS IDENTITY MEANING 161 AND BEHAVIORS CHAPTER 10 CONCLUSION AND IMPLICATIONS 200 REFERENCES 220 APPENDICES 229 viii LIST OF TABLES Table 6.1: Demographics as Compared to the 2009 City Homeless Survey: 65 Number, Frequency and Standard Deviation for Race/Ethnicity, Gender, Military Service, Age and Living with Family Members Table 6.2: Demographics: Number, Frequency and Standard Deviation 67 for Relationship Status, Parent, Education Level, Length of Time Living in Atlanta and Income for Previous Month Table 6.3: Causes of Homelessness: Number and Frequency for Economic 69 Reasons, Health Reasons, Family Reasons, Housing Reasons and Other Reasons Table 6.4: Homeless History: Number, Frequency and Standard Deviation 72 for Usually Slept at Night Over Past Month, Different Times Homeless, Current Length of Homelessness and Total Length of Homelessness within Past Three Years Table 6.5: Number, Frequency, Type and Standard Deviation of Disabilities 74 Table 6.6: Number, Frequency and Standard Deviation for Acceptance 77 of the Homeless Identity Table 6.7: Centrality to the Homeless Identity: Number, Mean and Standard 78 Deviation of respondents who ranked the role as an important identity at some level ix Table 6.8: Centrality to the Homeless Identity by Roles: Number and Frequency 79 for Family member, Worker, Friend, Relationship Status, Religious, Physically Healthy, Education Level and Homeless Table 6.9: Centrality to the Homeless Identity by Roles: Number and Frequency 82 for Race/Ethnicity, Gender, Mentally Healthy, Age, Student, Veteran, Sexual Orientation and Alcoholic/Addict Table 6.10: Salience to the Homeless Identity: Number and Frequency 84 of Respondents who Discussed the Topics Most Often Across the Different Situation Table 6.11: Salience to the Homeless Identity: Number, Frequency and Standard 85 Deviance of Homelessness, Work, Physical Health and Religion Table 6.12: Mean and Standard Deviation for Affective Commitment 87 to the Homeless Identity Table 6.13: Number and Frequency for Interactive Commitment to the 89 Homeless Identity Table 6.14: Mean and Standard Deviations for Self-Esteem 91 Table 6.15: Mean and Standard Deviations for Self-Efficacy 92 Table 6.16: Mean and Standard Deviation for Homeless Identity Meaning 95 Table 6.17: Number, Frequency and Means for Homeless Identity Behaviors 97 Table 6.18: Number, Frequency and Standard Deviation for Exiting the 99 Homeless Situation Table 6.19: Mean and Standard Deviation for Family Support 101 Table 6.20: Homeless Experience: Religious Faith 103x Table 6.21: Homeless Experience: Alcohol / Drug Abuse and Mental Health 105 Table 6.22: Homeless Experience: Learning and Bad Experiences 106 Table 6.23: Homeless Experience: Employment and Housing 107 Table 7.1: Multiple Regression of the Relationship between Interactive 112 Commitment and Salience of the Homeless Identity (H1A) Table 7.2: Multiple Regression of the Relationship between Affective 114 Commitment and Salience of the Homeless Identity (H1B) Table 7.3: Multiple Regression of the Relationship between Interactive 115 Commitment and Centrality of the Homeless Identity (H1C) Table 7.4: Multiple Regression of the Relationship between Affective 116 Commitment and Centrality of the Homeless Identity (H1D) Table 7.5: Multiple Regression of the Relationship between Salience 118 of the homeless identity and Current Length of Time Homeless (H2A) Table 7.6: Multiple Regression of the Relationship between Centrality 119 of the Homeless Identity and Current Length of Time Homeless (H2B) Table 7.7: Multiple Regression of the Relationship between Total Length 122 of Time Homeless and Acceptance of the Homeless Identity (H3A) Table 7.8: Multiple Regression of the Relationship between Number 123 of Different Times Homeless and Acceptance of the Homeless Identity (H3B) Table 7.9: Multiple Regression of the Relationship between Acceptance 125 of the Homeless Identity and Attempts to Exit the Homeless Situation (H4) xi Table 7.10: Multiple Regression of the Relationship between Interactive 127 Commitment to the Homeless Identity and Attempts to Exit the Homeless Situation (H5A) Table 7.11: Multiple Regression of the Relationship between Affective 128 Commitment to the Homeless Identity and Attempts to Exit the Homeless Situation (H5B) Table 7.12: Multiple Regression of the Relationship between Salience 129 of the Homeless Identity and Attempts to Exit the Homeless Situation (H5C) Table 7.13: Multiple Regression of the Relationship between Centrality 130 of the Homeless Identity and Attempts to Exit the Homeless Situation (H5D) Table 7.14: Multiple Regression of the Relationship between Family Support 133 and Attempts to Exit the Homeless Situation (H6) Table 8.1: Multiple Regression of the Relationship between Total Length 137 of Time Homeless and Self-Esteem (H7A) Table 8.2: Multiple Regression of the Relationship between Number 138 of Different Times Homeless and Self-Esteem (H7B) Table 8.3: Multiple Regression of the Relationship between Acceptance 140 of the Homeless Identity and Self-Esteem (H8) Table 8.4: Multiple Regression of the Relationship between Interactive 142 Commitment to the Homeless Identity and Self-Esteem (H9A) Table 8.5: Multiple Regression of the Relationship between Affective 143 Commitment to the Homeless Identity and Self-Esteem (H9B) xii Table 8.6: Multiple Regression of the Relationship between Salience 144 of the Homeless Identity and Self-Esteem (H9C) Table 8.7: Multiple Regression of the Relationship between Centrality 145 of the Homeless Identity and Self-Esteem (H9D) Table 8.8: Multiple Regression of the Relationship between Family Support 147 and Self-Esteem (H10) Table 8.9: Multiple Regression of the Relationship between Family Support 149 and Self-Efficacy (H11) Table 8.10: Multiple Regression of the Relationship between Acceptance 151 of the Homeless Identity and Self-Efficacy (H12) Table 8.11: Multiple Regression of the Relationship between Interactive 153 Commitment to the Homeless Identity and Self-Efficacy (H13A) Table 8.12: Multiple Regression of the Relationship between Affective 154 Commitment to the Homeless Identity and Self-Efficacy (H13B) Table 8.13: Multiple Regression of the Relationship between Salience 155 Commitment of the Homeless Identity and Self-Efficacy (H13C) Table 8.14: Multiple Regression of the Relationship between Centrality 156 of the Homeless Identity and Self-Efficacy (H13D) Table 8.15: Multiple Regression of the Relationship between Total Length 158 of Time Homeless and Self-Efficacy (H14A) Table 8.16: Multiple Regression of the Relationship between Number 159 of Different Times Homeless and Self-Efficacy (H14B) xiii Table 9.1: Multiple Regression of the Relationship between Safe Place 163 and Homeless Identity Meanings (H15A) Table 9.2: Multiple Regression of the Relationship between Person 164 To Trust and Homeless Identity Meanings (H15B) Table 9.3: Multiple Regression of the Relationship between Best Meal 165 Sites and Homeless Identity Meanings (H15C) Table 9.4: Multiple Regression of the Relationship between Helping Others 166 and Homeless Identity Meanings (H15D) Table 9.5: Multiple Regression of the Relationship between Telling Storied 167 and Homeless Identity Meanings (H15E) Table 9.6: Multiple Regression of the Relationship between Staying Sober 168 and Homeless Identity Meanings (H15F) Table 9.7: Multiple Regression of the Relationship between Sharing 169 Information and Homeless Identity Meanings (H15G) Table 9.8: Multiple Regression of the Relationship between Interactive 171 Commitment to the Homeless Identity and Homeless Identity Meanings (H16A) Table 9.9: Multiple Regression of the Relationship between Affective 172 Commitment to the Homeless Identity and Homeless Identity Meanings (H16B) Table 9.10: Multiple Regression of the Relationship between Salience 173 of the Homeless Identity and Homeless Identity Meanings (H16C) Table 9.11: Multiple Regression of the Relationship between Centrality 174 of the Homeless Identity and Homeless Identity Meanings (H16D) xiv Table 9.12: Multiple Regression of the Relationship between Total Length 176 of Time Homeless and Homeless Identity Meanings (H17A) Table 9.13: Multiple Regression of the Relationship between Number 177 of Different Times Homeless and Homeless Identity Meanings (H17B) Table 9.14: Multiple Regression of the Relationship between Self-Esteem 179 and Homeless Identity Meanings (H18A) Table 9.15: Multiple Regression of the Relationship between Self-Efficacy 180 and Homeless Identity Meanings (H18B) Table 9.16: Multiple Regression of the Relationship between Best Meal 182 Site and Interactive Commitment to the Homeless Identity (H19A-1) Table 9.17: Multiple Regression of the Relationship between Helping Others 184 and Interactive Commitment to the Homeless Identity (H19A-2) Table 9.18: Multiple Regression of the Relationship between Telling Stories 185 and Interactive Commitment to the Homeless Identity (H19A-3) Table 9.19: Multiple Regression of the Relationship between Staying Sober 186 and Interactive Commitment to the Homeless Identity (H19A-4) Table 9.20: Multiple Regression of the Relationship between Share 187 Information and Interactive Commitment to the Homeless Identity (H19A-5) Table 9.21: Multiple Regression of the Relationship between Helping Others 188 and Affective Commitment to the Homeless Identity (H19B-1) Table 9.22: Multiple Regression of the Relationship between Telling Stories 189 and Affective Commitment to the Homeless Identity (H19B-2) xv Table 9.23: Multiple Regression of the Relationship between Share 190 Information and Affective Commitment to the Homeless Identity (H19B-3) Table 9.24: Multiple Regression of the Relationship between Helping Others 192 and Salience of the Homeless Identity (H19C-1) Table 9.25: Multiple Regression of the Relationship between Telling Stories 193 and Salience of the Homeless Identity (H19C-2) Table 9.26: Multiple Regression of the Relationship between Staying Sober 194 and Salience of the Homeless Identity (H19C-3) Table 9.27: Multiple Regression of the Relationship between Share 195 Information and Salience of the Homeless Identity (H19C-4) Table 9.28: Multiple Regression of the Relationship between Staying Sober 197 and Self-esteem (H20A) Table 9.29: Multiple Regression of the Relationship between Staying Sober 198 and Self-efficacy (H20B) 1 CHAPTER 1 INTRODUCTION Homelessness is a pervasive social problem in the United States. On any given night, over 650,000 homeless people are found sleeping on the streets, at homeless shelters or in transitional housing programs. The U.S. Department of Housing and Urban Development (HUD) (2009) defines homeless individuals as persons without adequate, stable nighttime residence. Middle aged single men who are members of a minority group and live in urban areas comprise the largest group of people who are homeless (HUD 2009). Despite a plethora of research on the extent of homelessness in the U.S., causes of homelessness, and behaviors of homeless people, relatively little has been written about how homelessness is incorporated into people’s sense of self and the significance of homelessness as a role identity. This dissertation addresses that gap by exploring the meanings and impact homelessness has for people’s self identity. The United States began experiencing an immense growth of the homeless population during the early 1980s (Snow and Anderson 1993). A change in the socio-demographic composition of the homeless population also occurred during this time period. An increase of single unaccompanied women and families who were homeless appeared requiring new social services to meet their needs (Rossi 1990; Snow and Anderson 1993). To understand the increasing numbers and changing composition, a growing literature developed focusing on issues of homelessness. Two broad explanations emerged from the past research as reasons for homelessness. The first of these linked the problem to personal issues such as physical and mental illness (Cohen 2 2001; Goering et al. 2002; Goldfinger et al. 1999; Rosenheck 2000), while the second reason for homelessness pointed to structural factors like economics and housing (Elliott and Krivo 1991). At the time, this research (mainly conducted using homeless counts and surveys) helped define the homeless problem and determine funding possibilities for solving the issue. Even though the data on socio-demographics, numbers and reasons were important in understanding the population, many of these studies overemphasized personal characteristics and individual pathologies of homeless persons (Snow, Anderson and Koegel 1994). In response, scholars turned their attention to other aspects of the homeless population, such as the strategies used to survive while homeless (Snow and Anderson 1987). People learn techniques in dealing with their homeless situation and to get their needs met. Food, water, and sleep are basic physiological needs that must be met to survive. Obtaining these necessities can be difficult for people because on most days only a few social service agencies and/or good Samaritans are providing these basic items at a few locations throughout an entire city. Getting adequate sleep can also be a problem because in several major U.S. cities police and security officers go around waking up homeless people beginning in the early morning hours. Safety is another important necessity for survival while homeless. The 2009 local city and two counties (Tri-J) homeless survey found that over a quarter of respondents had been the victim of a violent attack since becoming homeless. In the six months prior to the Tri-J survey, almost half of the respondents had visited the emergency room (ER). The fourth most mentioned reason for going to the ER was because of being attacked or for fighting (Parker and Regus 2009). Other primary needs are social psychological in nature focusing on the requirements for emotional support, a sense of belonging and a feeling of self worth. According to the 2007 local 3 city and two counties (Tri-J) homeless survey, the majority of respondents (70 percent) stated that they were getting the emotional support that they needed, with a small number (30 percent) of them getting that support from other homeless people. Predominately, the respondents had people to rely on when they became sick (77 percent), with the majority depending on family members. In addition, over half of the respondents stated that they were in contact with their family on a weekly basis (Massey, Runkle and Parker 2008). Finally, average to high levels of self-esteem are essential for feeling confident about the self, even while homeless. Predominately, the homeless literature has centered on the physiological and belonging needs and techniques used to get them met. Snow and Anderson (1987) were among some of the first researchers to examine the social psychological strategies people use to develop and maintain the self while homeless, specifically focusing on aspects of the self concept relating to identity, meaning, and self esteem. These researchers concentrated on the personal identity, whereby meaning for the self was created during interaction. One dominant strategy for establishing the self to others was through verbal communication. Verbal construction of the self as homeless was accomplished through either embracing the homeless identity or it was resisted by distancing oneself from the homeless identity. Other strategies involved activities centered on getting needs met, life on the streets and interacting with other homeless people. According to Snow and Anderson (1987), people created identity and meaning in homelessness by actively presenting themselves to others and interacting with them, while they gained or lost self-esteem based on the perceived responses of others to their presentation of self. Thus homeless individuals either had their self-concept verified or not verified in this process. Negative feelings resulted if the image of themselves was not verified by others. 4 Much of the homeless research on social psychological aspects of life on the streets has been addressed by scholars empirically but has not been well developed theoretically. I propose filling the gap by applying Stryker’s structural identity theory (Stryker and Serpe 1994). In this theory, a person’s multiple identities are organized into a hierarchy. This hierarchy of identities is based around three important elements: salience, commitment and centrality. First, salience is the probability of invoking a certain identity in a series of situations consistently over time. The structural identity theory is thus conceived as situationally based identity as opposed to psychological theories which are personality based. Next, commitment is how attached people are to an identity. It can be measured in two ways: the number of known others who are connected to the same identity is “interactive commitment,” and the stronger the ties attached to the identity is “affective commitment.” Finally, centrality is how important the identity is to individuals. Identities that people possess are tied to the social positions or categories in society that they occupy in their everyday lives (James 1890). Common examples of established social positions that people occupy are parent, spouse, student and worker. Other social categories that are not regularly held by people, and thus are seen as not part of the normal life trajectory, include addict, prostitute and homeless. To know how to act in the social category, there are behavioral expectations attached to the positions defined as roles. For example, students are expected to attend classes, read books, write papers, and study for exams. Along with the activities attached to a role, there are also the meanings that people impute upon the role, such as students being people who learn and gain knowledge (Burke and Stets 2009). For those persons with identities that are not seen as part of the traditional life trajectory, and thus their role behavior does not meet set expectations, they are often viewed negatively by 5 others and labeled as stigmatized (Goffman 1963). Much of the past research has investigated role identities based on the traditional life trajectory such as student (Burke and Reitzes 1981; Stryker and Serpe 1994). Instead, this study examines the non-traditional and stigmatized role identity of homelessness. In a similar vein, most self-esteem research has also focused on assessing traditional role identities. The premiere self-esteem scale, developed by Rosenberg (1979), was tested using high school students. For Rosenberg, self-esteem was the assessment people make of their overall personal worth based on the opinion of others, comparison to others, and a self assessment of successes versus failures. Based on findings from his student testing, Rosenberg viewed people with a negative role identity as having a lower self-esteem. Goffman (1963) and Gecas and Schwalbe (1983) believed similarly that stigmatized people would possess a lowered sense of self worth. On the other hand, Croker and Major (1989) thought that stigmatized persons would have a middle to high range of self-esteem based on techniques that they use to compensate for their situation. For this study, Rosenberg’s self-esteem scale (1979) and Gecas and Schwalbe’s self-efficacy concepts (1983) are tested on people who are stigmatized with the homeless identity to determine whether they will have a low, middle or high self-esteem level. A further area of investigation is the acquiring and exiting of the homeless role identity. People who are precariously domiciled can become homeless for a number of reasons such as unemployment or addiction, thus changing their identity from housed to homeless. The length of time that people are homeless influences the embracing of the new homeless identity (Snow and Anderson 1993). Conversely, individuals who are homeless can become housed. As for exiting the homeless situation, I examine what homeless people believe they need in order to no longer be homeless and the number of attempts they have made to leave the situation. 6 For this study, I base my inquiry around the following questions: While homeless, do individuals adopt a homeless identity (i.e., do they embrace or distance themselves from this identity)? Is a homeless identity the primary or most salient identity for people while in the situation? How committed are people to the homeless identity during their homelessness? What activities are important for individuals who are homeless? Do people think of themselves in positive or negative terms while homeless? How does being homeless affect people’s self-worth? What attempts do individuals make to exit the homeless situation? In the past, researchers predominately conducted ethnographies to understand the construction of a homeless identity (Cohen 2001; Snow and Anderson 1993). For this study, I build on the previous field research by testing Stryker’s structural identity theory on the three elements: salience, centrality and commitment (Stryker and Serpe 1994). I also examine identity meanings based on self descriptive terms and activities (Burke and Reitzes 1981; Burke and Stets 2009), along with the self esteem (Rosenberg 1979) and self-efficacy (Gecas and Schwalbe 1983) of homeless people. These theories are all examined using a standardized questionnaire. A survey provides the ability to test for validity, reliability and statistically significant relationships among the multiple variables. In addition, a new approach to measuring salience is tested based on conversational topics with others in various situations. This measurement is different than past research in which students were asked about meeting people for the first time and what they would tell others about themselves (Stryker and Serpe 1994). Also, the homeless condition, exiting the homeless situation and socio-demographic characteristics are measured in the surveys. As in previous studies, these variables are assessed across time by looking at how long and the number of times 7 people had been homeless. Finally, the respondents are asked whether they identify themselves as homeless or not. This research is innovative and significant in four ways. First, whereas past research investigated homeless identity, the meaning of being homeless and the self-worth of homeless people by conducting ethnographies, this study tests these variables utilizing a standardized questionnaire. Second, instead of only considering homeless identity in terms of whether people embrace it or not, this research also examines homeless identity theoretically (and empirically) using Stryker’s structural identity theory: salience, centrality, and commitment. Third, it applies a new approach to measuring identity salience based on conversational topics with others in various situations. Finally, this project expands on the common pathology model of homelessness, which focuses on fixing the problems of individuals, and applies a new identity centered approach to policies and program development. This new approach focuses on the positive meanings, activities and self-worth of people who are homeless by emphasizing their strong survival skills, ability to provide emotional support to other homeless people and their average to high self-esteem. This new method also highlights and evaluates other identities that homeless people possess such as parent and worker, instead of only focusing on the stigmatized homeless identity. The following chapters describe more comprehensively the goals, theory, methods, and findings of my research. Chapter two discusses the current and past state of homelessness, including the highly debated definition of homelessness, the individual and structural causes, and demographic characteristics. Chapter three then states the various strategies that people use to maintain and construct their identities, meaning, and self-worth while homeless. In chapter four, the theory of Stryker, Burke, Goffman, and Rosenberg used in this research is given in detail, 8 along with other supporting theoretical frameworks. Chapter five explains the sampling and survey research methods utilized in this study while chapter six provides the reliability data and descriptive statistics. Chapters seven (homeless identity), eight (self-esteem and self-efficacy) and nine (homeless identity meanings and behaviors) present the findings from the regression analysis used to test the hypotheses. After the results are presented, chapter ten offers the conclusion to the study, its key theoretical contributions, its relevance in the context of social service programs and policies, and future research possibilities. 9 CHAPTER 2 CONDITIONS OF HOMELESSNESS In order to understand homelessness, it must first be defined. Reviewing the literature, most often homelessness is identified as the lack of a permanent place to sleep at night, such as an apartment or house. Reasons provided for becoming homeless range from individual problems of addiction to structural factors such as the lack of affordable housing. In addition to knowing the causes of homelessness, researchers and policy makers throughout the U.S. want to understand the scope of the problem by determining the number and characteristics of the homeless population. The purpose of this chapter is to discuss these various aspects that encompass the past and current conditions of homelessness. HOMELESS DEFINITIONS An issue for people attempting to solve homelessness is how to identify the scope of the problem. One aspect of that is deciding who is homeless. Currently, there are a multitude of homeless definitions from various perspectives. The federal definition of homelessness was amended in 2009 as The Homeless Emergency Assistance and Rapid Transition to Housing (HEARTH) Act. The original McKinney-Vento Homeless Assistance Act (1987) focused on an individual or family who was literally homeless, such as those sleeping on the streets, in abandoned buildings, and in temporary shelters or housing programs. The 2009 amendment expands the federal homeless definition to include people who are to lose their housing within fourteen days and cannot get another permanent 10 place to stay due to a lack of resources or support networks. The updated definition also includes unaccompanied youth and families with children who have been without permanent housing for an expanded period, frequently moved around and will continue to be unstably housed due to someone in the household experiencing a disability or several obstacles to employment. A final factor for defining an individual or family as homeless is if they are fleeing from domestic violence or some other life-threatening situation. This federal definition is administered by the U.S. Department of Housing and Urban Development (HUD) and used by state and local governments and non-profit agencies receiving federally based funds. Another federal agency, the U.S. Department of Health and Human Services (2003), defines homelessness by length of time: temporary, episodic and chronic. People who are temporarily homeless experience the homeless spell for only a short period of time. Once they are housed, they are no longer in need of homeless services ever again. Episodically homeless people are homeless sporadically. Thus their use of homeless services are intermittent and usually for only short periods of time. Finally, it is important to note that the chronically homeless has a much limited definition than the previous two time frames in that it is only for unaccompanied individuals and certain sleeping locations (HUD 2007). Thus chronically homeless individuals are those who are continuously homeless for a year or more or have been homeless at least four times over the past three years. The individuals must also have a disabling condition such as an addiction or mental illness. For Snow and Anderson (1993), homelessness has three dimensions: residential, family support networks, and role and self-worth. The first dimension indicates a lack of conventional permanent housing and is typically the start of how people are defined as homeless. The second dimension focuses on the available support that homeless people may or may not receive from 11 their family. Familial support ties into the “traditional image” of home as being shared with family. The third dimension refers to the homeless role as being a master status. The homeless role or identity is highly visible and seen as negative, especially when interacting with others, and thus can affect the dignity and self worth of people who are homeless (Snow and Anderson 1993). Other researchers also indicate that the primary definition of homelessness begins with the lack of a conventional place to sleep such as a house or apartment. Rossi (1989) though challenges the concept of what is a “conventional dwelling”. He asks whether or not a hotel room, single occupancy or rented room, recreational vehicle or even a tent could be included. By reviewing how other agencies and researchers have defined homelessness, we see that it is not universally defined but instead created by each developer based on their research or purpose. TYPES OF SLEEPING LOCATIONS For people who are homeless, there are several types of sleeping locations. Emergency shelters provide homeless individuals and families with places to sleep nightly for up to three months. These facilities provide a minimum of basic services, such as a place to shower and evening and morning meals. On the other hand, a transitional housing program allows homeless people to stay for up to two years. The agencies also provide additional supportive services, such as case management and job training, to help facilitate moving into an independent living situation. Aside from these formally designated shelters and housing programs, some homeless people find or make improvised sheltered places to live or sleep (e.g., abandoned buildings, make shift sheds, bus stations or airport terminals). Finally, if individuals, or less likely families, 12 cannot find sheltered locations to sleep, then they will sleep on the streets, in parks, in doorways, under bridges or bushes, or in other unsheltered locations. CAUSES OF HOMELESSNESS The homeless literature has primarily focused on two main causes of homelessness (Main 1998; Morris 1998). One reason points to issues with the social structure such as affordable housing and political and economic factors as sources for the problem (Morrell-Bellai, Goering, and Boydell 2000). The other cause indicates the pathological behavior of individuals as the reason for people becoming homeless (Wright, Rubin, and Devine 1998). A continuing debate exists as to the importance of each in explaining the cause of homelessness. Structural explanations attribute the problem of homelessness to the dominant institutions of the United States such as the government, educational system and corporate business. Homeless advocates blame these institutions for not providing enough support for the most vulnerable and disadvantaged members. With continued systematic problems, an increasing unequal distribution of wealth and poverty creates a widening gap between the rich and the poor. Criticism is placed on the federal and local governments for failing to create social policy that adequately addresses the homeless problem (Elliott and Krivo 1991; Main 1998) and for changing policy to the detriment of the homeless. Policy changes, such as the 1996 Welfare Reform Act, adversely affect the social subsidies that the homeless use to survive. The welfare benefits prior to 1996 allowed homeless families the ability to leave emergency shelters within a six week time period. However with the changes to the welfare system, families need to stay for several months at long term transitional family shelters in order to stabilize and obtain permanent housing (Gerstel et al. 1996). 13 The main concern for homeless advocates is the failure to aggressively address the overarching societal factors that result in high levels of poverty that put people at risk of becoming homeless. For instance, the public school system does not provide the education and skill level that high risk students need to compete in the current global job market. Job opportunities for people with low skill levels are diminishing. Working at the minimum wage is not sufficient to provide a living for individuals or enough money for heads of households to support families. At the same time, inadequate mass transit systems and long commutes in urban areas makes it difficult for the homeless to travel the distance necessary to obtain possible jobs in prosperous suburban neighborhoods where a majority of businesses that pay livable wages currently are located. Often homeless people are located in the downtown sections of cities to be closer to social service agencies which are concentrated in the central city. Navigating the urban space between agencies is a problem because this places the homeless in direct contact with others using the same downtown space. Conflicts arise when the use of space is at odds between the homeless and businesses or residents (Snow and Mulcahy 2001). Various major cities throughout the United States such as Atlanta, Miami Beach, and San Francisco have passed ordinances to limit panhandling around businesses and tourist attractions (Copeland and Jones 2005). Even though both structural factors and personal problems are discussed in the literature, primary attention is placed on individual difficulties (Snow et al. 1994). Focus on personal troubles leads to the assumption that it is the pathology that is responsible for persons being homeless. The issue of homelessness is shifted then to a problem inherent in individuals and not in societal factors. It is assumed that people have the power to prevent homelessness. The solution for homelessness is then determined to be treatment of the pathology of individuals. 14 Much of the emphasis on personal characteristics has to do with the theoretical orientation of researchers from other disciplines such as psychology, social work and the medical profession. These researchers list multiple individual attributes, conditions, and characteristics as reasons for people becoming homeless. Mental illness (along with the deinstitutionalization of mentally ill people) has been the main problem discussed with substance abuse being a close second (Cohen 2001; Coldwell and Bender 2007; Goering et al. 2002; Goldfinger et al. 1999; Main 1998; Rosenheck 2000). Negative childhood experiences like poverty and family problems are found to create an increased risk for becoming homeless as adults (Koegel, Melamid, and Burnam 1995; Morrell-Bellai et al. 2000; Tam et al. 2003). Other factors that contribute to homelessness include spousal abuse, evictions, lack of familial support and physical disabilities (Goering et al. 2002; Morrell-Bellai et al. 2000). DEMOGRAPHICS WITHIN U.S. HISTORICAL CONTEXT In the U.S., the size and composition of homelessness fluctuates depending on the economic situation. Prosperous times obviously lead to a decreased homeless population. Unfortunately these periods do not last and are often followed by downward economic cycles that are difficult times for those living in poverty. After World War I, the United States economy experienced a number of disruptions: the stock market crash of 1929, a drought in the Midwest region from 1933 to 1935, and the hardships of the Great Depression in the 1930s. The homeless population was mainly comprised of young unattached men who moved from place to place looking for employment; however, families were also a part of the population (Ry 1993). Transient centers and overnight shelters were established by cities without any financial assistance from the federal government. 15 Unfortunately with so many people struggling to survive, providing assistance became a burden for cities and states which they could not handle. Because of the high demand, local agencies filled in the need but limited their aid to residents who had lived in the community for at least a year (Freidel 2001). With entry into World War II, the U.S. experienced a reduction in the size of the homeless population as men and women joined the military or found work in the factories (Hopper and Hamburg 1984). Following the war in the 1950s, the homeless population was predominately comprised of white adult men in their fifties who lived regularly or sporadically in single room occupancy hotels (SRO). The skid row section of the central cities is where they were concentrated. This part of cities catered to transients, poor and the homeless by providing lodging, restaurants, bars, pawn shops and cheap clothing stores (Bogue 1963; Rossi 1989). These homeless men were unmarried with few friends and little contact with their relatives. They experienced alcoholism, mental illness and physical disabilities. They lived off their social security benefits and menial jobs in agriculture, construction, and the loading docks. Missions and shelters were concentrated in the skid rows to provide the men who were not working with food and beds (Bogue 1963). From the 1950s to the 1970s the American economy prospered with an expansion of job opportunities and a focus on urban renewal. The skid row sections of downtowns where the homeless lived were viewed as problem areas. They became a target for redevelopment due to the cheap land and housing. The urban renewal projects of that era resulted in much of the affordable housing that supported the homeless being renovated or demolished and replaced with parking lots, office buildings and luxury housing (Rossi 1989). 16 In the 1980s, the size of the homeless population increased and a shift in the composition occurred (Rossi 1990). Unlike the skid row homeless of the previous years, this group was not concentrated in one area of downtown but was living on the streets and in shelters scattered throughout U.S. cities. The homeless population was more visible to the public due to their larger numbers, dispersion throughout cities, and because they were literally sleeping on park benches and doorways of office buildings. One main concern was the increase in the number of single women and families. Transitional and domestic violence shelters appeared to accommodate the larger numbers and changing needs of homeless (Weinreb and Rossi 1995). Racial and ethnic minorities came to comprise the largest percentage. The overall median age of this group was much lower at 34 years, mainly due to the younger ages of heads of families. As in previous years, this group experienced extreme levels of poverty due to inconsistent and non-existent employment. Unaccompanied adults also faced high levels of mental illness, alcoholism, and physical disabilities (Rossi 1990). LOCAL DEMOGRAPHICS With the increasing concern about the rise in the homeless population, communities throughout the U.S. began to be concerned about the scope of the problem locally. In 2002, the Southern city and its two counties (Tri-J) had decided that assessing the number of people homeless in the community and their needs was critical in addressing the problem. Beginning in 2003 and every two years thereafter, the Tri-J conducted an enumeration of the homeless population, which was then followed several months later with a survey. The homeless count takes place during the last two weeks in January based on a HUD mandate and gathers basic demographics along with the numbers (Parker et al. 2011). Several months later the homeless 17 survey is conducted to get a more in-depth understanding of the characteristics and needs of people who are homeless in the community (Parker 2009). On January 25, 2011, the city and its two counties (Tri-J) counted 6,838 as being homeless in the community (Parker et al. 2011). The majority of homeless people (5,987 persons) were located in the city where most social service agencies are located. Of the overall homeless number, there was a three way split in sleeping locations – emergency shelters (36 percent), unsheltered (35 percent) and transitional housing (29 percent). Adult males were the largest group of people homeless (68.5 percent) with adult females a distant second (15.3 percent) and children in families third (10.3 percent). A trend analysis of all four local homeless surveys revealed that the majority of respondents were black, single, non-veteran, middle aged men. For both the 2007 and 2009 homeless surveys (Massey et al. 2008; Parker 2009), the majority of respondents were homeless six months or less with this being their first time homeless, although these numbers had decreased from 2007 to 2009. As for the chronically homeless percentage, it had increased from 20 percent to 23 percent between 2005 and 2009 (Johnson, Ashley, and Pandley 2005; Parker 2009). The top causes of homelessness for both 2007 and 2009 were unemployment, alcohol/drug use and relocation from out of town. An interesting finding was that alcohol/drug use was the primary reason given for the 2007 Tri-J homeless survey but it was the lack of jobs for 2009. This is a possible reflection of the current difficult times that the U.S. economy is experiencing (Massey et al. 2007; Parker 2009). Another indication of the current economic hardships for this population was obtained from a question that asked about their current working situation. For 2005, the responses revealed that 18 over a quarter of people worked a full time job (27 percent). Yet in 2009 when that same question was asked, the number decreased significantly (7 percent) for people employed full time. Also, the number of respondents receiving governmental assistance almost doubled from 2005 to 2009 (Johnson et al. 2005; Parker 2009). SUMMARY A review of the literature finds several definitions of homelessness based on the agenda and specialization of the agencies and people creating the definition. Homeless definitions created by the federal government are for funding purposes. In the past, they were limited in their focus by concentrating on one main element of the homeless situation such as sleeping location or length of time. With the HEARTH Act amendment however, the federal homeless definition has expanded to include more dimensions similarly to that of homeless researchers. This study also starts with sleeping location for defining people homeless and expands the definition by investigating length of time homeless and other dimensions such as cause of homeless and experiencing a disability. In addition, the dissertation allows the respondents to state whether or not they identify as homeless, and if so, how strongly. A similar limited focus is found with discussions regarding causes of homelessness. Many researchers and advocates focus on the cause of homelessness as uni-dimensional - either structural factors or individual characteristics. However, by placing the homeless population and composition in a historical and local context as this study has done, the changing nature of the population can be seen. As for reasons of homelessness, a multi-dimensional approach is applied. Normally there is not just one cause as to why persons or families become homeless but several. The multiple factors are interconnected, even if on different levels. People without stable, adequate housing do in fact suffer disproportionately from a wide variety of chronic and acute 19 illnesses. Nevertheless, individual deficiencies cannot be blamed solely for a person becoming homeless. Several structural factors, such as being born into poverty, place persons at risk of considerable homelessness. A disability such as a mental illness, a natural disaster such as Hurricane Katrina, or an unexpected event such as a sudden illness can push individuals or families who are at risk past the level of maintaining a home and onto the streets (Wright et al. 1998). In working to eliminate homelessness, both individual and structural factors must be taken into account to come up with an appropriate response that meets the immediate needs of homeless people while addressing long-term causal factors. Finally, with this study, it is important to place the demographics, composition and number of homeless people in the community within a historical and geographical context. For instance, the local homeless population is definitely a reflection of the demographics and economic possibilities within the city. Another factor that will be considered is the city’s goal to focus on eliminating chronic homelessness in ten years by moving away from emergency shelters which sustain homelessness and towards transitional and permanent housing which appears to end the cycle of homelessness. 20 CHAPTER 3 STRATEGIES OF THE HOMELESS In the 1980s, with a change in the social demographics of the homeless population, also came a change in the homeless literature. Previously research had primarily focused on the causes of homelessness and the social demographic characteristics of the homeless. Additionally, Bogue (1963) also investigated the personality structure and self-image of homeless people on skid row. One area not studied by homeless researchers at the time was identity. Snow and Anderson (1987) realized this gap existed in the homeless research and focused their attention on this neglected area. Since the work of Snow and Anderson (1987), other researchers expanded their ideas on the construction and maintenance of the homeless identity. Studies also looked at extending other social psychological elements of the homeless. One area of interest focused on the meaning that homeless people give to their lives while on the street (Boydell, Goering, and Morell-Bellai 2000; Cohen 2001; Hill 2003; Snow and Anderson 1993). Another subject of importance has been the self worth of individuals who are homeless (Miller and Keys 2001; Morris 1998; Osborne 2001; Snow and Anderson 1993). Along with social psychological strategies of the homeless, this chapter also presents the work of researchers who studied the strategies of people who attempt to exit the situation. 21 HOMELESS IDENTITY To fill the void in the literature, Snow and Anderson (1987, 1993) examined the survival strategies of the homeless. The researchers observed homeless people in their daily routines and engaged them in conversations to understand their life on the streets. A main objective of their study was to advance the understanding of how individuals in the lowest strata of society survive socially and psychologically. Snow and Anderson (1987, 1993) found that people were faced with establishing who they are during this homeless situation by constructing and maintaining an identity that supported a sense of worth and dignity. Identity was not viewed by the researchers as a single element, but was instead separated into three types: social identity, personal identity and self-concept. Social identity is the identity given to persons by others based on the individuals’ appearance or behavior in a situation. This concept of social identity is consistent with Goffman’s (1963) definition. On the other hand, personal identity differs from Goffman (1963) and is instead seen as the attribution of meaning given to oneself during interaction. Finally, the self-concept is the overarching image one has of oneself. According to Snow and Anderson (1993), one of the main activities used when proclaiming a personal identity is that of verbal communication. Verbal construction, called identity talk, consists of three forms: distancing, embracement and fictive storytelling. The first form of identity talk is distancing, when individuals separate themselves from the homeless role and from homeless institutions. They believe that the homeless identity is inconsistent with their self-image or personal identity. There are three types of distancing. Associational distancing is when homeless people separate themselves from other homeless individuals. The second type of 22 distancing indicates that persons disassociate themselves from the role of homelessness (role distancing). Institutional distancing is the final technique and occurs when homeless individuals talk derogatorily of the service agencies that attend to the needs of homeless people such as Salvation Army. Embracement is the second type of identity talk. Individuals accept a homeless identity because it is consistent with their self-concept. Role embracement is seen as accepting a homeless category such as hippie tramp or dumpster diver. Associational embracement is when homeless people take seriously and are committed to social relationships with other homeless individuals. Accepting a set of beliefs congruent to the homeless identity is called ideological embracement (Snow and Anderson 1993). The final type of identity talk is that of fictive storytelling. Experiences and accomplishments of the individuals’ past, present or future selves are exaggerated during discussion to enhance self presentation. Additionally, homeless individuals may fantasize about the future indicating that they would be self-employed, possess money and material items, and be in a significant relationship (Snow and Anderson 1993). Snow and Anderson (1993) indicated that survival strategies and identity construction varied with time that homeless individuals spent on the streets. Homeless people on the streets for a short period of time distance themselves from the homeless identity, whereas those on the streets longer than two years embrace the homeless identity in conversation. Therefore, the identities that homeless people construct and maintain are not static but instead change with the passage of time. Being homeless creates identity problems for people. Positive identities that existed prior to becoming homeless are upheld, but the established past self is often lost and the homeless 23 identity of the present is devalued. For the future, different non-homeless identities are always indicated as emerging by people. Thus homeless identity formation is not static over time, but instead is a process that progresses towards accepting a homeless identity the longer individuals are on the streets (Boydell et al. 2000). MEANINGS AND BEHAVIORS OF HOMELESS PEOPLE To create meaning while in the homeless situation, individuals invoke “casual accounts or commonsense attributions” (Snow and Anderson 1993) to make sense of their circumstances. One account (“I’m down on my luck”) is that they credit their situation to unanticipated occurrences that are outside of their control such as bad luck or an unfortunate incident. A second attribution (“What goes around, comes around”) is that since they are suffering now, their luck will change for the better in the future and they will be on top. With the final account (“I’ve paid my dues”), it is believed that they have met the preconditions for a run of good luck due to their homeless experiences. Unfortunately, while homeless, individuals must deal with more than simply trying to meet their basic daily needs. They are also often faced with disabling conditions such as mental illness and alcohol or drug abuse. For mentally ill homeless persons, there is a continuous search to discover significance in their lives. Meaning and a positive self-image is infused into their lives by the telling of stories which can involve elaborate fantasies and important people and events. For alcohol and drug abusers, meaning is instilled through staying busy with activities such as finding sources by which to purchase alcohol or drugs, by pursuing obtainable goals like getting drunk or high, and by participating in the social life of the streets. These behaviors are viewed by 24 mainstream domiciled people as pathological, but for the homeless, the use of stories and the pursuit of these goals can be adaptive strategies to appear “normal” (Cohen 2001). According to Boydell et al. (2000), for homeless people, the loss of their past identity is important in understanding the meaning of their present situation. A number of homeless people express their current identity in positive terms using adjectives like honest, kind, hardworking, resourceful and independent, whereas others spoke of their selves in negative terms. The sense of a positive self, as well as a devalued self, is usually the result of interactions with other people. Many homeless people conceive of a future self that will no longer be homeless (i.e., a non-homeless identity). Their future self is seen as being healthy and stable. Once housed, they feel that they will be able to assist other marginalized people because, due to the homeless experience, they feel as if they have a deeper understanding of the purpose of life (Boydell et al. 2000). SELF-WORTH OF HOMELESS PERSONS Just as homeless people attempt to live lives of value, they also strive to have selves of value. According to Miller and Keys (2001), homeless people feel a sense of self-worth when they are validated by being treated with dignity. This respect is validated and sustained through receiving care, being treated as individuals, and receiving personalized service from others. By feeling good about themselves, individuals are more motivated to become self-sufficient, exit homelessness and contribute to others (Miller and Keys 2001). On the other hand, while homeless, people feel doubts about their self-worth because on a regular basis they experience negative attention, being ignored or avoided by those who are domiciled (Snow and Anderson 1993). A consequence of being treated badly is an invalidation 25 of their dignity (Burke and Stets 2009; Miller and Keys 2001). They feel demeaned and diminished as people. The dignity of homeless persons is violated in several ways such as through poor service and being unfairly treated which led to feelings of anger, depression and that no one cares (Miller and Keys 2001). Miller and Keys (2001) identified six factors that negatively affect the dignity of homeless people. One way to violate homeless persons’ dignity is to treat individuals like a number or to lump all homeless into one homogenous group with no personal identity. Another way is to treat people like children or animals. A third way is to give impersonal service by making people wait in a long line for a lengthy time or else force people to rush through a service line. A fourth way is to violate the rights of individuals. A fifth way is to assault the persons physically or verbally such as by yelling insults. A final way is to have an excessive number of rules, have staff enforce rules at their own discretion, have staff enforce rules with no rationale, and have staff give explicit orders of how to behave. For mainstream society, the number, quality and cost of material possessions that people own evoke a sense of status or social worth for the self. Unfortunately, homeless persons are unable to act as consumers and purchase items due to a lack of financial resources and a place to house the material objects. The homeless are thus socially isolated and ostracized because they cannot participate in the modern American consumer driven society (Hill 2003). They may feel that someone who possesses so little of material value must themselves be worthless (or they suspect that others think this about them). According to Miller (1998), gender affects the self-esteem of individuals. The self-esteem for unaccompanied women is found to be slightly higher than that for unaccompanied men. It is more difficult for men to accept their position of dependence while homeless as compared to 26 women, because they have been socialized to be independent and self-sufficient. On the other hand, women are more willing not to blame themselves for accepting assistance while homeless through their socialization process (Morris 1998). To protect their self esteem, the newly and chronically homeless create an “us” versus “them” dichotomy. Each group believes that they are different from the other group by not being as lazy or as unmotivated. One group presents themselves favorably while speaking of the other group in negative terms (Boydell et al. 2000). EXITING THE HOMELESS SITUATION With the focus of research on conditions of homelessness and social psychological factors, few studies have addressed attempts of persons who are homeless to transition off the streets. MacKnee and Mervyn (2002) determined nineteen categories that facilitated and four that hindered homeless people’s attempts to exit homelessness based on face to face interviews with seventeen formerly homeless individuals. From these twenty-three categories, five themes emerged: theme one – established supportive relationships, theme two – enhanced self-esteem, theme three – accepted personal responsibility, theme four - accomplished mainstream lifestyle goals and finally theme five - changed perceptions that street life was acceptable. Of these five themes, this study assesses themes one, two and four as they relate to the social psychological variables being investigated. Supportive relationships (theme 1) means having someone provide instrumental and/or socioeconomic support, such as a family member, social worker, friend, or pastor. When other people provide support, homeless people experience encouragement, acceptance and trust. Creating new relationships with mainstream people who have stable jobs and stable homes and 27 reestablishing family relationships help them to discover their personal identity and exit life on the streets. A vital factor in exiting homelessness and then not returning back to street life is the ability to sever ties with friends from the street. An obstacle to exiting the street life fully is feeling a sense of loyalty towards the street friends who they think of as “family”, feeling guilty over leaving this constructed street “family”, and a sense of loss for giving up these relationships (MacKnee and Mervyn 2002). Being treated with dignity by others is motivation to improve oneself, become self-sufficient, and exit homelessness (Miller and Keys 2001). Key factors to enhancing self-esteem (theme 2) is establishing confidence and belief in people’s talents. Accomplishing mainstream goals (theme 4) is also important in attempting to exit life on the streets. These goals include attending alcohol/drug rehabilitation programs to stop using substances, obtaining a legitimate job to maintain financial stability and finishing a school program and/or attaining a GED. Achieving these goals promotes a sense of pride, accomplishment, and level of confidence (MacKnee and Mervyn 2002). Various possibilities exist for individuals who maintain their domiciled identity in their attempts to leave the streets versus people who create a homeless identity (Osborne 2001). The individuals who strongly identify as homeless are more self-reliant and less likely to transition off the streets. Within a year time period, these people are significantly less likely than others to have made attempts to exit the streets. On the other hand, people who did not identify as being homeless are more likely to use social services and make attempts to get off the streets. The reason for the difference is that people who identify as homeless are fully integrated in the homeless role and thus use different strategies to maintain a homeless status. For people who have been on the streets for less than sixteen months, the average number of attempts to exit the street life is less than four. 28 Farrington and Robinson (1999) divided the twenty-one respondents from their participant observation study into categories based on length of time on the streets. For “aspirant exiters,” the length of time on the streets was less than one year. “Deniers” had been on the streets from fourteen to eighteen months. “Subgroupers” had been homeless from two to four years. “Carers and sharers”, “family” and “typicals” were homeless the longest time at more than three and a half years. Aspirant exiters, deniers and subgroupers all discussed leaving their homeless situation. Aspirant exiters did not identify with other homeless people, but instead distanced themselves both behaviorally and cognitively by stressing their skills, travel experience and coping abilities. Deniers disavowed group membership as homeless and instead asserted new identities. This allowed individuals the ability to exit the homeless situation, since their identity was not tied to it. Homeless identities were integral to carers, sharers, family, and typicals in the situation making it difficult for them to escape the street life. The ability to exit the homeless situation appears to be the result of a number of factors at both the structural and individual level. Structural reasons that enable individuals to leave a life of homelessness include a livable minimum wage, decent employment, and affordable housing. Personal issues that contribute to people remaining on the streets include impoverished support networks and alcohol or drug problems (Morrell-Bellai et al. 2000). SUMMARY The current literature views homeless individuals as coming to accept a homeless identity the longer that they are homeless (Snow and Anderson 1993). In this dissertation the current dualistic “either/or” approach (i.e., viewing someone simply as either homeless or not homeless) is expanded by investigating not only the acceptance of a homeless identity but also the strength 29 of that identity. Instead of Goffman’s concept of identity, in this study I use Stryker’s structural identity theory (Stryker and Serpe 1994), which will be discussed in the next chapter. Specifically, the focus is on what Snow and Anderson (1987) term the personal identity and self-concept of homeless people rather than the social identity. Often the definition and meaning of homelessness is defined by those in power such as policy makers, medical personnel, social workers, and directors of agencies. It is important to realize that the meaning that individuals attribute to the homeless role identity can be different than that assigned to it by others. Therefore, a strength of this study is that the respondents are asked as to what it means for them to be homeless. Being fully integrated into the situation, people who are homeless use different strategies to maintain their self worth (Osborne 2001). This research study looks at the self-esteem of homeless people (Rosenberg 1979) to learn, among other things, how it varies by length of time homeless. This study also examines the attempts of homeless people to exit the homeless situation. The attempts to transition out of the homeless situation are predicted by supportive relationships, specific behaviors like obtaining a job, and structural factors of affordable housing. 30 CHAPTER 4 SELF-CONCEPT OF HOMELESS PERSONS Much of the homeless research after Snow and Anderson (1987) has been empirically driven. Christian and Abrams (2003) found the lack of a theoretical framework in the current homeless identity literature to be a concern and filled the gap by using the social identity theory (SIT) of Tajfel and Turner (1986). SIT proposes that people create social identities based on social group membership. Members in a group are seen as sharing similar characteristics to each other and as differing from those outside the group (Stets and Burke 2000). Using SIT, Farrington and Robinson (1999) were able to predict that long term homeless people would stop making social comparisons with other groups and would instead compare themselves to others who are also homeless. Reviewing the homeless identity literature in the United States, I find a similar theoretical gap. Instead of using SIT, I apply the identity theory of Stryker to the current literature of homelessness and homeless identity. Specifically, I examine the identity within the context of a stigmatized group such as that of homelessness. Also, I address the meaning and behaviors that are attached to the homeless role identity and the factors involved with entering and exiting the role of homeless. Finally, I investigate the impact of being homeless on the self-esteem and self-efficacy. 31 STIGMATIZED IDENTITY According to Goffman (1963), in certain situations when people present their personal characteristics or beliefs, they are perceived by others as not fitting the traditional cultural norms and labeled as stigmatized. Possessing such a stigma does not allow individuals full acceptance by others and affectS their identity negatively. Assumptions exist as to how people are labeled as having a stigma and the process of managing the spoiled identity. People can acquire a stigma either at birth or at any time during their life. For Goffman (1963), three types of stigma exist. The first type is that of individuals with a stigmatized body or physical deformity. The second type, tribal stigma, bases the fault with people’s race, nationality or religion. The final stigma views the individuals’ character as being reduced in value. Homeless people are faced with one or more of these types of stigmas by first being black, disabled and/or an addict and then second by being reduced in value because they have no home. When individuals encounter persons for the first time, the interaction is based on initial impressions and by placing the other into an established category which allows people to know how to behave. For instance, as domiciled persons walk past homeless pan handlers on the streets, the pan handlers know to ask for spare change and the housed individuals know that one option is to not look at the homeless and to mumble something about not having any money at the moment. The established category is based upon the persons’ social identity which is the attribute that people possess such as the social trait of being homeless and a personal quality of being friendly. Individuals who possess a social identity that is less desirable than others are considered stigmatized. Stigmatized individuals are determined to be tainted, failing or incomplete. Over time, a stigma becomes a stereotype such as homeless people being labeled as 32 scam artists. Because persons are negatively stereotyped, they can suffer from discrimination which can result in reduced life chances (Goffman 1963). For Goffman (1963), a stigma which is known about by others such as a physical trait is considered discredited. A discreditable stigma, on the other hand, is not known by others in a social situation. For individuals with a known stigma, encounters with others are about managing the impression that the individuals are making. For people with an unknown stigma, the situation is about managing the information so that others do not discover the stigma. It is termed passing when the stigma is not detected by others, only identified by a few, or recognized and accepted. Unfortunately homeless people are rarely able to hide the fact that they suffer from the discredited stigma of not having a home. The stigma of being homeless is quite visible often by the clothes people are wearing and/or the belongings that individuals are carrying. Sometimes people can use their stigma of being homeless to their advantage to receive services such as food or shelter. However, homeless individuals attempt to conceal the possession of an unknown stigma such as being an addict, because knowledge of active addiction can cause persons to lose access to assistance. Stigmatized people are constantly aware as to the impression they are making to others and the consequences of that impression (Goffman 1963). When homeless people believe that they are not making a favorable impression due to possessing characteristics that don’t fit the traditional cultural norms, they experience feelings of shame (Goffman 1963). A gap exists between the homeless individuals’ virtual social identity, what people believe are the “ideal” traditional cultural characteristics that they are supposed to possess, and the actual social identity, what attributes persons really possess. This discrepancy 33 between the expected and the actual creates the stigmatized view that being homeless is a negative attribute (Adams and Sydie 2002). Sympathy and camaraderie, rather than shame and embarrassment, often exist among people who share the same stigmatized identity. For instance, homeless individuals are likely to group together as friends due to their shared stigma. Another group of sympathetic others are domiciled people who are intimately knowledgeable about the stigmatized group due to a special situation such as working with the group. With both of these groups, stigmatized people do not have to feel shame or guilt (Goffman 1963). When interacting with strangers, the communication is based on social identity or stereotyped responses; however, when interacting with others who are known personally, the communication is based on personal identity. As people become closer, a more realistic assessment and better understanding of the personal qualities of individuals are apparent. This means that people are seen as being uniquely different from others (Goffman 1963). IDENTITY THEORY Stryker (1989) developed identity theory to look at how the self process and outcomes are affected by social structure. People occupy roles and thus possess as many identities or selves as they do roles (James 1890). In this way, the diligent worker, the caring mother and the humorous friend are all identities that can be held by one individual. The multiple identities are organized into a hierarchal structure (Stryker 1989). The hierarchy of identities is based on identity salience “defined as a readiness to act out an identity as a consequence of the identity’s properties as a cognitive structure or schema” (Stryker and Serpe 1994: 17). Salience provides a framework that is used for interpreting events. Thus, the 34 identities are organized based on salience or the probability of invoking a certain identity in a series of situations. Consistency and predictability result when there is a tendency to invoke the same identity in different situations. Given the choice, people will invoke a salient identity in more situations and in different situations. For instance, individuals are likely to consistently invoke their homeless identity at shelters, housing programs and other social service agencies. If the homeless identity is confirmed by others in multiple situations, then salience is confirmed. This results in stability of identity salience across time and situation (Serpe 1987). On the other hand, when no identity is salient, then individuals will enact different identities in various situations. Salience is a function of people’s commitment to an identity (Stryker and Serpe 1994). For instance, the more committed people are to the role, then the more likely they would be to invoke the role identity. Commitment relates to how attached people are to the role and therefore personal and social costs are incurred when people are no longer fulfilling a role identity. Commitment is comprised of two dimensions: interactive and affective (Stryker and Serpe 1994). The interactive commitment is the quantitative component and is based on the number of other people individuals are tied to due to their role identity. Affective commitment, the qualitative component, relates to the strength of the attachment to others (Stryker and Serpe 1994). In the past, studies have emphasized one type of commitment over another (Hoyt and Babchuck 1983; Vannoy-Hiller and Philibar 1989). With this study, commitment is analyzed on both dimensions. Centrality is also a function of people’s commitment to an identity (Stryker and Serpe 1994). The central identity relates to how important the role identity is to individuals, whether the identity is central or peripheral (major or minor). For individuals, a particular identity such as 35 that of homeless may be considered primary or secondary. Often it represents the desirability from a personal point of view as to the preferred identity. Stryker and Serpe (1994) believe that a positive role will be most important to people. However, it must be considered for many that their homeless identity is their most important even though it is a stigmatized one (Goffman 1963). Identity theory argues that “the predominant direction of influence is from commitment to salience” (Stryker and Serpe 1994: 20). The other influence of commitment is towards that of centrality. Thus, hypotheses one (A - D) test the relationships between identity commitment, salience and centrality. H1A) The stronger the interactive commitment to the homeless identity, the greater the salience of the homeless identity. H1B) The stronger the affective commitment to the homeless identity, the greater the salience of the homeless identity. H1C) The stronger the interactive commitment to the homeless identity, the greater the centrality of the homeless identity. H1D) The stronger the affective commitment to the homeless identity, the greater the centrality of the homeless identity. In addition, identity salience and centrality predict time in role, although salience does a much better job explaining the time factor than centrality. Thus, the second hypotheses (A-B) investigate identity salience and centrality in relation to the length of time participating in a role. H2A) The greater the salience of the homeless identity, the longer people are homeless. 36 H2B) The greater the centrality of the homeless identity, the longer people are homeless. ACCEPTING THE HOMELESS IDENTITY Past research has focused on whether or not homeless people identify as homeless. Snow and Anderson (1993) determined that the discussion and perception of persons’ homeless identity changes with the length of time that individuals are without a permanent home. Homeless identity formation is a process that progresses towards accepting a homeless identity the longer individuals are homeless (Boydell et al. 2000). The taking on of a new role such as homelessness is a process that involves four stages – anticipatory, formal, informal and personal (Thorton and Nardi 1975). For the first stage, participants learn about expectations of the upcoming role. The second stage involves people actually participating in the role instead of just being outside observers. In the third stage, individuals learn the informal practices that help to navigate the formal system of the role. For instance, homeless people might learn from other homeless individuals where the best shelter in the city is located and what the rules are for getting in for the night. Finally, the last stage is about people making the role their own by adapting the role expectations to fit their own unique personalities. Therefore, hypotheses three (A-B) attempt to verify that the length of time and number of times that people are homeless affects the adoption of a homeless identity. H3A) The longer the length of time that individuals are homeless, the more likely people are to identify as homeless. H3B) The more times that people are homeless, the more likely people are to identify as homeless. 37 EXITING THE HOMELESS IDENTITY AND SITUATION Exiting the homeless role involves a disengagement process from a current role into a new role or re-establishing a past role. As part of the process, people begin to withdraw from the past role expectations in such areas as behavior and people associated with the role. Another part of the exiting process is disidentification, where individuals stop thinking of themselves as in the role such as being homeless. At the same time, people are in the process of taking on the expectations of a new role or re-establishing a past role. This is referred to as role socialization (Ebaugh 1988). Osborne (2002) has determined both a positive and negative impact of identifying as homeless. A negative result of people who identify as homeless is that they are significantly less likely to make attempts to transition from the homeless situation. Hypotheses four and five (A-D) investigate the relationship between homeless identity and attempts made to exit homelessness. In addition, hypothesis five tests the impact of the homeless identity on attempts to exit the situation. H4) People who identify strongly as homeless are less likely to make attempts to exit the homeless situation than those who do not identify as homeless. H5A) The greater the interactive commitment to the homeless identity, the less likely people are to attempt to exit the homeless situation. H5B) The greater the affective commitment to the homeless identity, the less likely people are to attempt to exit the homeless situation. H5C) The greater the salience of the homeless identity, the less likely people are to attempt to exit the homeless situation. 38 H5D) The greater the centrality of the homeless identity, the less likely people are to attempt to exit the homeless situation. Supportive relationships such as with family members encourage homeless people to make attempts to exit the situation (MacKnee and Mervyn 2002; Morrell-Bellai et al. 2000). The following hypothesis therefore investigates the support of families to relatives who are homeless in their attempts to exit the situation and role. H6) The more support from family members, the more likely people are to attempt to exit the homeless situation. SELF-ESTEEM Global self-esteem is the self evaluation of people’s overall personal worth. Most often people desire to think well of themselves. Self-esteem is an intrinsic motive in everyday life that influences what people say, how people act, what people attend to and how people direct their efforts (Rosenberg 1979). Three principles are important in the formation of self-esteem: reflected appraisals, social comparison, and self-attribution. Reflected appraisals, based on Cooley’s looking-glass self, implies that individuals’ self concept is influenced by what people believe are the perception of other people’s judgments towards them. An important factor of reflected appraisals is that the other persons need to be significant for the opinion to be strongly valued. Significant others are often parents, spouses, friends or bosses. Social comparison entails individuals judging themselves in reference to others when there is otherwise no objective information available. Self attribution involves individuals ascribing characteristics, motives, causes, etc. to themselves based on the success or failure of their actions (Rosenberg 1979). 39 People with high self-esteem have self-respect, self-acceptance and self-worth. Individuals will be satisfied with who they are as people and satisfied with their life. While people understand their own value, they also acknowledge their own faults and hope to overcome them. Individuals with high self-esteem who succeed in a particular situation or in life will explain the achievement in terms of their own personal merit. On the other hand, people with high self-esteem who fail in certain situations or in life will attribute the shortcoming to external factors beyond their control. Both the interpretation for success and failure are means for protecting the high self-esteem. When comparing the self to others, people with high self-esteem will not believe themselves to be better or worse than other individuals (Rosenberg 1979). People with low self-esteem hold a negative attitude regarding their self-concept and consider themselves to be unworthy and inadequate. Individuals lack self-respect and believe themselves to be deficient as people. Persons with low self-esteem who succeed or fail at specific events blame the poor performance on internal factors and refuse to accept information that is to the contrary (Rosenberg 1979). Individuals are always alert, avoiding, protecting and coping with possible threats to the self-esteem. Several defense mechanisms are employed in the service of self-esteem protection and enhancement: rationalization, compensation, projection, displacement, reaction formation and repression. Rationalization involves using a socially admired reason for behavior that might not otherwise be accepted. Compensation entails achieving extraordinary success in one area to overcome failure in another area. Projection includes placing on others undesirable traits that in fact individuals possess and would be problematic for the self-esteem if recognized. Displacement, also known as scape- goating, is when people boost their own self-esteem by asserting their superiority over others because they are frustrated and humiliated by those who 40 are more powerful. Reaction formation emphasizes the feelings or characteristics that are reverse of the actual undesirable feelings or characteristics that people possess. Repression involves suppressing unconscious impulses that would upset self-esteem if recognized (Rosenberg 1979). Self-esteem can be conceptualized in three ways: as an outcome, buffer and motive. Most often self-esteem is conceived of as the product of accomplishing an ideal goal with the actual performance. On the other hand, self-esteem can also be thought of as protecting the self when behavior of people does not match up to the anticipated results. Finally, people are motivated to either maintain or improve their current level of self-esteem (Cast and Burke 2002). Self-Esteem of Stigmatized Individuals According to Croker and Major (1989), previous theoretical research has indicated that the self-esteem of people who are labeled as stigmatized would be low. However, empirical research has not often supported those findings. One reason for this discrepancy is that past research has seen self-esteem as “a stable trait that is consistent across different situations” (Crocker and Quinn 2003: 153). Instead, Crocker and Quinn view self-esteem as being constructed in the situation based on meanings that individuals attach to the situation. People in a stigmatized group use several strategies as part of the group to protect their self-concept. One strategy is to attribute a negative feedback or outcome to being a part of the group in the sense that people are prejudiced towards the group. Therefore, the failure is not seen as a personal fault but as being part of the group. The second method is to make in group comparisons with others who are part of the stigmatized group. Comparisons with people in an advantaged group may be painful and result in lowered esteem (Tajfel and Turner 1986). A third 41 technique is to devalue performance outcomes in which they or their group do poorly and to value those in which they or the group does well (Rosenberg 1979). These strategies are not consistent across the board for each member of the stigmatized group. Instead there are factors that indicate the prevalence of the use of each, such as time since acquiring the stigma, visibility of the stigma, acceptance of negative views by others towards the stigmatized group and who holds responsibility for the stigma (Crocker and Major 1989). Therefore, hypotheses seven (A-B) investigates the correlation between the length of time and number of times in the homeless role and self-esteem. H7A) The longer people are homeless, the lower the self-esteem. H7B) The more times people are homeless, the lower the self-esteem. Hypothesis eight tests homeless identification as one of the possible factors that predicts self-esteem. H8) People who identify as homeless are more likely to possess a lower self-esteem than those who do not identify as homeless. Additionally, hypotheses nine (A-D) assess the impact that (interactive and affective) commitment, salience and centrality of the homeless identity have on people’s self-esteem as an outcome. H9A) The greater the interactive commitment to the homeless identity, the lower the self-esteem. H9B) The greater the affective commitment to the homeless identity, the lower the self-esteem. H9C) The greater the salience of the homeless identity, the lower the self-esteem. 42 H9D) The greater the centrality of the homeless identity, the lower the self-esteem. The following hypothesis examines the support of significant others to self-esteem. H10) The greater the support of homeless people’s families, the higher the self-esteem. SELF-EFFICACY For Gecas and Schwalbe (1983), self-efficacy is based on people’s competence level and not their sense of worth. An efficacious approach to self-esteem instead focuses on individuals as causal agents. Through the consequences of the individuals’ actions, people come to evaluate and understand the self, and this creates the base for the experience of the self-efficacy. People with high levels of efficacy will view the self as capable. The actions of people take place within a physical and social structure. The conditions of the structure can both enable and constrain the actions of individuals. One such condition is that of the interaction of people within a power structure. Power relationships are based on access to resources, autonomy and control. Those who lack resources are on the lower end of the power hierarchy and are often more dependent on others for survival. Because they are less autonomous and have little control, Gecas and Schwalbe (1983) believe that the opinions of others will matter more. Therefore, the process of reflected appraisals may be most applicable to those on the lower end of the power hierarchy such as homeless people for determining their self-esteem, whereas for those with more resources and higher up on the power hierarchy, social comparisons as a process will be more relevant (Gecas and Schwalbe 1983). Thus, hypothesis eleven attempts to verify that the support from family members affects self-efficacy. 43 H11) The greater the support of homeless people’s families, the higher the self-efficacy. The next hypothesis investigates the acceptance of the homeless identity in relation to self-efficacy. H12) People who identify as homeless are more likely to possess a lower self-efficacy than those who do not identify as homeless. Hypothesis thirteen (A-D) assess the impact that commitment, salience and centrality of the homeless identity have on self-efficacy. H13A) The greater the interactive commitment to the homeless identity, the lower the self-efficacy. H13B) The greater the affective commitment to the homeless identity, the lower the self-efficacy. H13C) The greater the salience of the homeless identity, the lower the self-efficacy. H13D) The greater the centrality of the homeless identity, the lower the self-efficacy. Finally, the correlation between the length of time and number of times homeless and self-efficacy is analyzed with hypotheses fourteen (A-B). H14A) The longer people are homeless, the lower the self-efficacy. H14B) The more times people are homeless, the lower the self-efficacy. 44 HOMELESS IDENTITY MEANINGS AND BEHAVIORS For each culture, identities are based on shared meanings (Burke 1980; Burke and Tully 1977) and behavioral expectations attached to social positions or roles (Stryker and Burke 2000). Meanings are understood through the interaction with others in a situation. As people perform in a particular role, they assess how to respond appropriately to others. Individuals grasp the meanings of the identity through the expected interaction and by seeing the reaction of others over time to their role performance (Mead 1934; Turner 1962). People judge each other’s behavior in a role to determine if it conforms to the defined role expectations within a particular society or culture. Over time, there is a desire to be consistent while in the role (Burke and Reitzes 1981). Behaviors attached to the role of homelessness include telling elaborate stories about important people and events to create a positive self-image and getting drunk or high for alcoholics or addicts living on the streets (Cohen 2001). In the end, the goal is for there to be a match between the meanings involved with occupying the role and the role behaviors that are performed while interacting with others (Burke 1980; Burke and Reitzes 1981). Therefore, hypotheses fifteen (A-G) examine the correlation between homeless identity behaviors and identity meanings. H15A-G) The more important the behaviors are to the homeless individuals, the more likely people are to describe their homeless identity meanings in negative terms. The homeless identity behaviors include having a safe place to sleep, knowing who to trust, knowing which are the best meal sites, helping others who are homeless, telling stories of past successes, staying sober and sharing information with other homeless people. 45 Each role identity has a set of meanings attached to them, which hypotheses sixteen (A-D) investigates. H16A) The greater the interactive commitment to the homeless identity, the more likely the respondents will be to describe their homeless identity meanings in negative terms. H16B) The greater the affective commitment to the homeless identity, the more likely the respondents will be to describe their homeless identity meanings in negative terms. H16C) The greater the salience of the homeless identity, the more likely the respondents will be to describe their homeless identity meanings in negative terms. H16D) The greater the centrality of the homeless identity, the more likely the respondents will be to describe their homeless identity meanings in negative terms. Comparatively, chronically homeless persons are more likely to speak negatively about themselves than the newly homeless (Boydell et al. 2000). Thus, hypotheses seventeen (A-B) examine the degree to which individuals describe themselves based on their time that they are homeless. H17A) The longer people are homeless, the more likely people will be to describe their homeless identity meanings in negative terms. H17B) The more often people are homeless, the more likely people will be to describe their homeless identity meanings in negative terms. 46 The following hypotheses test the prediction that self-esteem and self-efficacy will have an effect on homeless identity meanings due to the meanings being described using descriptive adjectives. H18A) The lower the self-esteem, the more likely the respondents will be to describe their homeless identity meanings in negative terms. H18B) The lower the self-efficacy, the more likely the respondents will be to describe their homeless identity meaning in negative terms. Identities are expressed by individuals through the behaviors attached to the role, which hypothesis nineteen examines. H19A) The more important the behaviors are to individuals, the greater the interactive commitment to the homeless identity. H19B) The more important the behaviors are to individuals, the greater the affective commitment to the homeless identity. H19C) The more important the behaviors are to individuals, the greater the salience of the homeless identity. H19D) The more important the behaviors are to individuals, the greater the centrality of the homeless identity. Hypotheses twenty (A-B) address the relationship between a specific homeless identity behavior, staying sober, and self-esteem and self-efficacy. H20A) The more important staying sober is to individuals, the higher the self-esteem. H20B) The more important staying sober is to individuals, the higher the self-efficacy. 47 SUMMARY Much of the past homeless research on social psychological aspects of life on the streets has been addressed by scholars empirically but not theoretically. I propose filling the gap by applying the structural identity theory of Stryker, utilizing all three elements: salience, commitment (interactive and affective) and centrality (Stryker and Serpe 1994). Also, much of the previous research has investigated role identities based on the traditional life trajectory such as student (Burke and Reitzes 1981; Stryker and Serpe 1994). Instead for this study, I examine the non-traditional and stigmatized role identity of homelessness. To know how to act in a traditional life trajectory role category such as parent, student or worker, there are set behavioral expectations attached to the position. For example, students are expected to attend classes, read books, write papers, and study for exams (Burke and Stets 2009). However, for a stigmatized non-traditional life trajectory social position such as homeless, what are the behavioral expectations and the meanings that are attached to the role identity? I investigate this question by examining homeless persons’ concept of meanings and behaviors attached to the homeless role identity. With acquiring the new identity of homelessness, I ask people as to how strongly they identify as homeless, if at all. I expect that identification as homeless will be higher among those who have been homeless longer. As for exiting the situation, I focus on what is needed to no longer become homeless and the number of attempts people have made to find housing. While homeless, individuals are constantly working to protect and maintain their self-esteem. For this study, I am not able to determine the particular defense mechanisms that homeless persons use to cope while in the situation. Instead the study investigates the negative 48 evaluation of self-esteem that people hold while homeless. This research examines both self esteem and self efficacy. For the dissertation, the independent variables are hypothesized to cause the dependent variable based on theoretical assumptions. However, it is important to note that these causal relationships can also be in the opposite direction. Thus, not only can staying sober effect individuals’ self-esteem but their self-esteem can also predict people staying sober. 49 CHAPTER 5 METHODOLOGY In the past, researchers predominately conducted ethnographies to understand the construction of a homeless identity (Cohen 2001; Snow and Anderson 1993). Instead for this study, homeless identity theories are examined using a standardized questionnaire. This data collection method allows for the testing of the relationship between the multiple social psychological variables in the study. The homeless identification measurement of Snow and Anderson (1193) is expanded by investigating Stryker’s structural identity theory on the three elements: centrality, salience, and commitment (Stryker and Serpe 1994). In addition, a new approach to measuring salience is tested based on conversational topics with others in various situations. Identity meanings are examined based on self descriptive terms and activities (Burke and Reitzes 1981; Burke and Stets 2009), along with self esteem (Rosenberg 1979) and self-efficacy (Gecas and Schwalbe 1983) of homeless people. Finally, socio-demographic characteristics, homeless condition, exiting the homeless situation and family support are also measured in the surveys. SAMPLE With homeless populations, collecting a probability sample is difficult due to the problem of creating an accurate homeless sampling frame or list. For instance, people living on the streets often sleep in isolated and hidden locations such as in bushes and under bridge overpasses for safety and warmth (Runkle and Parker 2007). The strengths of this study, therefore, are that it is 50 based on a stratified sample from a variety of service agency types (e.g. meal site, general service provider, emergency shelter and transitional housing) and it accurately reflects the demographics of the homeless population in this community (e.g., its racial, gender, age, and military service composition). These characteristics and their frequencies were determined using the 2007 Tri-J homeless survey for the city and its counties (Massey, Runkle and Parker 2008). For example, the 2007 homeless survey found that 86 percent of respondents were black and 10 percent were white, while this study’s respondents were 82 percent black and 11 percent were white, non-Hispanic (Table 1.1). The sample size necessary for the study was calculated at 350 cases. The case numbers were based on a power test that determines the sample size for the population (Sudman 1976) and the 2007 homeless census for the city and its counties where 5,978 adults were counted in one night (Runkle and Parker 2007). In total, the study yielded 380 surveys. Unfortunately, 54 surveys were unusable due to being incomplete, duplicates, the respondents indicating that they were not homeless or the respondents living in their own place and thus not being homeless. With this final sample size of 326, the confidence level is at 95 percent with a margin of error (confidence interval) at 2.25. Ten service provider agencies were approached to determine if they would participate in the study. Out of the providers asked, six different homeless service providers gave permission to conduct surveys on site. The service agencies, referred to as A to F, were selected based on a range of criteria from size of facility to demographics of clients, such as household composition and gender, to type of services provided, such as outreach services, emergency shelter, transitional housing and/or meals. 51 Site A is a service provider agency which operates on weekdays from 9 a.m. to noon. The service provider assists almost 200 homeless or nearly homeless people each workday. People wanting assistance must make an appointment. The agency provides services such as emergency food, assistance in obtaining identification, and referrals to emergency shelters, transitional housing programs, rehabilitation services and health services. Site B serves a morning meal to over 500 people every Saturday. The meal site predominately caters to homeless single men and women, along with a smaller number of homeless families. Those desiring a meal must arrive around 9 a.m. to receive a ticket. Beginning around 11 a.m., people are called in groups by their ticket numbers to line up for a meal. The guests are seated at tables where volunteers serve the meal and beverages, along with a dessert. A second meal site (C) serves a Sunday midday meal to over 300 people each week. This location is set up similar to the previous in that it mostly serves single men and women, the people need to arrive about an hour or so prior to the meal to get a ticket, and then are served the meal at tables by volunteers. Site D also provides a meal but on Tuesday afternoons. Unlike the other meal sites, individuals must sign up for and participate in the client program to receive the meal. The program assists people with additional needed supportive services such as case management and provides basic supplies such as toiletry items, socks, coats and blankets. As with the other organizations, the clients are seated around tables where they are served a meal by volunteers. The other two agencies provide sleeping facilities for homeless people. The first bed provider (E) offers both an emergency shelter and a transitional housing program for single men. The programs do not accept drug using or dually diagnosed individuals (diagnosed with both mental health and drug or alcohol problems). The emergency shelter is on a first come, first serve 52 basis, and homeless men must be at the facility by 2 p.m. for the intake process. Whereas the emergency shelter only provides a bed and a meal for one night at a time for 150 men, the transitional housing program also provides supportive services such as job training to about fifty men. Individuals must stay a minimum of six months with a maximum length of two years. The clients in the transitional program who are working pay a fee of 15 percent of their income. The other bed provider agency (F) also offers several emergency shelter and transitional housing programs which include employment, recuperative care, mental health treatment and veterans programs. These programs assist over 200 single men and 50 single women and women with children. Together, the six agencies (A-F) serve a broad and diverse set of homeless people. SURVEY DESIGN The questionnaire (see Appendix A) was designed over the course of several months in 2007. The survey is nine pages in length and included all close ended questions or statements, except for the last question which was open ended to allow the respondents to express their experience of being homeless. A majority of the questions were taken from other research studies. For example, the questions regarding homeless history were from the local city and counties’ (Tri-J) homeless survey (Massey, Runkle and Parker 2008). With several of the borrowed questions or statements, I utilized the exact same wording that the other studies used to measure their concepts, such as Rosenberg’s self esteem scale (Rosenberg 1979). Other borrowed questions were modified for this study. For instance, the question of where the respondents had usually slept at night during their current homeless situation was modified from “since you became homeless” (Massey, Runkle and Parker 2008) to “the past month prior to the 53 survey”. Other items for the questionnaire such as the salience measurement were created for this research study. The questionnaire was designed to be self-administered. A strength of a self-administered format is that respondents appear more comfortable answering sensitive or embarrassing questions rather than as face to face interviews (Czaja and Blair 1996). By using a self-administered survey, the questionnaire must have instructions and questions that are clear and able to be understood by a variety of respondents. This method allows the respondents to read a list of possible answers, whereby the participants are more likely to choose among the listed answers instead of selecting the other category and providing their own responses. The survey is comprised of eight parts. Part I of the survey examines the demographic characteristics of the respondents and asks identifying information to check for duplication. Part II focuses on two aspects of identity theory – identity centrality and identity salience. Part III asks about self-esteem and self-efficacy of the respondents. Part IV discovers the factors of homelessness and acceptance of the homeless identity. Part V looks at affective and interactive commitment to the homeless identity. Part VI investigates the identity meanings of being homeless and identity behaviors performed while homeless. Part VII addresses exiting the homeless situation and family support. Finally, Part VIII asks the respondents to share their experience of being homeless. MEASUREMENT For this study, demographic characteristics (see items 1-11 of Part I of questionnaire in Appendix) are measured with the dummy variables race/ethnicity, gender, veteran status, relationship status and parental status. Demographic interval variables include age and previous 54 month’s income. Education is assessed by asking the respondents to provide their highest level completed. To determine household type, the respondents are asked whether or not they are living with family members such as children or a spouse during the time of the survey. Respondents are asked how long they have lived in the city to calculate if they are long term or short term residents. Finally, two questions, initials and date of birth, are used as identifiers to check for duplication and age of respondents. The age of respondents are grouped into several categories: 17 years or younger, 18-24, 25-34, 35-44, 45-54, 55-64 and 65 years or older. In Part II of the survey, identity is conceptualized and measured utilizing Stryker’s identity theory and variables of centrality, salience and commitment (Stryker and Serpe 1994). For centrality (see item 1 of Part II of questionnaire in Appendix), the respondents are provided a list of roles or groups including family member, friend, homeless, veteran and so forth (Thoits 1992). The respondents are then asked to rank the roles or groups as to the five most important (Brown 1998). The second identity measurement, salience, is developed for this study (see Part II, items 2-6). Again, the respondents are asked to look at a similar list of roles or groups such as family member, friend, homeless, veteran and so forth as they did for the centrality measurement. From the list, the subjects are requested to choose their first and second most discussed role or group when conversing in various situations such as with friends, with family, at service agencies, at doctors’ offices and at emergency shelters. The salience of the homeless identity is determined by the total number of times that homelessness is discussed with others in various situations. Thus the scores range from homelessness discussed in four to five different situations = 4, homelessness discussed in three different situation = 3, homelessness discussed in two different situations = 2, homeless discussed in one situation = 1 and homelessness not discussed = 0. 55 Self-esteem (see items 1-10 of Part III of questionnaire in Appendix) is assessed using the Rosenberg Self-Esteem Scale which consists of ten items. The items range from “I feel that I'm a person of worth, at least on an equal plane with others” to “At times I think I am no good at all.” The items are measured on a four point Likert scale of strongly agree = 3, agree = 2, disagree = 1 and strongly disagree = 0. Six additional statements are provided to calculate self-efficacy (see Part III, items 11-16). These statements range from “I can pretty well control things that happen to me” to “sometimes I feel that I’m being pushed around” (Reitzes and Jaret 2009). They are also evaluated on the same four point Likert scale as self-esteem ranging from strongly agree to strongly disagree. Homeless identity is determined by asking “Do you identify as a homeless person?” (see item 2 of Part IV of questionnaire in Appendix). The response options include strongly identify, identify, not sure if identify and do not identify. Another item in Part IV of the survey inquires about the causes of homelessness ranging from personal reasons of addiction to structural factors such as the economy. The other five questions in Part IV ask about the history of homelessness. One question investigates where homeless persons had slept most often in the month prior to the survey. The answers to sleeping location are grouped into two categories to create a dummy variable: unsheltered locations such as on the streets and in abandoned buildings and sheltered locations including emergency shelters, transitional housing which includes treatment programs, anywhere that they could find a place to sleep, hotel or motel room and staying in apartment or house of family or friends often called doubled up. Question four of Part IV of the survey asks about the number of different times that people had been homeless over the past three years. Questions five and six gauge the current length of time homeless and the length of time that people had been continuously homeless over the past 56 three years. Finally, the last question of this section investigates whether or not the respondents were experiencing any disabilities such as chronic health problems or mental illness during the time period of the survey. Questions two through seven of Part IV are taken from the local city and counties’ homeless survey (Massey, Runkle and Parker 2008; Parker 2009) with slight modifications made for this study. For affective commitment identity to the homeless identity (see items 1-10 of Part V of questionnaire in Appendix), the respondents are provided ten statements that analyze the strength of their ties to other homeless people (Reitzes and Jaret 2009). The statements range from “I don’t feel connected to others who are homeless” to “people who are homeless understand me better than most other people do”. The items are measured on a four point Likert scale: strongly agree = 3, agree = 2, disagree = 1 and strongly disagree = 0. Finally, interactive commitment to the homeless identity is determined by asking individuals the number of homeless friends that they have (see Part IV, item 11). The response choices include 0, 1-2, 3-5, 6-10, 11-15, and 16 or more friends who are homeless. To measure homeless identity meanings (see items 1-7 of Part VI of questionnaire in Appendix), respondents are asked to complete the sentence: “As a homeless person, I am . . . “ by selecting their preferred descriptor in a series of seven adjectives pairs arranged in a five point semantic differential format (Osgood 1976). The adjective pairs include honest-dishonest, friendly-unfriendly, hardworking-lazy, resourceful-not capable, independent-dependent, kind-mean, and motivated-unmotivated (Boydell, Goering and Morrell-Bellai 2000). The items are coded as very positive = 4, positive = 3, neutral = 2, negative = 1 and very negative = 0. To study what behaviors homeless people consider most important (see Part VI, items 8-14), respondents are given a list of seven behaviors that are generally associated with being 57 homeless and are asked to rate how important or unimportant each behavior is to them as homeless persons (Brown 1998; Reitzes and Jaret 2009). The activities include finding a safe place to sleep, hanging out with friends and getting drunk or high, telling stories, sharing ideas, knowing who to trust, and knowing which meal sites serve the best tasting food (Cohen 2001). The items are coded as very important activities while homeless = 3, important = 2, not so important = 1 and not at all important behavior = 0. Obtaining permanent housing is an important component for transitioning out of homelessness (see items 1-3 of Part VII of questionnaire in Appendix). The first question about exiting the homeless situation inquires about what people need to find permanent housing (Morrell-Bellai, Goering, and Boydell 2000; MacKnee and Mervyn 2002). The next question asks about the number of attempts made to acquire housing such as viewing an apartment or filling out an application for an apartment during the month prior to the survey. The final permanent housing question addresses the emotional difficulty (i.e., feeling some guilt) that persons might experience by moving into housing and leaving their friends who are still homeless (MacKnee and Mervyn 2002). The last four questions of Part VII of the survey investigate the support that the homeless respondents receive from their family members. These statements are taken from the Multidimensional Scale of Perceived Social Support (MSPSS) (Zimet et al. 1988) and are measured using a four point Likert scale ranging from strongly agree to strongly disagree. Family support focuses on emotional support, becoming sick, when things go wrong, a source of comfort, and financial assistance. 58 Finally, the last question of the survey is open ended (see item 1 of Part VIII of questionnaire in Appendix). This question asks the respondents to share their experience of being homeless. It provides them a chance to express personal comments in their own voice. DATA COLLECTION A pre-test of the survey had been conducted in May 2008. Approximately twenty people were asked to fill out the questionnaire. After completing the survey, the respondents were asked to provide feedback in a debriefing session regarding such issues as the overall survey design, specific questions or sections, and length of time. The questionnaire design and implementation procedure was revised based upon the debriefing responses provided. The data were collected from June to August 2008 over several days each week at six different homeless service provider agencies. An administration schedule for the survey was created prior to collection. The schedule was determined by when homeless people would be seeking services at the agencies and a convenient date and time for the staff of the agencies. Along with the researcher, at several of the sites one or two research assistants helped with collecting the data. One research assistant was a formerly homeless veteran, while the other was a graduate student from the same university program as the researcher. At the sites, as many clients as possible who were seeking services at the agency were asked if they would like to participant in the study. Respondents were provided with a brief verbal overview of the study, along with a consent form to read. If the respondents agreed to take the survey and were able to fill out the questionnaire on their own, they were provided with the survey, a clipboard and pen. The researcher and assistants were available to answer any questions. After completing the questionnaire on their own, the respondents returned the form to 59 the researcher or the assistants who then reviewed the survey for any sections that were missed or filled out improperly. For missed or improperly answered questions or sections, the researcher or assistant explained to the respondent that they missed a section or filled it out improperly and were asked if they would fill that part out again. If the participants had problems with reading and comprehension, then the survey was read to them by the researcher or an assistant. The questionnaire took approximately twenty to minutes to answer. Once the survey was completed and reviewed, the respondents were thanked for participating and provided a small participation gift of a $1 fast food coupon. DATA ANALYSIS A code sheet was created as a guide for entering the data generated by the survey questionnaire into SPSS. The last item on the questionnaire was open-ended and therefore the respondents’ exact words were typed into an excel spreadsheet for analysis. Entering the survey data into SPSS and Excel took several months. The data were then cleaned by checking for inconsistent or invalid values, probably caused by coding or data-entry errors. The data were run through a series of cleaning techniques such as running frequencies and by sorting. Once any data errors were identified, they were corrected, if possible, with the valid values. This procedure of preparing a clean data file took several weeks (Czaja and Blair 1996). After the data had been cleaned, some of the raw data were constructed into new variables such as number of disabilities, re-coded into different variables such as age groups or sleeping location group, or totaled into a range of scores for scale items such self-esteem or affective commitment identity. Due to the dependent variables being continuous in this study, all the hypotheses were tested using multivariate regression. Another reason for choosing multivariate regression was 60 that it is a useful method for modeling. This study typically ran two models for each hypothesis. Model one provided a baseline for the stepwise regression addition method based on theoretical reasons that the independent variables provided a strong basis for explanation of the dependent variable (Agresti and Finlay 1997). With model two, I also included the demographic or characteristic variables to add depth and control to the analysis. Therefore, model two included race/ethnicity, gender, relationship status, military status, parent, family member, education level, age group, years living in Atlanta, last month income, sleeping location groups, and number of disabilities. The first three hypotheses investigate the homeless identity. Hypothesis one tests that interactive commitment (H1A) and affective commitment (H1B) to the homeless identity will predict homeless identity salience, and that interactive commitment (H1C) and affective commitment (H1D) to the homeless identity will also predict homeless identity centrality. The next set of hypotheses analyze homeless identity salience (H2A) and homeless identity centrality (H2B) in relation to the length of time in the homeless role. Hypotheses three (A-B) examine the acceptance of the homeless identity as the dependent variable with length of time and number of times homeless as the independent variables. The next group of hypotheses tests the impact of the homeless identity and social relationships on attempts to exit the homeless situation. The fourth hypothesis examines the acceptance of the homeless identity in relation to attempts made to no longer be homeless. Attempts made to exit homelessness is the dependent variable for hypotheses five (A-D), while (interactive and affective) commitment, salience and centrality to the homeless identity are the independent variables. With hypothesis six, the correlation of family support and attempts to leave the homeless role is investigated. 61 Self-esteem is examined as the outcome for hypotheses seven through ten. Hypotheses seven (A-B) attempt to verify that length of time and number of times homeless have an effect on self-esteem. Next, hypothesis eight assesses the impact that accepting the homeless identity has on self-esteem. For hypothesis nine (A-D), (interactive and affective) commitment, salience and centrality of the homeless identity are the independent variables while self-esteem is the dependent variable. The final self-esteem hypothesis examines the support of significant others such as family members on the self-esteem of their homeless relatives. Self-efficacy is measured similarly to self-esteem with hypotheses eleven through fourteen. The independent variables are acceptance of the homeless identity (H11), acceptance of the homeless identity (H12), (interactive and affective) commitment, salience and centrality (H13A-D), and length of time and number of times homeless (H14A-B) while self-efficacy is the dependent variables. The last set of hypotheses analyzes homeless identity meanings and behaviors. First, hypotheses fifteen (A-G) measure the correlation between homeless identity meanings and behaviors. Hypotheses sixteen (A-D) investigate (interactive and affective) commitment, salience and centrality of the homeless identity as having a set of identity meanings attached to them. Hypotheses seventeen (A-B) examine homeless identity meanings as the dependent variable and length of time and number of times homeless as the independent variables. The next set of hypotheses test self-esteem (H18A) and self-efficacy (H18B) in relation to the homeless identity meanings. Hypotheses nineteen (A-D) examine the salience, centrality and (interactive and affective) commitment to the homeless identity as expressed through the behaviors attached. Finally, the last hypotheses address the relationship between staying sober and self-esteem (H20A) and self-efficacy (H20B). 62 The independent variables are hypothesized to cause the dependent variable based on theoretical assumptions. However, it is important to note that these causal relationships can also be in the opposite direction. Thus, not only will accepting the homeless identity effect the length of time and number of times homeless but time homeless will also predict accepting the homeless identity. SUMMARY A diverse sample of 326 homeless people were taken at six different service agencies providing meals, supportive services, shelter beds and housing programs. Although not a random sample, the survey sample did closely reflect the known homeless population in this metropolitan city. The stratified sample was based on aspects of the local homeless population regarding race/ethnicity, gender, age, household type and veteran status. A ten page self-administered standardized questionnaire was utilized for this study. Areas of focus for the survey included demographics, homeless history and characteristics, homeless identity, self-esteem and self-efficacy, identity meaning and behavior, family support and attempts to exit the homeless situation. The data was collected during the summer of 2008. The analysis of the data was conducted using SPSS. Frequency distributions were tabulated for all the variables in the study. Multiple regressions were run on the twenty hypotheses. An OLS was conducted due to the dependent variables being continuous. This study ran two models for each hypothesis. Model one provided a baseline for the stepwise regression addition method based on theoretical and predictive reasons. The second model was run with the control variables of race/ethnicity, gender, relationship status, military status, parent, family members, education 63 level, age group, years in Atlanta, last month income, sleeping location and number of disabilities. 64 CHAPTER 6 CHARACTERISTICS OF THE HOMELESS SAMPLE The study survey is comprised of eight parts: 1) demographic characteristics, 2) identity centrality and identity salience, 3) self-esteem and self-efficacy, 4) homeless history, characteristics and identity, 5) affective commitment and interactive commitment to the homeless identities, 6) identity meaning and behaviors, 7) exiting the homeless situation and family support, and 8) the experience of the respondents while homeless. Descriptive statistics have been calculated on all the variables for each part of the survey. The frequency results are discussed in this chapter. DEMOGRAPHICS The respondents are overwhelmingly single, middle-aged black men (see Table 1). The dissertation study’s demographic profile is similar to that of the city and two counties’ (Tri-J) homeless survey (Parker 2009; see Table 6.1). These findings reflect a homeless population who live in an urban area and are literally homeless. As previously indicated, the majority of the study respondents are African American (82 percent; see Table 6.1). A much smaller number are white, non-Hispanic (11 percent). In comparison, the racial composition of the general city residents is slightly more than half black (54 percent) and slightly more than a third white (36 percent; U.S. Census Bureau 2010). Thus, the study population shows an overrepresentation of African Americans as compared to the residents of the city. 65 Table 6.1: Demographics as Compared to the 2009 City Homeless Survey: Numbers, Percentages, and Standard Deviations for Race/Ethnicity, Gender, Military Service, Age and Living with Family Members Dissertation Study 2009 Tri-J Survey City Residents Variable N Percentage N Percentage N Percentage Race/Ethnicity Black 265 82 433 87 226,802 54 Other race 58 18 63 13 193,201 46 Total 323 100 496 100 420,003 100 SD = 0.38 Gender Male 287 89 396 84 210,001 50 Female 34 11 78 16 210,002 50 Total 321 100 474 100 420,003 100 SD = 0.31 Veteran No 245 75 383 78 390,603 93 Yes 80 25 106 22 29,400 7 Total 325 100 489 100 420,003 100 SD = 0.43 Age Under age 17 1 0 0 0 29,400 7 18 – 24 12 4 7 2 67,200 16 25 – 34 33 11 49 11 96,601 23 35 – 44 82 26 76 17 96,601 16 45 – 54 133 42 208 47 96,601 16 55 – 64 51 16 99 22 43,200 11 65 or older 4 1 7 1 43,200 11 Total 316 100 446 100 420,003 100 SD = 10.38 Living with family members No 395 91 51 90 163,801 39 Yes 29 9 446 10 256,202 61 Total 324 100 497 100 420,003 100 SD = 0.29 66 In the city, over a quarter of the residents (27 percent) are between the ages of 45 to 64 (U.S. Census Bureau 2010; see Table 6.1). However, the dissertation study’s homeless respondents between the ages of 45 to 64 are the majority of the population (58 percent). The range of people between the ages of 25 and 44 is similar for the study respondents (37 percent) as the city residents (39 percent; U.S. Census Bureau 2010). However, young adults between the ages of 18 and 24 are underrepresented among the study respondents (4 percent) as compared to the residents of the city (16 percent; U.S. Census 2010). The average age for the city residents is 33 years (U.S. Census Bureau 2010) while the average age for the study respondents is 46 years. This indicates a homeless population that is skewed older than the city residents. With gender, the city residents are evenly split between males and females (U.S. Census Bureau 2010; see Table 6.1). The study respondents, on the other hand, have an overrepresentation of men (89 percent). Along the same line, the number of study respondents who served in the military (25 percent) is higher than that of the city (7 percent) and U.S. population (10 percent; U.S. Census Bureau 2010). For this study, the majority of the respondents are unaccompanied (91 percent; see Table 6.1). Predominately, they are single, separated, divorced or widowed (85 percent; see Table 6.2). This indicates a population that probably has difficulty with being in a committed relationship. Even though the majority of respondents are single, over half are parents and may have solid relationships with their children and possibly other family members. The majority of study respondents (65 percent) have a high school diploma / GED or less (see Table 6.2). In comparison, almost half of the adult city residents (46 percent) have a college degree (U.S. Census Bureau 2010). In 2005 for the Tri-J homeless survey, the respondents indicated that their most frequent types of employment were labor jobs such as construction, 67 Table 6.2: Demographics: Numbers, Percentages, and Standard Deviations for Relationship Status, Parent, Education Level, Length of Time Living in Atlanta and Income for Previous Month Variable N Percentage Relationship status Single and/or dating 205 63 Separated, divorce or widowed 71 22 Committed relationship / Married 48 15 Total 324 100 SD = 0.90 Parent No 136 42 Yes 189 58 Total 325 100 SD = 0.49 Education level 11th grade or less (no diploma) 75 23 High school diploma or GED 135 42 Some college (no diploma) 49 15 Technical, 2 or 4 year degree 53 16 Some graduate school or grad degree 12 4 Total 324 100 SD = 1.98 Length of time living in Atlanta Less than 1 month 21 7 1 – 6 months 31 10 7 months – 1 year 20 6 1 year – 4 years 44 14 5 – 9 years 45 14 10 – 15 years 42 13 More than 15 years 115 35 Total 318 100 SD = 33.53 Income for previous month $0 136 43 $1 - $250 69 22 $251 - $500 33 10 $501 - $750 36 11 $751 - $1,000 23 7 $1,001 - $1,500 14 4 Over $1,501 8 3 Total 319 100 SD = 3.95 68 landscaping and custodial. On the other hand, professional jobs which often require a college education were rarely indicated (Ashley, Johnson and Pandey 2005). The lack of a stable well paying job is also reflected by the majority (93 percent) of the dissertation study respondents making $1,000 or less in the month prior to the survey (see Table 6.2) which is less than $12,000 a year. According to the U.S. Department of Health and Human Services (2010), a person making $10,830 or less a year is considered to be living in poverty. Finally, the study respondents are mainly local, long-term residents with most having lived in the city for more than five years (62 percent; see Table 6.2). This indicates a community that could benefit from preventative services as well as homeless services. As for recent arrivals to the city, less than a fifth of the respondents have lived here for six months or less. According to the 2009 Tri-J homeless survey, the top reason for moving to the area was for job opportunities (Parker 2009). HOMELESS CHARACTERISTICS AND IDENTITY The respondents are asked in the dissertation survey about their main reasons for becoming homeless (see Table 6.3). They are allowed to provide multiple reasons. On average, the respondents indicate three causes for their homelessness. For example, several respondents indicate drug abuse, unemployment and the inability to stay with their families as the reasons. These multiple factors could be interrelated such that the drug abuse impacts the ability of the respondents to maintain a job and stay with their families long term. An unemployment category can be created by combining the economic reasons of fired / quit job and laid off work, and then de-duplicating the two causes for the respondents. The new unemployment (60 percent) category thus becomes the main reason for respondents becoming 69 Table 6.3: Causes of Homelessness: Numbers and Percentages for Economic Reasons, Health Reasons, Family Reasons, Housing Reasons and Other Reasons* Variable N Percentage Economic reasons Fired or quit job 157 48 Unable to pay rent 85 26 Laid off work 69 21 Not making enough money 52 16 No jobs for skill level 40 12 Government cutbacks 17 5 Health reasons Alcohol or drug abuse 63 19 Mental illness 41 13 Medical / health problems 41 13 Physically disability 35 11 Family reasons Can’t stay with family or friends 52 16 Divorced or separated 27 8 Death in family 24 7 Family violence 16 5 Fight with family 5 2 Housing reasons Can’t find affordable housing 38 12 Public or section 8 housing not available 17 5 Lost housing due to non-economic reasons 13 4 Lost public or section 8 housing 10 3 Other reasons Bad luck 51 16 Relocation 42 13 Personal choice 39 12 Hurricane Katrina 8 3 Prison, jail or criminal background 7 2 *Multiple answers were allowed 70 and in emergency shelters depending on the weather and if they have money to pay for a night at the emergency shelter. Less than a fifth (16 percent) stay in transitional housing or in treatment programs. This group is receiving supportive services such as job placement or drug treatment along with a bed. Some of the folks (13 percent) are sleeping anywhere they can find a place. A few of the respondents (4 percent) indicate they are living in hotel or motel rooms which were actually designed for people to stay on a temporary basis for vacations or business trips and not long term. Finally, a group of people state that they are living with family or friends in their apartments or homes (10 percent). This means that the respondents are sleeping doubled up, often on couches (known as “couch surfing”) and are not legally on the lease if the home is rented. For the analysis, the sleeping locations are grouped into two categories: unsheltered and sheltered locations. As can be seen by the sleeping locations, none of the respondents have their own permanent place to live and are thus considered to be without their own home. homeless (see Table 6.3). Often the lack of a job results in the respondents not having enough money (16 percent) to pay bills such as rent (26 percent). Besides economic reasons, health problems are another major cause of homelessness (see Table 6.3). Alcohol and drug abuse (19 percent) is the most common health concern among this homeless population with mental illness (13 percent) another major health issue. Problems with family members such as fighting, violence, death in family and the inability to stay with family can also cause people to become homeless (28 percent after de-duplication). The main issue that the homeless respondents face with their family is not being able to live with them (16 percent). Reasons include family members living in Section 8 or public housing whereby the rules do not allow for additional people not on the lease to stay. In addition, family members may kick out 71 others who are abusing drugs or alcohol or who have mental health problems because the individuals are difficult to live with on a regular basis. Several other factors are mentioned as reasons for becoming homeless including bad luck (16 percent), relocation (13 percent) and personal choice (12 percent; see Table 6.3). Bad luck as an answer signifies that the respondents believe their homelessness is by chance and thus beyond their control. This is in contrast to the people who indicate that becoming homeless is their own personal decision. By relocating, it appears that the respondents believed that they had better options in this particular city than where they were before. All in all, structural factors such as economics and personal problems like health concerns are both found to be instrumental in causing people to become homeless. As shown in Table 6.4, over a third of the study respondents (34 percent) report staying in emergency shelters while almost a quarter (23 percent) are sleeping in unsheltered locations such as on the street or at the airport. Often people move back and forth between sleeping outdoors and in emergency shelters depending on the weather and if they have money to pay for a night at the emergency shelter. Less than a fifth (16 percent) stay in transitional housing or in treatment programs. This group is receiving supportive services such as job placement or drug treatment along with a bed. Some of the folks (13 percent) are sleeping anywhere they can find a place. A few of the respondents (4 percent) indicate they are living in hotel or motel rooms which were actually designed for people to stay on a temporary basis for vacations or business trips and not long term. Finally, a group of people state that they are living with family or friends in their apartments or homes (10 percent). This means that the respondents are sleeping doubled up, often on couches (known as “couch surfing”) and are not legally on the lease if the home is rented. For the analysis, the sleeping locations are grouped into two categories: unsheltered and 72 Table 6.4: Homeless History: Numbers, Percentages, and Standard Deviations for Usually Slept at Night Over Past Month, Different Times Homeless, Current Length of Homelessness and Total Length of Homelessness within Past Three Years Variable N Percentage Usually slept at night over past month Unsheltered 74 23 Emergency shelter 109 34 Transitional housing program 50 16 Anywhere I can 41 13 Hotel 13 4 House or apt. of family or friend 32 10 Total 319 100 SD = 1.54 Current length of homelessness Less than 1 month 42 14 1 – 3 months 70 23 4 – 6 months 43 14 7 – 11 months 28 9 1 year 35 11 2 years 26 8 3 years 18 6 4 years 15 5 5 years or more 32 10 Total 309 100 SD = 19.45 Total length of homeless within past 3 years Less than 1 month 31 10 1 – 3 months 51 17 4 – 6 months 49 17 7 – 11 months 25 8 1 year 43 14 2 years 40 13 3 years 63 21 Total 302 100 SD = 13.12 Different times homeless within past 3 years 1 time 156 50 2 times 64 20 3 times 40 13 4 times 15 5 5 or more times 37 12 Total 312 100 SD = 1.37 73 sheltered locations. As can be seen by the sleeping locations, none of the respondents have their own permanent place to live and are thus considered to be without their own home. For half of the study respondents, this is their first time homeless while the other half have experienced multiple homelessness over the past three years (see Table 6.4). Being homeless multiple times indicates a population that has suffered from episodic homelessness. They attempt to stay in housing but are unable to permanently. Of particular concern are the respondents who have been homeless five or more times over the last three years (12 percent), moving in and out of homelessness at least twice a year during that time. Almost half of the study respondents (44 percent) have been homeless for a total length of time of six months or less during the past three years (see Table 6.4). People who have been homeless for a short period of time are not as entrenched in homelessness as those who have been homeless long term. They can see being homeless as a short-term, temporary situation. Long term homelessness is often specified as one year or more. Accordingly, less than half of the respondents (40 percent) have been continuously homeless for one year or more. This long term homeless population struggles to leave the homeless situation, especially people homeless five years or more (10 percent). The majority of the study respondents (73 percent) suffer from at least one disabling condition as shown in Table 6.5. In comparison, the Tri-J homeless survey found 63 percent of the respondents experienced at least one disabling condition (Parker 2009). The top disability for this study is depression (35 percent) with drug addiction a close second (29 percent) and alcoholism third (26 percent). They are followed by physical disabilities such as back and leg problems, severe mental illness such as bipolar disorder and post traumatic stress disorder, anxiety, chronic health problems such as asthma and high blood pressure, and HIV/AIDS. The 74 Table 6.5: Numbers, Percentages, Type and Standard Deviations of Disabilities Variable N Percentage Number of disabilities No disabilities 86 27 1 disability 122 39 2 disabilities 68 22 3 disabilities 29 9 4 - 5 disabilities 9 3 Total 314 100 SD = 1.04 Type of disabilities* Depression 113 35 Drug abuse 93 29 Alcohol abuse 86 26 Physical disability 63 19 Mental illness 53 16 Anxiety 53 16 Chronic health problems 32 10 HIV/AIDS 19 6 No disabilities 86 27 *Multiple answers were allowed 75 disabilities can be grouped into four categories and then de-duplicated for similar disabling conditions. Thus, mental health including depression and anxiety is still the dominant issue (42 percent), followed closely by substance dependence including alcohol and drug abuse (41 percent), with physical disabilities a distant third (19 percent) and severe health problems including HIV/AIDS fourth (15 percent). By comparing respondents in my sample to the U.S. population, we can understand the magnitude of the problem. For example, 22 percent of the U.S. population experiences a serious debilitating mental illness (National Institute of Mental Health 2010) while 42 percent of the study population has a mental disorder. These figures indicate a population that is disproportionately suffering from mental health problems as compared to the national level. On the other hand, respondents in the sample studied in this dissertation report a similar number of physical disabilities as the national average. According to U.S. Census Bureau, 17 percent of the U.S. population has a disability that limits their mobility while my study finds 19 percent of the respondents with a physical disability (Brault 2008). To determine whether people are chronically homeless, questions one and three through seven from Part IV of the survey are utilized. These questions capture the HUD definition of chronic homelessness (i.e., an unaccompanied individual who suffers from a disability and who has either been continuously homeless for a year or more, or has experienced at least four episodes of homelessness over the past year [HUD 2007]). According to my calculations, 23 percent of the study population meets the definition of chronic homelessness. This is in exact agreement with the 2009 city and counties’ homeless survey (Parker 2009), which also found 23 percent of the respondents were chronically homeless. The chronic homeless figure is concerning because this population is one of the most difficult to get out of the situation and into permanent 76 housing due to the fact that not only are they facing long term homelessness but are also suffering from a disabling condition. Table 6.6 indicates that when asked directly (i.e. Do you identify as a homeless person?) the majority of respondents (61 percent) identify or strongly identify as homeless. On the other hand, less than a quarter state that they do not identify as homeless (23 percent) with almost a fifth indicating that they are unsure if they identify as homeless or not (16 percent). What is interesting is that all the respondents are considered homeless by researchers, policy makers and service providers based on sleeping location. Specifically, of the people who do not identify as homeless, many are staying at emergency shelters (30 percent) with the next largest group sleeping outdoors (21 percent). These findings therefore indicate that even if people label others as homeless, it does not mean that the individuals themselves will wholeheartedly accept that identity. IDENTITY THEORY The 289 respondents who answered the centrality of the homeless identity question correctly was much fewer than most of the other survey items. One reason was that many of the respondents had difficulty ranking the role identities overall from 1 (most important) to 5 (fifth most important). Instead twenty six of the respondents ranked each role identity separately. In addition, eight respondents did not answer this question. Of all the roles listed for the centrality identities (see Table 6.7), being the member of a family is chosen as important for the majority of the respondents (78 percent). In fact, almost half ranked family membership as their most important identity (47 percent; see Table 6.8). This indicates that the family is critical even for those who state that family problems were a reason 77 Table 6.6: Numbers, Percentages, and Standard Deviations for Acceptance of the Homeless Identity Variable N Percentage Acceptance of homeless identity Strongly identify 89 28 Identify 105 33 Not sure if identify 52 16 Do not identify 72 23 Total 318 100 SD = 1.11 78 Table 6.7: Centrality to the Homeless Identity: Numbers, Percentages, Means, and Standard Deviations of respondents who ranked the role as an important identity at some level* Variable N Percentage Mean SD Roles Family member (parent, child, sibling, etc.) 224 78 3.28 2.05 Worker 174 60 1.72 1.72 Friend 162 56 1.70 1.74 Relationship status (married, dating, single, etc.) 138 48 1.55 1.84 Religious affiliation 127 44 1.35 1.86 Educational level 92 32 0.81 1.38 Physically healthy 83 29 0.83 1.49 Homeless 60 21 0.56 1.26 Race / ethnicity 51 18 0.48 1.11 Gender 48 17 0.45 1.10 Mentally healthy 46 16 0.44 1.15 Age 34 12 0.25 0.81 Student 33 11 0.27 0.86 Veteran 31 11 0.28 0.93 Sexual orientation 28 10 0.20 0.73 Alcoholic or addict 18 6 0.12 0.59 *All items were scored by ranking: most important identity = 5, second most important identity = 4, third most important identity = 3, fourth most important identity = 2, fifth most important identity = 1 and not ranked = 0. 79 Table 6.8: Centrality to the Homeless Identity by Roles: Numbers and Percentages for Family member, Worker, Friend, Relationship Status, Religious, Physically Healthy, Education Level and Homeless* Variable N Percentage Variable N Percentage Family member Worker 1st 135 47 1st 21 7 2nd 47 16 2nd 36 13 3rd 12 4 3rd 46 16 4th 19 6 4th 38 13 5th 11 4 5th 33 11 Not ranked 65 23 Not ranked 115 40 Friend Relationship status 1st 20 7 1st 13 5 2nd 36 13 2nd 64 22 3rd 52 18 3rd 29 10 4th 33 11 4th 8 3 5th 24 8 5th 24 8 Not ranked 124 43 Not ranked 151 52 Religious Physically healthy 1st 40 14 1st 12 4 2nd 16 6 2nd 17 6 3rd 16 6 3rd 19 7 4th 24 8 4th 19 6 5th 31 10 5th 16 6 Not ranked 162 56 Not ranked 206 71 Education level Homeless 1st 9 3 1st 9 3 2nd 12 4 2nd 10 3 3rd 21 8 3rd 10 3 4th 27 9 4th 16 6 5th 23 8 5th 15 6 Not ranked 197 68 Not ranked 229 79 N = 289 *All items were scored by ranking: most important identity = 5, second most important identity = 4, third most important identity = 3, fourth most important identity = 2, fifth most important identity = 1 and not ranked = 0. 80 for their becoming homeless. Over half of the 2007 local Tri-J homeless survey (Massey, Runkle and Parker 2008) respondents (56 percent) indicated that they had contact with an adult family member sometime during the previous week of the questionnaire. In addition, a third of the respondents had contacted their family twelve or more times during the month prior to the survey. These findings indicate that they still view family membership as a central role, if not the most important, even while homeless. The second most chosen identity overall for centrality is worker (60 percent; see Table 6.7). Only a small number of respondents (7 percent) list it as the most important however (see Table 6.8). The worker identity is frequently ranked as the third most important identity right after the friend identity (see Table 6.8). As previously stated, few of the respondents had a permanent full time job at the time of the survey. The worker identity could then be the reflection of a past and future identity that is still of significance for this population because money is needed to obtain and maintain permanent housing and thus no longer be homeless. The third most picked identity overall is friend (56 percent; see Table 6.7) and is ranked by more respondents as the third most important identity (18 percent; see Table 6.8). When asked in 2007 by the Tri-J homeless survey (Massey, Runkle and Parker 2008) who the respondents rely on for emotional support, the most frequent response wasanother homeless person. In that same survey, the respondents indicated that the persons they spend most of their time with are other homeless individuals. These findings suggest that even though family is most important, homeless people also rely heavily on other homeless individuals during this tough situation. Relationship status is the fourth most checked role (48 percent; see Table 6.7) and is ranked by more individuals as the second most important identity (22 percent; see Table 6.8). This finding is interesting because most of the respondents are single (85 percent). Perhaps this 81 population is still quite interested in being in a committed relationship even while they are homeless. The fifth top ranked identity is religion (39 percent; see Table 6.7), and it is frequently ranked by the respondents as the most important identity right after the family identity (see Table 6.8). This finding could be a result of the city being located in the Bible belt of the south. As for a homeless identity, a fifth of the respondents (21 percent) indicate that being homeless is one of their top five identities (see Table 6.7). Only nine respondents rank the homeless identity as their most important (3 percent; see Table 6.8). The interesting finding is that when asked directly if they identify as homeless, then over half agree. However, when asked to rank the homeless identity among multiple roles, the identity is not chosen often and not highly ranked (18 percent). This could be due to the respondents believing that this is a negative, stigmatizing identity due to a temporary difficult situation. The identity chosen least often is alcoholic / addict (6 percent; see Table 6.7). Only two people ranked it as their most important identity (1 percent; see Table 6.9). As with the homeless identity, being an alcoholic or addict is a negative and stigmatizing identity. In addition, people who are alcoholics or addicts are thought to be so for life, according to Alcoholics and Narcotics Anonymous programs, and must drastically change their behavior to become sober and maintain their sobriety. Salience is based on the tendency to invoke the same identity in different situations. For this study, invoking is measured by discussing the same identity more often in the various situations. This results in stability of the identity across time and situation (Serpe 1987). The situations include hanging out with friends, spending time with family, when at a service provider agency, when at a doctor’s office and when at a shelter. On the other hand, when 82 Table 6.9: Centrality to the Homeless Identity by Roles: Numbers and Percentages for Race/Ethnicity, Gender, Mentally Healthy, Age, Student, Veteran, Sexual Orientation and Alcoholic/Addict* Variable N Percentage Variable N Percentage Race/Ethnicity Gender 1st 2 1 1st 2 1 2nd 9 3 2nd 8 3 3rd 16 6 3rd 19 6 4th 20 7 4th 12 4 5th 4 1 5th 7 3 Not ranked 238 82 Not ranked 241 83 Mentally Healthy Age 1st 7 2 1st 3 1 2nd 8 3 2nd 2 1 3rd 8 3 3rd 6 2 4th 14 5 4th 7 2 5th 9 3 5th 16 6 Not ranked 243 84 Not ranked 255 88 Student Veteran 1st 2 1 1st 4 1 2nd 5 2 2nd 5 2 3rd 6 2 3rd 7 2 4th 10 4 4th 5 2 5th 10 4 5th 10 4 Not ranked 256 87 Not ranked 258 89 Sexual orientation Alcoholic/Addict 1st 0 0 1st 2 1 2nd 5 1 2nd 1 0 3rd 3 1 3rd 2 1 4th 10 4 4th 2 1 5th 10 4 5th 11 4 Not ranked 261 90 Not ranked 271 93 N = 289 *All items were scored by ranking: most important identity = 5, second most important identity = 4, third most important identity = 3, fourth most important identity = 2, fifth most important identity = 1 and not ranked = 0. 83 no identity is salient, then individuals will enact different identities in various situations. For this study, only a small number of respondents do not indicate a salient identity (10 percent). When looking at the five different situations, homelessness is discussed by the respondents more frequently than any of the other roles in the situations (38 percent) and is thus the most salient identity overall (see Table 6.10). The homeless identity is most salient, meaning the identity is discussed in at least two or more situations, for almost half of the respondents (48 percent; see Table 6.11). In contrast, homelessness is the eighth most important identity overall for centrality while the most central identity is family. A possible reason for the high salience of the homeless identity, even though it is a negative stigmatizing identity, is that people who are homeless must identify as such to receive services whether the assistance is from a provider agency, housing facility or family members. In addition, when hanging out with friends who are also homeless, it is likely that their situation will be discussed. Also, the homeless identity is not often one that can be easily hidden from others because people living on the streets will tend to have dirty clothes and an unkempt look while people in transitional housing programs will be seen coming and going from those locations. The second most salient identity overall is worker (30 percent; see Table 6.10). The worker identity is also the second most central identity. Specifically, the worker identity is salient, meaning the identity is discussed in at least two or more situations, for almost half of the respondents (42 percent; see Table 6.11). What is interesting is that “at a job” is not one of the situations, yet employment is discussed quite often by the respondents. As with identity centrality, the significance of the worker identity for this population can be seen by its high salience. 84 Table 6.10: Salience to the Homeless Identity: Numbers and Percentages of Respondents who Discussed the Topics Most Often Across the Different Situations Variable N Percentage Topics Homelessness 102 38 Work 82 30 Physical health 23 9 Religion 12 4 Family 9 3 Sports 8 3 Mental health 8 3 Education 5 2 Finances 4 2 Relationships 3 1 Friends 3 1 Alcoholism/addiction 3 1 Politics 2 1 Self 2 1 Life 2 1 Race/Ethnicity 1 0 Total 269 100 85 Table 6.11: Salience to the Homeless Identity: Numbers, Percentages, and Standard Deviances of Homelessness, Work, Physical Health and Religion Variable N Percentage Homelessness Discussed in four or five situations 16 5 Discussed in three situations 33 10 Discussed in two situations 104 33 Discussed in one situation 109 35 Not discussed 54 17 Total 316 100 SD = 1.05 Work Discussed in four or five situations 22 7 Discussed in three situations 39 12 Discussed in two situations 71 22 Discussed in one situation 87 28 Not discussed 97 31 Total 316 100 SD = 1.23 Physical health Discussed in four or five situations 3 1 Discussed in three situations 14 4 Discussed in two situations 38 12 Discussed in one situation 205 65 Not discussed 56 18 Total 316 100 SD = 0.75 Religion Discussed in four or five situations 0 0 Discussed in three situations 6 2 Discussed in two situations 22 7 Discussed in one situation 75 24 Not discussed 213 67 Total 316 100 SD = 0.71 *All items were scored by ranking: discussed in 4 or 5 situations = 4, discussed in 3 situations = 3, discussed in 2 situations = 2, discussed in 1 situation = 1 and not discussed = 0. 86 Physical health is a third identity overall that is frequently talked about in the situations (9 percent; see Table 6.10). As a central identity, it is chosen as the seventh most important identity overall. A possible reason for the increased salience of the identity is that one of the situations is a visit to the doctor’s office whereby physical health is the primary topic of conversation for the majority of the respondents (68 percent). Identity theory uses two measurements for commitment: affective and interactive. Affective commitment to the homeless identity is the strength of ties that homeless people have with other homeless people. For this study, ten statements are used to calculate that commitment such as “my homeless friends are a real source of comfort to me” (see Table 6.12). The answers from the affective commitment homeless identity statements are totaled and range from 0 to 30. Scores between 15 and 25 are within the middle range while scores 14 and below indicate a low commitment to other homeless people. Internal reliability of the multiple items in the index is measured with Cronbach’s Alpha. For the alpha, a score of 0.70 or higher on a scale of 0 to 1 suggests that the index is measuring the same thing and thus reliable (Knoke, Bohrnstedt and Mee 2002; Vogt 1999). Affective commitment homeless identity has a Cronbach’s Alpha score of 0.77. Based on the total scores, over half of the respondents (53 percent) indicate that they have a low affective commitment to other homeless people with only a small number (3 percent) expressing a high level of commitment. The rest are within the middle range. Therefore, the majority of the respondents show a low affective attachment to other people who are homeless and thus the homeless identity. Of the ten statements for the affective commitment homeless identity, the survey respondents only agree with three (see Table 6.12). These statements indicate that some of the respondents have made good friends while homeless (mean = 1.52) and feel that 87 Table 6.12: Means and Standard Deviations for Affective Commitment to the Homeless Identity* Variable Mean SD I enjoy and value the social ties and contacts that I’ve made 1.60 0.97 as a homeless person. Most other homeless people don’t treat me well.** 1.56 0.87 I haven’t made good friends with others who are homeless.** 1.52 0.93 People who are homeless understand me better than most other 1.43 0.95 people do. I get the emotional help and support I need from my friends 1.40 0.98 who are homeless. My homeless friends are a real source of comfort to me. 1.39 0.90 I can’t depend on my homeless friends when things go wrong.** 1.38 1.02 I’m happy when I’m with friends who are also homeless. 1.34 0.96 I don’t feel connected to others who are homeless.** 1.30 0.94 I count on my homeless friends when I get sick. 1.03 0.89 Alpha = 0.77 N = 322 *All items were scored on a four point scale: strongly agree = 3; agree = 2; disagree = 1; and strongly disagree = 0. **Items recoded in positive direction. 88 they can “enjoy and value the social ties” that they’ve made while homeless (mean = 1.60; see Table 6.12). Additionally, they believe that they will be treated well by other people who are in the same situation (mean = 1.56). On the other hand, a majority of the respondents do not agree with seven of the ten affective commitment homeless identity statements (see Table 6.12). Even though the respondents have developed good friendships with other homeless people, these relationships do not make them happy (mean = 1.34) and are not a real source of comfort (mean = 1.39). Most often their emotional help and support does not come from other people who are in the same situation (mean = 1.40). This shows a population that cannot depend on other homeless people when things go wrong (mean = 1.38). In fact, in their most vulnerable state of being sick, they tend not to rely on their homeless friends (mean = 1.03) but on their housed family members (Massey, Runkle and Parker 2008). Finally, interactive commitment, unlike affective commitment, is based on the number of social ties with other homeless people. Accordingly, about half of the respondents (46 percent) report having either no friends or few friends (1-2 friends) who are homeless (see Table 6.13). Only about a quarter of this population (22 percent) suggest that they have a high number of friends (11 or more friends) who are also homeless. Similarly to affective commitment, these findings indicate a low commitment to the homeless identity. SELF-ESTEEM AND SELF-EFFICACY Self-esteem is assessed using the Rosenberg Self-Esteem Scale (1989) which consists of ten items such as “I feel that I'm a person of worth, at least on an equal plane with others” (see Table 6.14). The items are measured on a four point Likert scale ranging from strongly agree to 89 Table 6.13: Numbers and Percentages for Interactive Commitment to the Homeless Identity Variable N Percentage Number of homeless friends No friends 84 27 1 – 2 friends 59 19 3 – 5 friends 67 21 6 – 10 friends 35 11 11 – 15 friends 10 3 16 or more friends 59 19 Total 314 100 90 strongly disagree. The responses from the ten statements are totaled and range from 0 to 30. Scores between 15 and 25 are within the middle range while scores 14 and below indicate low self-esteem. The alpha score is 0.81 which indicates a high reliability. The average self-esteem score for this homeless population is 20.60. The majority of respondents (65 percent) indicate a self-esteem score in the middle range. Only a small number of people demonstrate low self-esteem (9 percent) with over a quarter showing high self-esteem (26 percent). This finding is contrary to past research which implies that most homeless people demonstrate low self-esteem (Diblasio and Belcher 1993). Of the self-esteem statements, nine of the ten item means are closer to either “agree” or “strongly agree” than to “disagree” (see Table 6.14). The one with the highest mean score indicates that the respondents feel that they have good qualities (mean = 2.62), no matter their situation. Two other statements with high averages include feeling that they are people of worth, at least on an equal plane with others, (mean = 2.47) and believing that they “do things as well as most other people” (mean = 2.41). These latter two statements are probably based on in group comparisons with other homeless people. The next statement shows that the respondents have a positive attitude towards themselves (mean = 2.40). The item with the lowest average, on the other hand, indicates that the respondents want to respect themselves more (mean = 1.39). Six additional statements are provided to calculate self-efficacy including “I can pretty well control things that happen to me” (see Table 6.15). They are also evaluated on the same four point Likert scale as self-esteem. The answers are totaled and the scores range from 0 to 18. Scores between 9 and 15 are within the middle range. Scores 8 and below indicate low self-efficacy. The Cronbach’s Alpha score for self-efficacy is lower than the self-esteem score at a 91 Table 6.14: Means and Standard Deviations for Self-Esteem* Variable Mean SD I feel that I have a number of good qualities. 2.62 0.59 I feel that I’m a person of worth, at least on an equal plan with others. 2.47 0.68 I am able to do things as well as most other people. 2.41 0.68 I take a positive attitude toward myself. 2.40 0.70 All in all, I am inclined to feel that I am a failure.** 2.11 0.91 I feel I do not have much to be proud of.** 1.88 1.01 At times I think I am no good at all.** 1.88 1.00 On the whole, I am satisfied with myself. 1.82 0.95 I certainly feel useless at times.** 1.63 1.01 I wish I could have more respect for myself.** 1.39 1.02 Alpha = 0.80 N = 324 *All items were scored on a four point scale: strongly agree = 3, agree = 2, disagree = 1 and strongly disagree = 0. **Items recoded in positive direction. 92 Table 6.15: Means and Standard Deviations for Self-Efficacy* Variable Mean SD What happens to me in the future mostly depends on me. 2.38 0.76 I can do just about anything I really set my mind to. 2.33 0.70 I can pretty well control things that happen to me. 1.85 0.85 When trying to learn something new, I soon give up if I am 1.83 0.94 not initially successful.** I often feel helpless in dealing with the problems of life.** 1.55 0.94 Sometimes I feel that I’m being pushed around.** 1.51 0.91 Alpha = 0.62 N = 323 *All items were scored on a four point scale: strongly agree = 3, agree = 2, disagree = 1, and strongly disagree = 0. **Items recoded in positive direction. 93 reliability level of 0.64. The original self-efficacy measurement was a ten item scale. By reducing the scale to six items for this study, it probably reduced the internal consistency. As with self-esteem, the majority of respondents (81 percent) indicate a self-efficacy score in the middle range, although at a higher percentage. There are more respondents with a low self-efficacy score (11 percent) as compared to self-esteem, while there are fewer people with high self-efficacy (8 percent). This can be seen by the mean score for self-efficacy being 11.45, based on an 18 point scale. All six of the self-efficacy statement means are closer to “agree” than “disagree” (see Table 6.15). An interesting finding is that the positive items had the three highest mean scores while the negative statements had the three lowest averages. The self-efficacy items with the highest means indicate that the respondents feel that they are in control of themselves (mean = 1.85), accomplishing what they set their minds to (mean = 2.33) and what happens in the future (mean = 2.38). Additionally, this homeless population does not give up when trying new things (mean = 1.83) and feels hopeful in dealing with life’s problems (mean = 1.55) such as getting out of their homeless situation. MEANINGS AND BEHAVIORS OF THE HOMELESS IDENTITY Homeless identity meanings are determined by asking respondents to select their preferred descriptor in a series of adjective pairs: honest-dishonest, friendly-unfriendly, hardworking-lazy, resourceful-not capable, independent-dependent, kind-mean and motivated-unmotivated (Boydell, Goering and Morrell-Bellai 2000). The items are measured using a five point semantic differential format from very positive to very negative (Osgood 1976). The responses from the homeless identity meaning pairs are totaled and range from 0 to 28. Scores between 14 and 23 are within the middle range while scores 13 and below indicate low or negative homeless 94 identity meanings. The Cronbach’s Alpha score for the homeless identity meanings scale is 0.83, indicating a strong internal reliability for the index. The average homeless identity meanings score for this population is 21.45. The majority of respondents (68 percent) indicate a homeless identity meanings score in the middle range. Only a small number of people demonstrate low homeless identity meanings (3 percent) with over a quarter showing high identity meanings (29 percent). Of the seven adjective pairs, the majority of the respondents identify themselves in positive terms (see Table 6.16). The respondents view themselves as hardworking (mean = 3.18), capable (mean = 3.09) and motivated (mean = 3.05). These descriptions of themselves probably relate to the significance of the worker status in our society. This can be seen with the worker identity being highly ranked for centrality and salience (see Tables 6.7 and 6.10) by the respondents even though they are not employed. Other adjective pairs indicate that the respondents believe themselves to be honest (mean = 3.08), kind (mean = 3.07) and friendly (mean = 3.05). This would be an advantage when seeking services because agency staff might provide better treatment to people who they believe are being nice versus those who they feel are acting rudely (Miller and Keys 2001). In addition, this would help in building relationships and a strong support network with other people who are homeless. The lowest average score is for the adjective pair independent/dependent (mean = 2.93). While people are in the homeless situation, they often must rely on others to get their basic needs met such as food, clothing and a place to sleep. For these respondents who are mostly men, being dependent on others can be a difficult situation when they have been traditionally socialized to be self-sufficient (Morris 1998). 95 Table 6.16: Means and Standard Deviations for Homeless Identity Meaning* Variable Mean SD Adjective pairs Hardworking/Lazy 3.18 0.77 Capable/incapable 3.09 0.87 Honest/dishonest 3.08 0.82 Kind/mean 3.07 0.74 Motivated/Unmotivated 3.05 0.86 Friendly/unfriendly 3.05 0.79 Independent/dependent 2.93 1.00 Alpha = 0.82 N = 304 *All items were scored on a five point scale: very positive = 4, positive = 3, neutral = 2, negative = 1 and very negative = 0. 96 To measure homeless identity behaviors, respondents are asked how important seven activities are to them as homeless people (see Table 6.17). The behaviors include finding a safe place to sleep, telling stories of past success, sharing information and helping other people who are homeless, knowing who to trust, staying sober and knowing which meal sites serve the best tasting food (Cohen 2001). The responses range from very important to not at all important. The behaviors that are very important overall are finding a safe place to sleep (mean = 2.74) and knowing which person to trust (mean = 2.64; see Table 6.17). These two activities indicate how crucial safety is while homeless. The other behaviors are considered important to the respondents. Staying sober has the third highest average (mean = 2.41). It is followed by helping other homeless people (mean = 2.21) and sharing information (mean = 2.13). If the respondents are having a difficult time helping themselves and other housed family members, then it might make them feel good by helping others who are in the same situation. Knowing which meal site serves the best tasting food has the second lowest mean of the activities (mean = 2.05). Perhaps the average is so low because the meal sites don’t serve very good tasting food or that what really matters is just to find a place that serves a meal nearby. The least important activity is telling stories of past successes which demonstrates the lowest mean score (mean = 1.63). A possible reason is that in the past the respondents have not experienced a lot of successes, but mostly failures in their lives. EXITING THE HOMELESS SITUATION According to the respondents, almost half have not attempted to obtain permanent housing in the month prior to the survey, which means that they have not viewed or filled out any applications for an apartment or housing during that time (see Table 18). The main reason for not 97 Table 6.17: Numbers, Percentages, and Means for Homeless Identity Behaviors* Variable N Percentage Variable N Percentage Finding safe place to sleep Knowing which person to trust Very important 241 74 Very important 226 69 Important 60 18 Important 63 19 Not so important 5 2 Not so important 13 20 Not at all important 4 1 Not at all important 8 3 Mean = 2.74 Mean = 2.64 Staying sober Helping other homeless people Very important 182 59 Very important 125 40 Important 89 29 Important 137 44 Not so important 23 7 Not so important 35 12 Not at all important 16 5 Not at all important 13 4 Mean = 2.41 Mean = 2.21 Sharing information with others Meal sites serving the best tasting food Very important 120 39 Very important 126 41 Important 131 42 Important 97 32 Not so important 37 12 Not so important 65 20 Not at all important 22 7 Not at all important 22 7 Mean = 2.13 Mean = 2.05 Telling stories of past success Very important 70 23 Important 93 30 Not so important 108 35 Not at all important 39 12 Mean = 1.63 N = 310 *All items were scored by ranking: very important = 3, important = 2, not so important = 1 and not at all important = 0. 98 attempting to get permanent housing is financial. The financial problems include not making a livable wage (67 percent), having a job that is forty hours a week (45 percent), maintaining a job (33 percent) and obtaining a better paying job (14 percent). Lack of employment and thus ample funds is also the predominate cause for becoming homeless. Without sufficient funds, it is impossible to obtain and maintain permanent housing. Another obstacle to leaving the homeless situation is the difficulty of finding housing that is affordable (52 percent) such as public housing or Section 8 housing (19 percent; see Table 6.18). An interesting finding is that the third most mentioned requirement for exiting homelessness is assistance from God (45 percent). To exit homelessness, respondents also believe that they need adequate transportation (32 percent), because without it, they will not be able to get to work reliably and not be able to look extensively for affordable housing. An interesting psychological issue that needs to be overcome to exit homelessness is a change in attitude (26 percent). Other requirements for leaving the homeless situation include education or job training, legal assistance, assistance with medical needs, drug or alcohol treatment, mental health treatment, counseling or case management, obtaining identification, good luck and obtaining social security benefits. The majority of the respondents (80 percent) indicate that they would not feel guilty if they were able to move into their own place and their homeless friends could not (see Table 6.18). This finding is reflected in over a quarter of the people making three or more attempts to obtain permanent housing during the month prior to taking the survey. Even though there are a number of obstacles to leaving the homeless situation, it does appear that quite a number of the respondents show hope in accomplishing the goal of finding permanent housing. 99 Table 6.18: Numbers, Percentages and Standard Deviations for Exiting the Homeless Situation Variable N Percentage Attempts to obtain permanent housing No attempts 136 48 1 attempt 39 13 2 attempts 33 12 3 attempts 30 11 4 attempts 14 5 5 or more attempts 32 11 Total 284 100 SD = 1.75 Would I feel guilty if move into own place? Yes 57 20 No 227 80 Total 284 100 SD = 0.61 What do I need to exit the homeless situation?* Livable wage 218 67 Affordable housing 169 52 Assistance from God 145 45 Job that is 40 hours a week 146 45 Save money 143 44 Maintain a job 107 33 Transportation 104 32 Change in attitude 83 26 Education or training 69 21 Public or Section 8 housing 62 19 Legal assistance 60 18 Assistance with medical needs 53 16 Drug or alcohol treatment 52 16 Mental health treatment 50 15 Counseling or case management 52 16 Obtain identification 49 15 Good luck 48 15 Better paying job 47 14 Obtain social security benefits 39 12 *Multiple answers were allowed. 100 FAMILY SUPPORT The support of family members for the homeless respondents is measured with five statements ranging from “I don’t get the emotional help and support I need from my family” to “my family helps out when I am in financial need” (see Table 6.19). The responses for family support are totaled and range from 0 to 15. Scores between 7 and 11 are within the middle range while scores 6 and below indicate a low support from family members. The alpha score is 0.80 for family support, indicating a strong internal reliability for the five item index. Almost half of the respondents (46 percent) indicate that they have no to low family support with a small number (12 percent) expressing a high level of family support. The rest are within the middle range. Therefore, based on these findings, the majority of the respondents demonstrate that they receive little support from their family members while homeless. Of the five family support statements, all item means are closer to “disagree” than to “agree.” The two statements with the highest average scores indicate that family members do not provide comfort (mean = 1.43) and do not help out when they become sick (mean = 1.40). This finding is contrary to the 2007 Tri-J homeless survey which found that the local homeless population primarily depend on family members or relatives from they become ill (Massey, Runkle and Parker 2008). The third item indicates that family does not help out their homeless relative financially (mean = 1.37). This is similar to the findings of the 2007 Tri-J homeless survey were only about a third of the respondents receive assistance from their family (Massey, Runkle and Parker 2008). The last two statements demonstrate that the homeless respondents also do not count on family members when things go wrong (mean = 1.33) and that they do not get the emotional support they need from relatives (mean = 1.26). This is not a population that relies on their 101 Table 6.19: Means and Standard Deviations for Family Support* Variable Mean SD My family is not a real source of comfort.** 1.43 1.03 I count on my family when I get sick. 1.40 1.00 My family helps out when I am in financial need. 1.37 1.07 I count on my family when things go wrong. 1.33 0.99 I don’t get the emotional help and support I need 1.26 1.02 from my family.** Alpha = 0.80 N = 304 *All items were scored on a four point scale: strongly agree = 3, agree = 2, disagree = 1, and strongly disagree = 0. **Items recoded in positive direction. 102 family for understanding about their situation. Instead they depend on other homeless people. Most often their emotional support comes from other people who are in the same situation (Massey, Runkle and Parker 2008). Therefore, they can depend on other homeless people more so than family members when it comes to receiving advice and as a support network. HOMELESS EXPERIENCES Finally, the last question of the survey is open ended. This question allows the respondents to share their experience of being homeless in their own words. Out of the 326 respondents, over a third of the population provided a comment (38 percent). The 132 responses were coded and grouped into similar concepts such as religion, disabilities, unmet needs and overall experience while homeless. These concepts were developed into several predominant categories: religious faith, substance abuse, mental health, affordable housing, employment, bad experience and learning experience. The most often discussed topic in the last section of the survey is religious faith (see Table 6.20). After family member, religious affiliation is chosen as the most important identity (see Table 6.7), and eighth overall in the identity centrality hierarchy (see Table 6.8). With the identity salience hierarchy, it is ranked fourth overall (see Table 6.10). For many of this population, their faith provides strength and comfort to help them while in this difficult situation. Whereas the respondents feel that others may judge them, God does not. Due to the bad life choices the respondents believe they have made in the past, they do not feel that they can survive homelessness on their own. Therefore, they place “God in charge”. God “will make the way” for them to no longer be homeless and to “become productive citizens.” 103 Table 6.20: Homeless Experience: Religious Faith* I am a minister and the past 2 years the Lord has given me a ministry "Lazarus House" for dissolved homeless so I spend all my funds to see how it feels and what aid is available. Thank God he is in charge and I am not. I've been homeless for 2 yrs and only by the Grace of God am I where I'm @ today. Prayers I have been an addict for 12 years. Now G.O.D. has saved me. I'm living my life in a God way. Thanks to G.O.D. I'm starting a new life. God is good! I pray to God for more acceptance of true homeless people. I feel a lot of different emotions on the homeless situation and their circumstances, especially if not of their on doing or being homeless. Just for God's blessing. Thank you. May the Lord Jesus Help us All. Amen. Keep the faith. It's been a test of faith. Don't let the devil trick you and turn to God before it to late! May God be with me in all things. Believe in God. I am getting help for my self and I know God is going to make a way. Want to get things better - praying, wearing out knees. Things are getting better. Stop doing dumb stuff. First time, I don't like it but I know God will make it right. Jesus was born homeless - to free us from this scourge - It's a spiritual nemesis - Only God can cure the land - We are in big trouble finally - only one step away!!! I learn that I can't do this by myself. I need High power (God and Jesus)! Own a queen bed under the Hwy by Gateway. Jesus Loves You! I guess sometime in the near future. I will become a productive citizen again, God willing. I have never been homeless before. I don't do drugs or alcohol. I have always worked and took care of myself. God willing. This is only a temporary situation. I'm a self motivated person. This experience I have is not just a wake-up call, but a situation of life and death for me. I'm ashamed of my being and I regret the choices I made in life but I put trust in God and I will push myself until I can do it no more. I would be very happy when this nightmare is over. (pray for me) *In the words of the respondents 104 Of concern are the numerous people suffering from disabling conditions such as substance abuse and mental health issues (see Table 6.21). Addiction and mental illness are primary causes of becoming homeless. Being homeless then exacerbates the problems. For one respondent, getting drunk and high helps him to get through the situation. Addiction and mental illness are huge obstacles that get in the way of people leaving homelessness. That same respondent “cannot save money because he spends it all on drugs and alcohol.” Extensive treatment and counseling are needed for them to be physically and mentally healthy enough to have a job and/or permanent housing. The overall homeless experience is described as a learning process that is quite tough (see Table 6.22). Viewing homelessness as a learning experience implies a level of hope. Homelessness is seen as a difficult situation that people can overcome. The future possibility is that they can take what they have learned from this unpleasant experience and share it with others. Two other topics mentioned frequently by the respondents are the need for well-paying jobs and affordable housing (see Table 6.23). These two topics are often interrelated. Finding employment that pays a livable wage is necessary for obtaining and maintaining permanent housing. SUMMARY The survey respondents are overwhelmingly single, middle aged black men. Most have a high school diploma or less. They are mainly local, long term residents of this Southern metropolitan city. Additionally, this local homeless population tend to be newly (first time) and recently (six months or less) homeless. 105 Table 6.21: Homeless Experience: Alcohol / Drug Abuse and Mental Health* It's very depressing living on the streets, drugs and getting drunk help me through this but I cannot save money because I spend it all on drugs and alcohol. I used do drugs and abuse alcohol. If you didn't pick up you won't use! At the moment I can't stop doing drugs and often have suicidal thoughts. Addiction, arrested, ready to find secure employment with benefits; that I can retire and be a productive citizen & family man. There should be a intensive drug prevention outreach program. And maybe advertising drug prevention downtown on signs or maybe infomercials. So these adult men and women can stop making these stupid decisions by spending all there money on crack. That crack use is the reason me and most of the homeless people in Atlanta, GA are homeless. Please clean up the street from drug sale on Broad Street and Peachtree and Pine. Invest more money in recovery centers that allow people to work but keep control of the money with money management. I have been an addict for 12 years. Now G.O.D. has saved me. I'm living my life in a God way. Thanks to G.O.D. I'm starting a new life. People are homeless for many reasons - mental health, drug/alc. use, choice, don't want to work, lost job. The worst enemy of homeless are the people who are suppose to help. They give what they want to give, not what you need. There are only two things we need, either rehab or a job. I have been harassed by the Government for a long time now. I need to stay focus on getting housing and my mental health. Since becoming more informal about myself, heretics back around getting to know people or being in the right places at the right time, get independent affordable housing and mental health care services with out the wasting lies and wrong health care. The areas I didn't fill out, I didn't like, I feel pressure - judgments, spies, control freaks - social community gangs - dictations and controlling lives - prejudges - to species and beliefs. The more I stay away from people, happier and more self-sufficient, caring I became protective of myself and not bullied - or live in fear. People are homeless for many reasons - mental health, drug/alc. use, choice, don't want to work, lost job. I wished there was someone I could talk to about my personal issues. I really feel scared right now. I don't know what to do. Sometimes I feel like killing myself. There just needs to be more mental health programs for the homeless. Homelessness is a depressing circumstance and there is too much bureaucracy in the agencies that help. Assistance is difficult to find and having self-worth seems to be the only solution. *In the words of the respondents 106 Table 6.22: Homeless Experience: Learning and Bad Experiences* I will not forget this experience. Being homeless causes you to want to quite and some people don't understand why you are homeless unless they themselves become that way for I have learn to understand now. Life in general - make the best of a bad situation. Wake up the next day and try again. Never give up. Not really, I will say that being homeless has been an interesting experience and I can't see myself like this for too long. It has been a learning process. I've entered into a subculture so set apart from mainstream America and at times it is a very frightening place. I won't ever forget it. But hope to never encounter it again. This is an experience that I can share with people who has never been homeless. It is very learning experience being homeless. Homelessness helped me learn about myself. I do a newspaper for GA Regional Hospital and I try to do my own paper for Atlanta and I also do a poetry class. I learn that I can't do this by myself. I need High power (God and Jesus)! As an unemployed and homeless writer, I've learned quite a bit about the physical, emotional and spiritual toil of being homeless. I will, when it is over, be able to share with others about the attitude and persistence needed to overcome the seemingly depressing lifestyle that is homelessness. I have learned that people who are homeless are more likely to help than the people that have more money then they need. I wouldn't wish this on anyone. I don't like being in this situation and I hope to be getting out of it soon. It's tough out here and it's getting tougher. Don't try it! It’s a bad experience for me. I never thought I would be in this situation. If you have jobs, keep it as best you can, and make your money work for you some kind of way. No fun being homeless. No one in the U.S. should have to live on the streets if they are really trying. I would be very happy when this nightmare is over. (pray for me) It's not very pleasant at all! Never thought would be here. Here because own hard headedness / stubbornness. Not going to rely on family. Here I sit. I know with in my heart that I can do better than I am doing right now in my life. *In the words of the respondents 107 Table 6.23: Homeless Experience: Employment and Housing* Need affordable housing and low income. 1. Jobs is the main issue. 2. Good housing. There have been some trials and tribulations; a test of patience. I been in a bind with housing and a job. This program is helpful when contributing to my needs (obtaining my children, housing and employment). I want my child, job, home. Can you give or help me at all? I want help get an apartment with one bedroom and seek for year round employment and retain it. The worst enemy of homeless are the people who are suppose to help. They give what they want to give, not what you need. There are only two things we need, either rehab or a job. Being here, I find that over half of the homeless just need work. I hope I can find somewhere to stay. I need to stay focus on getting housing and my mental health. Since becoming more informal about myself, heretics back around getting to know people or being in the right places at the right time, get independent affordable housing and mental health care services with out the wasting lies and wrong health care. Resources like bus card, food and shelter would help me get on my feet. I have ran (off and on) a housing program. I have had 6 houses. The last house made me homeless. I am so mad at the homeless that I have issues due to this, but I will, soon try it once more. *In the words of the respondents 108 The main cause of homelessness is due to economic factors such as unemployment and the inability to pay rent. Besides economic reasons, health problems and family issues are other dominant causes for homelessness. Having a number of disabling conditions such as a mental illness and chronic health problems appear to be a huge factor in the issues surrounding homelessness. A majority of the respondents identify as homeless when asked directly. Interestingly, when applying identity theory, homelessness is the most salient identity, while family is the most central identity. Consistent with the low centrality of the homeless identity, my indexes of both my indexes of affective and interactive commitment to the homeless identity indicate a low attachment to the homeless identity. Most of the survey respondents demonstrate self-esteem and self-efficacy scores within the middle range and not on the low end as previous research predicted. For self-esteem, the top characteristics are having good qualities, being of worth and doing things well, while the lowest characteristics are not being satisfied with themselves, feeling useless and not being able to feel respect for themselves. Self-efficacy indicates that the respondents feel in control of themselves but probably not of their situation. Similarly to self-esteem and self-efficacy, homeless identity meanings are also within the middle range. The top adjectives for homeless identity meanings are hardworking and capable while the lowest scored adjective is that of independence. With homeless identity behaviors, finding a safe place to sleep and knowing who to trust are the top activities while knowing the meal site that serves the best tasting food is not as important. According to the respondents, almost half have not attempted to obtain permanent housing in the month prior to the survey. The main obstacle to leaving homelessness is due to lack of employment and the resulting inability to pay for permanent housing. The majority of 109 respondents indicate that they would not feel guilty if they were able to move into their own place and their homeless friends could not. Finally, almost half have no to low family support while in this difficult situation. 110 CHAPTER 7 RESULTS FOR IDENTITY THEORY, ACCEPTING THE HOMELESS IDENTITY AND EXITING HOMELESSNESS The results of the analysis performed on the twenty hypotheses are presented in the following three chapters. The hypotheses are examined using multivariate regression. This study explores two models for each hypothesis. Model one provides a baseline for the regression method based on theoretical and predictive reasons. The second model adds the control variables of race/ethnicity, gender, relationship status, military status, parent, family members, education level, age group, years in Atlanta, last month income, sleeping location and number of disabilities. For this study, the adjusted R2 is presented to provide the total amount of linear variation explained in the dependent variable by the independent variables taking into account the number of independent variables in relation to the number of cases (Knoke, Bohrnstedt, and Mee 2002). In addition, the standardized regression coefficients are shown for the study because they provide a way to compare the relative effects of different independent variables in the regression by putting the variable measurements into a common metric of standard deviation units. The standardized coefficient or beta tells the relative strength of the variables (Vogt 1999). Finally, the significant variables, those that do not occur by chance and are thus meaningful, and the standard error, standard deviation of a sampling distribution from the mean, are also provided. This chapter presents the findings of the regression analysis for the homeless identity. Specifically, homeless identity is analyzed utilizing Stryker’s identity theory (1994) and time 111 spent homeless. Additionally, aspects of people acquiring the homeless identity are addressed, along with individuals leaving the homeless situation and thus disengaging from the role. IDENTITY THEORY Commitment to the Homeless Identity Predicts Homeless Identity Salience and Centrality Identity theory argues that “the predominant direction of influence is from commitment to salience” (Stryker and Serpe 1994: 20). The other influence of identity commitment is on identity centrality. The prediction is that identity commitment will have a greater impact on identity salience than on identity centrality. Thus, hypotheses 1A and 1B state: The greater the (interactive and affective) commitment to the homeless identity, the greater the salience of the homeless identity. Hypotheses 1C and 1D read: The greater the (interactive and affective) commitment to the homeless identity, the greater the centrality of the homeless identity. Salience of the homeless identity, the tendency to invoke the homeless identity in different situations, is the dependent variable for both hypotheses 1A and 1B. Centrality reflects how important the homeless identity is to individuals and is the dependent variable for hypotheses 1C and 1D. Interactive commitment to the homeless identity is the independent variable for both hypotheses 1A and 1C. Affective commitment to the homeless identity is the independent variable for both hypotheses 1B and 1D. The interactive commitment is based on the number of other people individuals are tied to due to their homeless identity, while affective commitment relates to the strength of the attachment to others who are homeless (Stryker and Serpe 1994). For hypothesis 1A (see Table 7.1), the significant relationship between the homeless identity interactive commitment and salience of the homeless identity is found across both model one (beta = .265; p<.001) and model two (beta = .227; p<.001). This means that as the number of 112 Table 7.1: Multiple Regression of the Relationship between Interactive Commitment and Salience of the Homeless Identity (H1A) Variable Model 1 Model 2 Beta SE Beta SE Interactive commitment to homeless identity .265*** .011 .227*** .011 Black (race) .064 .092 Male (gender) .022 .227 Single (relationship) -.044 .078 Veteran .107 .158 Parent .075 .136 Living with family members -.044 .251 Education level -.016 .033 Age group .001 .007 Years in Atlanta .052 .002 Last month income -.072 .016 Unsheltered sleeping locations .084 .139 Number of disabilities .104 .062 Adjusted R2 .067 .072 SE 1.015 1.012 Dependent variable: Salience of the homeless identity *** p<.001 ** p<.01 * p<.05 113 friends who are homeless increases, the likelihood of invoking the homeless identity in different situations also increases. For model two, no other variables are predictors and have no effect on the dependent variable. There is only a slight increase in the adjusted R2 from model one to model two (.5 percent). Unlike hypothesis 1A, homeless identity salience is not explained by the independent variable, affective commitment to the homeless identity, in hypothesis 1B for either model (see Table 7.2). Among the control variables, only number of disabilities (beta = .130; p<.05) has an effect on salience of the homeless identity indicating that as the number of disabilities that people possess increases, the salience of the homeless identity increases. Neither interactive nor affective commitment to the homeless identity explain homeless identity centrality for hypotheses 1C and 1D. Only race demonstrates a significant relationship with centrality of the homeless identity in model two across both Tables 7.3 (beta = .154; p<.05) and 7.4 (beta = .166; p<.05). This means that individuals who are black are more likely to view homelessness as one of their top ranked central identity as compared to people of other races. In summary, this study finds a relationship only between homeless identity interactive commitment and salience of the homeless identity. On the other hand, there is no influence of (interactive and affective) commitment to the homeless identity towards homeless identity centrality. This indicates that the findings only partially support the prediction of commitment to the homeless identity having an effect on homeless identity salience (Hypothesis 1A) and does not support at all the commitment to the homeless identity having an effect on homeless identity centrality (Hypotheses 1C and 1D). In addition, the link between commitment and salience to the homeless identity is stronger than that between homeless identity commitment and centrality as predicted (Stryker and Serpe 1994). 114 Table 7.2: Multiple Regression of the Relationship between Affective Commitment and Salience of the Homeless Identity (H1B) Variable Model 1 Model 2 Beta SE Beta SE Affective commitment to homeless identity .096 .012 .088 .012 Black (race) .066 .094 Male (gender) .014 .234 Single (relationship) -.082 .079 Veteran .108 .160 Parent .079 .138 Living with family members -.023 .259 Education level -.023 .034 Age group .035 .007 Years in Atlanta .063 .002 Last month income -.079 .016 Unsheltered sleeping locations .103 .139 Number of disabilities .130* .063 Adjusted R2 .006 .029 SE 1.049 1.037 Dependent variable: Salience of the homeless identity *** p<.001 ** p<.01 * p<.05 115 Table 7.3: Multiple Regression of the Relationship between Interactive Commitment and Centrality of the Homeless Identity (H1C) Variable Model 1 Model 2 Beta SE Beta SE Interactive commitment to homeless identity -.002 .014 -.016 .014 Black (race) .154* .121 Male (gender) .017 .308 Singe (relationship) -.091 .099 Veteran .090 .198 Parent -.098 .174 Living with family members .033 .337 Education level .009 .043 Age group .037 .009 Years in Atlanta -.016 .003 Last month income -.021 .021 Unsheltered sleeping locations .058 .177 Number of disabilities .015 .078 Adjusted R2 -.004 .004 SE 1.253 1.248 Dependent variable: Centrality of the homeless identity *** p<.001 ** p<.01 *p<.05 116 Table 7.4: Multiple Regression of the Relationship between Affective Commitment and Centrality of the Homeless Identity (H1D) Variable Model 1 Model 2 Beta SE Beta SE Affective commitment to homeless identity -.016 .014 -.008 .015 Black (race) .166* .120 Male (gender) .035 .309 Singe (relationship) -.105 .098 Veteran .093 .193 Parent -.095 .171 Living with family members .017 .336 Education level .013 .042 Age group .038 .008 Years in Atlanta -.009 .002 Last month income -.021 .020 Unsheltered sleeping locations .060 .171 Number of disabilities .021 .077 Adjusted R2 -.004 .011 SE 1.244 1.235 Dependent variable: Centrality of the homeless identity *** p<.001 ** p<.01 *p<.05 117 Salience and Centrality of the Homeless Identity Predict Time Homeless According to Stryker and Serpe (1987), salience and centrality of the identity predict time in role, although identity salience does a much better job explaining the time factor than identity centrality. Thus, the second hypotheses (A and B) investigate homeless identity salience and centrality in relation to the length of time participating in the homeless role. Hypothesis 2A states: The greater the salience of the homeless identity, the longer people are homeless. Hypothesis 2B reads: The greater the centrality of the homeless identity, the longer people are homeless. Salience of the homeless identity is the independent variable for hypothesis 2A while the independent variable for hypothesis 2B is centrality of the homeless identity. The dependent variable for both is current length of time homeless. Tables 7.5 and 7.6 demonstrate the results of the analysis. For hypothesis 2A model one, the homeless identity salience (beta = .168; p < .01) does predict the length of time that people are homeless. Thus, the more often that individuals discuss their homelessness consistently across situations, the longer people will likely be homeless. When the control variables are added for model two, however, the salience of the homeless identity no longer has a statistically significant effect on the length of time that people are in the homeless role. In model two of Table 7.5, there are four control variables that demonstrate a correlation with current length of time homeless: living with family members (beta = -.168; p<.01), years in Atlanta (beta = .144; p<.05), unsheltered sleeping location (beta = .135; p<.05) and number of disabilities (beta = .138; p<.05). Thus, people living alone have been homeless longer than those living with family members. Additionally, the greater the length of time that people have lived in 118 Table 7.5: Multiple Regression of the Relationship between Salience of the Homeless Identity and Current Length of Time Homeless (H2A) Variable Model 1 Model 2 Beta SE Beta SE Salience of homeless identity .168** 1.099 .088 1.080 Black (race) -.042 1.643 Male (gender) -.098 4.174 Single (relationship) -.066 1.375 Veteran .112 2.843 Parent -.099 2.438 Living with family members -.168** 4.752 Educational level -.028 .578 Age group .117 .117 Years in Atlanta .144* .034 Last month income -.095 .285 Unsheltered sleeping locations .135* 2.412 Number of disabilities .138* 1.105 Adjusted R2 .025 .128 SE 18.897 17.863 Dependent variable: Current length of time homeless *** p<.001 ** p<.01 * p<.05 119 Table 7.6: Multiple Regression of the Relationship between Centrality of the Homeless Identity and Current Length of Time Homeless (H2B) Variable Model 1 Model 2 Beta SE Beta SE Centrality of homeless identity .047 .994 .014 .961 Black (race) -.036 1.810 Male (gender) -.097 4.725 Single (relationship) -.056 1.460 Veteran .102 2.972 Parent -.095 2.613 Living with family members -.165* 5.304 Education level -.055 .618 Age group .151* .126 Years in Atlanta .115 .037 Last month income -.113 .304 Unsheltered sleeping locations .190** 2.549 Number of disabilities .148* 1.152 Adjusted R2 -.002 .118 SE 19.558 18.348 Dependent variable: Current length of time homeless *** p<.001 ** p<.01 * p<.05 120 this large Southern city, the longer they have been homeless. Individuals sleeping in unsheltered locations have been homeless for a greater length of time than those staying in sheltered places. Finally, the more disabilities people posses, the longer they have been homeless. According to the beta, the number of disabilities contributes more to explaining length of time homeless than any other variable in model two. With hypothesis 2B (see Table 7.6), homeless identity centrality does not show a relationship with the current length of time that people are homeless in either model. However, model two does demonstrate four statistically significant relationships with the dependent variable: living with family members (beta = -.165; p<.05), age (beta = .151; p<.01), unsheltered sleeping locations (beta = .190; p<.05) and number of disabilities (beta = .148; p<.05). As with hypothesis 2A, people living alone have been homeless longer than those living with family members, individuals sleeping on the streets have been homeless longer than those staying in sheltered locations and the more disabilities people possess, the longer they will be homeless. Additionally for Table 7.6, people who are older have been homeless longer than younger individuals. Unsheltered sleeping location, according to the beta, has the strongest relationship with length of time homeless than any other variable in model two. In conclusion, this study does support hypothesis 2A and previous research that salience to the homeless identity predicts time in role (Stryker and Serpe 1994). However, that relationship is no longer significant and is reduced in strength when the control variables are added. The findings do not indicate that homeless identity centrality predicts time in role at all and thus does not support hypothesis 2B. As with hypotheses 1 (A-D), salience to the homeless identity does a much better job of predicting the length of time in the homeless role than homeless identity centrality for hypotheses 2 (A-B). 121 ACCEPTING THE HOMELESS IDENTITY Relationship between Time Homeless and Acceptance of the Homeless Identity Identity construction varies with the time that homeless individuals spend on the streets. Homeless persons on the streets for a short period of time distance themselves from the homeless identity, whereas those on the streets longer than two years embrace the homeless identity (Snow and Anderson 1987). Hypotheses 3 (A-B) tests the assumption that the longer and more often individuals are homeless, the more likely people are to identify as homeless. The dependent variable is the acceptance of a homeless identity. The total length of time homeless over the past three years (H3A) and the number of times homeless during the last three years (H3B) prior to the survey are the measurements for length of time homeless and thus the independent variables. The results of the regression analysis are shown in Tables 7.7 and 7.8. Hypothesis 3A (6.2 percent) explains about the same amount of linear variance for accepting the homeless identity as hypothesis 3B (7 percent) does after including the additional twelve demographic and characteristic independent variables in model two. The significant relationship for length of time homeless over the past three years and accepting the homeless identity holds true for hypothesis 3A model one (beta = .218; p<.01) and model two (beta = .168; p<.05). Similarly, number of different times homeless over the past three years shows an effect on accepting the homeless identity for hypothesis 3B models one (beta = .173; p<.01) and two (beta = .125; p<.05). However, the strength of that relationship is reduced for both measures of time homeless after adding the control variables. Thus, the increasing time that people are homeless is a good predictor of their increasing acceptance of the homeless identity. The findings of this hypothesis do confirm prior research (Snow and Anderson 1987). 122 Table 7.7: Multiple Regression of the Relationship between Total Length of Time Homeless and Acceptance of the Homeless Identity (H3A) Variable Model 1 Model 2 Beta SE Beta SE Total length of homelessness (past 3 years) .218*** .005 .168* .005 Black (race) -.075 .095 Male (gender) .010 .248 Single (relationship) .014 .083 Veteran .061 .166 Parent .028 .141 Living with family members -.013 .282 Education level .073 .034 Age group .021 .007 Years in Atlanta -.092 .002 Last month income -.029 .017 Unsheltered sleeping locations .062 .143 Number of disabilities .196** .065 Adjusted R2 .044 .062 SE 1.045 1.034 Dependent variable: Acceptance of homeless identity *** p<.001 ** p<.01 * p<.05 123 Table 7.8: Multiple Regression of the Relationship between Number of Different Times Homeless and Acceptance of the Homeless Identity (H3B) Variable Model 1 Model 2 Beta SE Beta SE Different times homeless (past 3 years) .173* .047 .125* .049 Black (race) -.093 .098 Male (gender) .016 .246 Single (relationship) .013 .081 Veteran .063 .162 Parent -.012 .140 Living with family members -.075 .270 Education level .042 .034 Age group .043 .007 Years in Atlanta -.070 .002 Last month income -.058 .017 Unsheltered sleeping locations .126* .138 Number of disabilities .206*** .066 Adjusted R2 .026 .070 SE 1.062 1.038 Dependent variable: Acceptance of homeless identity *** p<.001 ** p<.01 * p<.05 124 Additionally, the number of disabilities affects acceptance of the homeless identity for both Hypotheses 3A (beta = .196; p<.01) and 3B (beta = .206; p<.001) in model two: as the number of disabilities that people experience increases, the acceptance of the homeless identity also increases. By comparing the standardized coefficients in model two, number of disabilities is seen contributing more to explaining acceptance of the homeless identity than any other variable for both hypotheses. Unsheltered sleeping location (beta = .126; p<.05) also shows a significant relationship with accepting the homeless identity in Table 7.8. Thus, individuals sleeping outside are more likely to accept the homeless identity than people staying in sheltered locations. EXITING HOMELESSNESS Relationship between Acceptance of the Homeless Identity and Attempts to Exit the Homeless Situation A negative result of people who identify as homeless is that they are significantly less likely to make attempts to transition from the homeless situation than those who do not identify as homeless (Farrington and Robinson 1999; Osborne 2002). This prediction is reflected in hypothesis four. Attempts made to exit homelessness such as filling out an apartment application is the dependent variable, while accepting a homeless identity and the strength of that identity is the independent variable. The findings of the multiple regressions are presented in Table 7.9. None of the linear variance for exiting the homeless situation is explained with the homeless identity in model one. Model two shows a slight increase in explanation after including the control variables (3.3 percent). For hypothesis four, accepting a homeless identity does not predict people’s attempts to exit the homeless situation in either model. This does not support hypothesis four and the 125 Table 7.9: Multiple Regression of the Relationship between Acceptance of the Homeless Identity and Attempts to Exit the Homeless Situation (H4) Variable Model 1 Model 2 Beta SE Beta SE Acceptance of homeless identity -.039 .102 .006 .107 Black (race) .044 .177 Male (gender) -.067 .425 Single (relationship) .098 .138 Veteran -.047 .272 Parent .087 .235 Living with family members .086 .481 Education level -.050 .057 Age group -.038 .011 Years in Atlanta .071 .003 Last month income .173** .030 Unsheltered sleeping locations .036 .243 Number of disabilities -.078 .112 Adjusted R2 -.003 .033 SE 1.714 1.683 Dependent variable: Attempts to exit homelessness *** p<.001 ** p<.01 * p<.05 126 previous research that a significant relationship exists between homeless identity and attempts to exit the homeless situation (Osborne 2002). The only variable to have an effect is last month’s income (beta = .173; p<.01) in model two. Thus, the more money that people have earned in the month prior to the survey, the more attempts made to no longer be homeless. This finding reflects the majority of respondents indicating that economics is their main reason for becoming homeless and the biggest obstacle for them to obtain permanent housing. Identity Commitment, Salience and Centrality to the Homeless Identity Predict Attempts to Exit the Homeless Situation As with hypothesis four, hypotheses 5 (A-D) also investigate the impact of the homeless identity on people’s ability to leave the homeless situation. For these hypotheses, the independent variable of homeless identity is measured by using identity theory: identity (interactive and affective) commitment, salience and centrality. Attempts made to exit homelessness is the dependent variable. Thus, hypothesis 5A reads: the greater the interactive commitment to the homeless identity, the less likely people are to attempt to exit the homeless situation. Hypothesis 5B states: the greater the affective commitment to the homeless identity, the less likely people are to attempt to exit the homeless situation. Hypothesis 5C states: the greater the salience of the homeless identity, the less likely people are to attempt to exit the homeless situation. Hypothesis 5D states: the greater the centrality of the homeless identity, the less likely people are to attempt to exit the homeless situation. The results of the analysis are shown in Tables 7.10 to 7.13. Of the four hypotheses, only hypothesis 5B demonstrates a correlation between attempts made to exit the homeless situation and the independent variable, affective commitment to the homeless identity. The significant 127 Table 7.10: Multiple Regression of the Relationship between Interactive Commitment to the Homeless Identity and Attempts to Exit the Homeless Situation (H5A) Variable Model 1 Model 2 Beta SE Beta SE Interactive commitment to homeless identity .072 .019 .081 .019 Black (race) .045 .175 Male (gender) -.052 .414 Single (relationship) .123 .136 Veteran -.067 .267 Parent .079 .234 Living with family members .082 .478 Education level -.029 .057 Age group -.045 .011 Years in Atlanta .063 .003 Last month income .196** .029 Unsheltered sleeping locations .016 .246 Number of disabilities -.076 .108 Adjusted R2 .001 .048 SE 1.720 1.680 Dependent variable: Attempts to exit homelessness *** p<.001 ** p<.01 * p<.05 128 Table 7.11: Multiple Regression of the Relationship between Affective Commitment to the Homeless Identity and Attempts to Exit the Homeless Situation (H5B) Variable Model 1 Model 2 Beta SE Beta SE Affective commitment to homeless identity .170** .020 .147* .020 Black (race) .040 .174 Male (gender) -.053 .409 Single (relationship) .110 .134 Veteran -.056 .264 Parent .085 .231 Living with family members .081 .473 Education level -.024 .056 Age group -.035 .011 Years in Atlanta .074 .003 Last month income .189** .029 Unsheltered sleeping locations .013 .241 Number of disabilities -.059 .107 Adjusted R2 .025 .065 SE 1.697 1.662 Dependent variable: Attempts to exit homelessness *** p<.001 ** p<.01 * p<.05 129 Table 7.12: Multiple Regression of the Relationship between Salience of the Homeless Identity and Attempts to Exit the Homeless Situation (H5C) Variable Model 1 Model 2 Beta SE Beta SE Salience of homeless identity .064 .106 .098 .109 Black (race) .041 .176 Male (gender) -.051 .414 Single (relationship) .128 .136 Veteran -.063 .272 Parent .076 .236 Living with family members .098 .481 Education level -.042 .057 Age group -.044 .011 Years in Atlanta .064 .003 Last month income .198** .029 Unsheltered sleeping locations .008 .246 Number of disabilities -.080 .109 Adjusted R2 .000 .051 SE 1.725 1.680 Dependent variable: Attempts to exit homelessness *** p<.001 ** p<.01 * p<.05 130 Table 7.13: Multiple Regression of the Relationship between Centrality of the Homeless Identity and Attempts to Exit the Homeless Situation (H5D) Variable Model 1 Model 2 Beta SE Beta SE Centrality of homeless identity .045 .094 .092 .093 Black (race) .042 .188 Male (gender) -.059 .456 Single (relationship) .113 .139 Veteran -.081 .272 Parent .129 .242 Living with family members .108 .522 Education level -.045 .058 Age group -.098 .012 Years in Atlanta .094 .004 Last month income .214*** .030 Unsheltered sleeping locations .030 .248 Number of disabilities -.064 .109 Adjusted R2 -.002 .075 SE 1.729 1.661 Dependent variable: Attempts to exit homelessness *** p<.001 ** p<.01 * p<.05 131 relationship is across both model one (beta = .170; p<.01) and model two (beta = .147; p<.05); however, it is in the opposite direction as predicted. In other words, the stronger the attachment to other homeless people, the more likely the respondents will be to make attempts to leave the homeless situation. Of the control variables in model two, only income shows an effect on the attempts made to exit the homeless situation. This significant relationship exists for Table 7.10 (beta = .196; p<.01), 7.11 (beta = .189; p<.01), 7.12 (beta = .198; p<.01) and 7.13 (beta = .214; p<.001). Thus, the higher the amount of income that people have made in the month prior to the survey, the more attempts made to exit the homeless situation. In summary, hypotheses 5 (A-D) also does not confirm the prediction that identifying as homeless will significantly reduce the number of attempts made to exit the situation. In fact, these results found the opposite with the number of attempts to exit homelessness increasing with the strengthening of ties to friends who are also homeless (affective commitment to the homeless identity). MacKnee and Mervyn (2002) believed that feeling a sense of loyalty towards friends who are homeless would be an obstacle to exiting the situation. However, this study finds that a positive result of strong friendships with other homeless people is finding housing. Perhaps homeless friends encourage each other and thus provide a support network that they might not otherwise have. Relationship between Family Support and Attempts to Exit the Homeless Situation Hypothesis six examines the idea that without support from housed family members, relatives who are homeless will make fewer attempts to find permanent housing (MacKnee and Mervyn 2002). Family support is the independent variable and attempts made to exit 132 homelessness is the dependent variable. Table 7.14 displays the findings from the multiple regression analysis. From model one to model two there is a slight increase (4 percent) in explaining the linear variance according to the adjusted R2. Family support (beta = .127; p<.05) does correlate with attempts to exit homelessness in model one. The relationship is in a positive direction. Therefore, the more support from housed family members that the respondents have, the more likely they are to make attempts to exit the homeless situation. However, when the control variables are added, family support is no longer significant and is reduced in strength according to the beta. The only control variable to demonstrate a correlation with attempts to exit homelessness is income (beta = .197; p<.01). Thus, the higher the amount of income that people have made in the month prior to the survey, the more attempts made to exit homelessness, no matter if they have family support or not. In conclusion, across all the hypotheses that investigate people making attempts to find permanent housing, only two independent variables show an effect on the dependent variable – affective commitment to the homeless identity and family support. Even though family support demonstrates a relationship with exiting homelessness, it is no longer correlated in model two and its strength is reduced. On the other hand, last month’s income prior to the survey has a significant relationship with attempts to exit homelessness in model two for all six tables. This indicates that income has the strongest relationship with respondents no longer being homeless, and thus implies that a livable wage is needed to obtain and maintain a permanent place to live. 133 Table 7.14: Multiple Regression of the Relationship between Family Support and Attempts to Exit the Homeless Situation (H6) Variable Model 1 Model 2 Beta SE Beta SE Family support .127* .029 .093 .030 Black (race) .040 .177 Male (gender) -.062 .412 Single (relationship) .112 .135 Veteran -.072 .268 Parent .076 .235 Living with family members .080 .477 Education level -.042 .056 Age group -.031 .011 Years in Atlanta .050 .003 Last month income .197** .029 Unsheltered sleeping locations .040 .242 Number of disabilities -.049 .109 Adjusted R2 .012 .052 SE 1.703 1.669 Dependent variable: Attempts to exit homelessness *** p<.001 ** p<.01 * p<.05 134 SUMMARY In regards to identity theory (Stryker and Serpe 1994), there is only partial support for hypotheses 1 (A-D). Interactive commitment to the homeless identity shows a correlation with homeless identity salience across both models (H1A). However, interactive commitment does not predict homeless identity centrality (H1C) and affective commitment to the homeless identity has no effect on either homeless identity salience (H1B) or centrality (H1D). Similarly to the previous hypotheses, hypotheses 2 (A-B) also only have partial support. Homeless identity salience predicts time in role (Stryker and Serpe 1994) for model one but not in model two after the control variables are added (H2A) while centrality of the homeless identity does not have an effect on time homeless at all. The results of this study confirm prior research (Snow and Anderson 1987) that homeless persons on the streets for a short period of time will distance themselves from the homeless identity, while those on the street for a longer stretch will embrace the homeless identity (H3A-B). When it comes to no longer being homeless, accepting a homeless identity does not predict people’s attempts to exit the homeless situation (H4). For hypothesis 5A-D, only affective commitment to the homeless identity demonstrates a significant relationship with leaving homelessness (H5B); however, it is in the opposite direction as predicted. As for increasing family support, it does correlate with increasing number of attempts made to exit homelessness in model one but not in model two (H6). 135 CHAPTER 8 RESULTS FOR SELF-ESTEEM AND SELF-EFFICACY The purpose of this chapter is to discuss the results of the analysis for self-esteem and self-efficacy. Model one is analyzed on dimensions of time homeless, acceptance of the homeless identity, identity theory (commitment, salience and centrality) and family support across the fifteen regression tables. The second model in the tables includes the independent variable, along with twelve demographic and homeless characteristics. SELF-ESTEEM Relationship between Time Homeless and Self-Esteem Previous theoretical research suggests that people who are labeled as stigmatized such as homeless persons would possess low self-esteem. Empirical research, however, has not supported those findings. Instead, Croker and Major (1989) indicate that several strategies would be used by homeless individuals to protect their self-concepts. The strategies include attributing failure to the group and not the person, making in group comparisons instead of out group (Tajfel and Turner 1986) and valuing behavior that they or the group perform well versus activities that they or the group do badly (Rosenberg 1979). These strategies are not universally effective for all people who are homeless. Instead, factors such as length of time in the homeless role indicate the prevalence of use for each strategy (Croker and Major 1989). Therefore, hypotheses 7 (A-B) propose that the longer and more often people are homeless, the lower their self-esteem. The dependent variable is self- 136 esteem and the independent variables are total length of time homeless and number of different times homeless over the past three years. As shown in Table 8.1, the explanation of linear variance for self-esteem increases dramatically from model one (3.2 percent) to model two (15.2 percent) after adding the control variables. In model one hypothesis 7A, total length of time homeless over the past three years (beta = -.189; p<.01) demonstrates a significant relationship with self-esteem. The relationship is in a negative direction, meaning that the longer people are in the homeless role, the lower their self-esteem. However, that correlation is not maintained in model two once the control variables are added. In model two, the only two predictors of self-esteem are education level (beta = .184; p<.01) and number of disabilities (beta = -.291; p<.001). Thus, the lower the education level of people who are homeless, the lower the self-esteem. For this hypothesis, number of disabilities has a negative effect on self-esteem. This means that the more disabilities a homeless person experiences, the lower the self-esteem. According to the beta, number of disabilities is the strongest relationship with self-esteem of any of the variables. Similarly to the previous table, the linear variance in Table 8.2 (H7B) shows a dramatic increase from model one (1.2 percent) to model two (14.5 percent). Additionally, the independent variable, different times homeless over the past three years, only has a statistically significant effect on self-esteem in model one (beta = -.126; p<.05). In model two, living with family members (beta = .178; p<.01) and number of disabilities (beta = -.298; p<.001) are still predictors of self-esteem. For this table, unsheltered sleeping location (beta = -.125; p<.05) also has a correlation with self-esteem. With the relationship in a negative direction, this indicates 137 Table 8.1: Multiple Regression of the Relationship between Total Length of Time Homeless and Self-Esteem (H7A) Variable Model 1 Model 2 Beta SE Beta SE Total length of homelessness (past 3 years) -.189** .023 -.118 .024 Black (race) .094 .426 Male (gender) -.062 1.070 Single (relationship) -.017 .368 Veteran -.016 .738 Parent .025 .630 Living with family members -.027 1.261 Education level .184** .151 Age group -.009 .031 Years in Atlanta .086 .009 Last month income .071 .076 Unsheltered sleeping locations -.103 .640 Number of disabilities -.291*** .290 Adjusted R2 .032 .152 SE 4.973 4.655 Dependent variable: Self-Esteem *** p<.001 ** p<.01 * p<.05 138 Table 8.2: Multiple Regression of the Relationship between Number of Different Times Homeless and Self-Esteem (H7B) Variable Model 1 Model 2 Beta SE Beta SE Different times homeless (past 3 years) -.126* .222 -.071 .218 Black (race) .093 .444 Male (gender) -.046 1.072 Single (relationship) -.016 .361 Veteran -.027 .728 Parent .033 .630 Living with family members -.039 1.215 Education level .178** .151 Age group -.031 .030 Years in Atlanta .056 .009 Last month income .086 .074 Unsheltered sleeping locations -.125* .623 Number of disabilities -.298*** .293 Adjusted R2 .012 .145 SE 5.067 4.715 Dependent variable: Self-Esteem *** p<.001 ** p<.01 * p<.05 139 that people sleeping on the streets have a lower self-esteem than persons staying in sheltered locations. In conclusion, these findings confirm the hypotheses only prior to the addition of the control variables, that the length of time and number of times homeless are factors in producing strategies to protect the self-concept. It does appear that there are other factors that have a stronger relationship with self-esteem than length of time and number of times in the homeless role. In other words, if homeless persons have multiple disabling conditions, the disabilities will contribute to a low self-esteem whether or not they are recently homeless or have been homeless for many years. The same is true of education level. If homeless people do not have a high school diplomas, that will contribute to a low self-esteem no matter if they have been homeless a month or are frequently homeless. Relationship between Acceptance of the Homeless Identity and Self-Esteem Hypothesis eight investigates if homeless identification could be a possible factor in producing strategies that protect the self-worth of people while homeless. Thus, hypothesis eight reads: people who strongly identify as homeless are more likely to possess a lower self-esteem than those who do not identify as homeless. Accepting the homeless identification is the independent variable while self-esteem is the dependent variable. The results in Table 8.3 do confirm hypothesis eight for model one (beta = -.196; p<.001). Therefore, a possible strategy for people who are homeless is to not accept the stigmatized identification at all to safeguard their self-worth during this troubling situation. However, once the control variables are added, accepting the homeless identity is no longer significant. Of the control variables, education level (beta = .175; p<.01) and number of disabilities (beta = -.291; 140 Table 8.3: Multiple Regression of the Relationship between Acceptance of the Homeless Identity and Self-Esteem (H8) Variable Model 1 Model 2 Beta SE Beta SE Acceptance of homeless identity -.196*** .281 -.112 .277 Black (race) .070 .430 Male (gender) -.046 1.096 Single (relationship) .000 .362 Veteran -.033 .728 Parent .021 .623 Living with family members -.075 1.176 Education level .175** .152 Age group -.010 .030 Years in Atlanta .049 .009 Last month income .085 .075 Unsheltered sleeping locations -.111 .626 Number of disabilities -.291*** .294 Adjusted R2 .035 .155 SE 5.037 4.714 Dependent variable: Self-Esteem *** p<.001 ** p<.01 * p<.05 141 p<.001) show an effect on self-esteem in model two. Therefore, lower education levels and possessing multiple disabilities also predict a low self-esteem. Number of disabilities has the strongest relationship with self-esteem of any of the variables according to the beta. Relationship between (Interactive and Affective) Commitment, Salience and Centrality of the Homeless Identity and Self-Esteem As with the previous hypothesis, hypotheses 9 (A-D) examine commitment, salience and centrality of the homeless identity as possible factors in provoking strategies to protect the self-esteem while homeless. The implication is that the greater the commitment, salience and centrality of the homeless identity, the lower the self-esteem. The dependent variable is self-esteem, while (interactive and affective) commitment, salience and centrality of the homeless identity are the independent variables. The findings of the analysis for hypotheses nine (A-D) are presented in Tables 8.4 to 8.7. Of the four hypotheses, only centrality of the homeless identity has a correlation with self-esteem for both model one (beta = -.125; p<.05) and two (beta = -.117; p<.05). The relationship is in a negative direction meaning that the higher the centrality of the homeless identity, the lower the self-esteem. These findings show limited support for hypotheses 9 (A-D). Thus, a possible strategy for persons who are homeless is to not have the stigmatized identity as their most important identity but to have a positive identity such as family membership as their central identity. For all four hypotheses in model two, three control variables have a statistically significant effect on self-esteem: education level, unsheltered sleeping locations and number of disabilities (see Tables 8.4-8.7). Thus, the lower the education level and the more disabilities homeless people experience, the lower the self-esteem. Additionally, people sleeping outdoors have a 142 Table 8.4: Multiple Regression of the Relationship between Interactive Commitment to the Homeless Identity and Self-Esteem (H9A) Variable Model 1 Model 2 Beta SE Beta SE Interactive commitment to homeless identity -.027 .053 .057 .052 Black (race) .083 .429 Male (gender) -.030 1.057 Single (relationship) .003 .361 Veteran -.036 .726 Parent .019 .628 Living with family members -.066 1.170 Education level .169** .154 Age group -.021 .030 Years in Atlanta .043 .009 Last month income .077 .075 Unsheltered sleeping locations -.136* .641 Number of disabilities -.323*** .288 Adjusted R2 -.003 .141 SE 5.106 4.725 Dependent variable: Self-Esteem *** p<.001 ** p<.01 * p<.05 143 Table 8.5: Multiple Regression of the Relationship between Affective Commitment to the Homeless Identity and Self-Esteem (H9B) Variable Model 1 Model 2 Beta SE Beta SE Affective commitment to homeless identity .055 .057 .097 .054 Black (race) .094 .425 Male (gender) -.019 1.056 Single (relationship) -.026 .356 Veteran -.030 .712 Parent .025 .616 Living with family members -.075 1.166 Education level .181** .151 Age group -.017 .030 Years in Atlanta .053 .009 Last month income .078 .073 Unsheltered sleeping locations -.127* .619 Number of disabilities -.312*** .281 Adjusted R2 .000 .149 SE 5.077 4.681 Dependent variable: Self-Esteem *** p<.001 ** p<.01 * p<.05 144 Table 8.6: Multiple Regression of the Relationship between Salience of the Homeless Identity and Self-Esteem (H9C) Variable Model 1 Model 2 Beta SE Beta SE Salience of homeless identity -.077 .295 -.008 .283 Black (race) .082 .432 Male (gender) -.029 1.062 Single (relationship) -.009 .361 Veteran -.032 .735 Parent .030 .634 Living with family members -.065 1.174 Education level .166** .153 Age group -.023 .030 Years in Atlanta .056 .009 Last month income .075 .076 Unsheltered sleeping locations -.121* .633 Number of disabilities -.322*** .290 Adjusted R2 .002 .143 SE 5.123 4.748 Dependent variable: Self-Esteem *** p<.001 ** p<.01 * p<.05 145 Table 8.7: Multiple Regression of the Relationship between Centrality of the Homeless Identity and Self-Esteem (H9D) Variable Model 1 Model 2 Beta SE Beta SE Centrality of homeless identity -.125* .258 -.117* .245 Black (race) .057 .458 Male (gender) -.080 1.158 Single (relationship) -.031 .371 Veteran -.058 .738 Parent .031 .654 Living with family members -.029 1.264 Education level .184** .157 Age group -.014 .032 Years in Atlanta .051 .009 Last month income .077 .077 Unsheltered sleeping locations -.132* .647 Number of disabilities -.329*** .292 Adjusted R2 .012 .160 SE 5.095 4.696 Dependent variable: Self-Esteem *** p<.001 ** p<.01 * p<.05 146 lower self-esteem than individuals staying in sheltered locations. Comparing the standardized coefficients, number of disabilities has the strongest relationship with self-esteem than any other variable in model two for all the hypotheses. Relationship between Family Support and Self-Esteem With self-esteem, the opinion of significant others such as parents can be strongly valued (Rosenberg 1979). Thus, hypothesis ten tests that the less support from family members, the lower the self-esteem of the relatives who are homeless. In this case, family support is the independent variable and self-esteem is the dependent variable. Table 8.8 presents the results of the regression analysis. There is a dramatic increase in the explanation of the linear variance from model one (2.3 percent) to model two (15.1 percent). Family support (beta = .162; p<.01) does demonstrate a significant relationship in model one to self-esteem. The relationship is in a positive direction. Therefore, the greater the family support, the higher the self-esteem of homeless relatives. However, when the control variables are added in model two, family support no longer indicates a correlation with self-esteem. This means that there is limited support for hypothesis ten. The control variables in model two that are good predictors of self-esteem are education level (beta = .149; p<.05) and number of disabilities (beta = -.311; .001). According to the beta in model two, number of disabilities has the strongest relationship with self-esteem, similarly to the previous tables. In summary, family support does have an effect on self-esteem until other variables such as education level and number of disabilities are added that have a stronger relationship. Thus, homeless persons with less than a high school diploma will be more likely to have a lower self-esteem than those with some graduate school, no matter if they have family support or not. 147 Table 8.8: Multiple Regression of the Relationship between Family Support and Self-Esteem (H10) Variable Model 1 Model 2 Beta SE Beta SE Family support .162** .083 .116 .082 Black (race) .117 .458 Male (gender) -.037 1.075 Single (relationship) .006 .366 Veteran -.020 .749 Parent .027 .643 Living with family members -.082 1.211 Education level .149* .157 Age group -.017 .031 Years in Atlanta .045 .009 Last month income .076 .077 Unsheltered sleeping locations -.096 .654 Number of disabilities -.311*** .300 Adjusted R2 .023 .151 SE 5.109 4.762 Dependent variable: Self-Esteem *** p<.001 ** p<.01 * p<.05 148 Additionally, if homeless individuals experience multiple disabilities, they will be more likely to possess a lower self-esteem than those without disabling conditions, whether or not they have support from their housed family members. SELF-EFFICACY Relationship between Family Support and Self-efficacy According to Gecas and Schwalbe (1983), the opinions of others will have a greater determination of self-efficacy for homeless people than social comparisons due to their lack of resources and thus limited ability to compare. Thus, hypothesis eleven states: the greater the support from homeless people’s families, the higher the self-efficacy. Family support is the independent variable and self-efficacy is the dependent variable. Table 8.9 displays the results from the multivariate regression. For model two, 14.2 percent of the linear variance for self-efficacy is explained with family support and the characteristic variables. That is an increase of over seven percent after including the control variables. The correlation is statistically significant between self-efficacy and family support in models one (beta = .258; p<.001) and two (beta = .212; p<.001). Therefore, the stronger the support from housed family members, the higher the self-efficacy of homeless relatives. This finding is consistent with the prediction for hypothesis eleven. Besides family support demonstrating a relationship with self-efficacy, number of disabilities (beta = -.261; p<.001) also shows an effect. The relationship is in a negative direction so that the more disabling conditions homeless people possess, the lower the self-efficacy. A comparison of standardized coefficients shows that the number of disabilities has a stronger relationship with self-efficacy than any other variable in the model two. 149 Table 8.9: Multiple Regression for Relationship between Family Support and Self-Efficacy (H11) Variable Model 1 Model 2 Beta SE Beta SE Family support .258*** .048 .212*** .049 Black (race) .102 .274 Male (gender) .040 .643 Single (relationship) .016 .219 Veteran -.038 .447 Parent -.006 .385 Living with family members .038 .724 Education level .015 .094 Age group -.003 .018 Years in Atlanta -.049 .006 Last month income .074 .046 Unsheltered sleeping locations -.116 .391 Number of disabilities -.261*** .179 Adjusted R2 .063 .142 SE 2.972 2.845 Dependent variable: Self-Efficacy *** p<.001 ** p<.01 * p<.05 150 Relationship between Acceptance of the Homeless Identity and Self-Efficacy The next objective is to explore the association between homeless identity and self-efficacy. People who strongly identify as homeless are more likely to possess a lower self-efficacy than those who do not identify as homeless. For hypothesis twelve, the dependent variable is self-efficacy while accepting a homeless identity is the independent variable. The findings displayed in Table 8.10 are consistent with hypothesis twelve. Acceptance of the homeless identity and self-efficacy are correlated in both models one (beta = -.262; p<.001) and two (beta = -.177; p<.01). The relationship is in a negative direction meaning that the stronger the identification as homeless, the lower the self-efficacy. Therefore, as with hypothesis eight, a strategy for protecting the self-concept is not to accept the stigmatizing identity. Regarding the control variables in model two, unsheltered sleeping locations (beta = -.126; p<.05) and number of disabilities (beta = -.243; p<.001) also demonstrate an effect on self-efficacy. Thus, persons sleeping outdoors have a lower self-efficacy than people staying in sheltered locations and individuals possessing multiple disabilities predict a lower self-efficacy than those without disabling conditions. According to the beta, number of disabilities has a stronger relationship with self-efficacy than any other variable in model two. Relationship between (Interactive and Affective) Commitment, Salience and Centrality of the Homeless Identity and Self-Efficacy Hypotheses 13 (A-D), similarly to the previous hypothesis, assess the impact of homeless identity on self-efficacy; however, this hypothesis uses (interactive and affective) commitment, salience and centrality of the homeless identity as the independent variables. The dependent variable is self-efficacy. Hypothesis 13A states: the greater the interactive commitment to the 151 Table 8.10: Multiple Regression of the Relationship between Acceptance of the Homeless Identity and Self-Efficacy (H12) Variable Model 1 Model 2 Beta SE Beta SE Acceptance of homeless identity -.262*** .164 -.177** .167 Black (race) .041 .259 Male (gender) .038 .660 Single (relationship) .013 .218 Veteran -.035 .439 Parent -.008 .376 Living with family members .054 .708 Education level .050 .092 Age group .000 .018 Years in Atlanta -.039 .005 Last month income .068 .045 Unsheltered sleeping locations -.126* .377 Number of disabilities -.243*** .177 Adjusted R2 .065 .125 SE 2.937 2.840 Dependent variable: Self-Efficacy *** p<.001 ** p<.01 * p<.05 152 homeless identity, the lower the self-efficacy. Hypothesis 13B reads: the greater the affective commitment to the homeless identity, the lower the self-efficacy. Hypothesis 13C states: the greater the salience of the homeless identity, the lower the self-efficacy. Hypothesis 13D reads: the greater the centrality of the homeless identity, the lower the self-efficacy. Tables 8.11 to 8.14 display the results of the analysis. Of the four hypotheses, only salience and centrality of the homeless identity have a significant relationship with self-efficacy. Salience of the homeless identity is correlated with self-efficacy in model one only (beta = -.152; p<.05). On the other hand, centrality of the homeless identity has an effect on self-efficacy in model one (beta = -.150; p<.05) and model two (beta = -.137; p<.05). Additionally, there is a correlation between unsheltered sleeping locations and number of disabilities with self-efficacy across all four hypotheses (see Tables 8.11-8.14). The strongest predictor of self-efficacy in model two across all tables is number of disabilities. To summarize, a strategy for people while homeless to protect their self-worth is to not have the stigmatized identity be the most central. Additionally, not invoking the homeless identity in as many different situations could help protect the self-concept. An important step to having a higher self-efficacy is to find a sheltered place to sleep such as transitional housing. Finally, people with multiple disabilities will struggle with a lower self-efficacy and so need to work on becoming physically and emotionally healthy. Relationship between Time Homeless and Self-Efficacy For hypotheses 14 (A-B), I explore whether there is a relationship between times homeless and self-efficacy. Total length of time homeless and number of different times homeless over the past three years are the independent variables while self-efficacy is the dependent variable. 153 Table 8.11: Multiple Regression for Relationship between Interactive Commitment to the Homeless Identity and Self-Efficacy (H13A) Variable Model 1 Model 2 Beta SE Beta SE Interactive commitment to homeless identity -.098 .031 -.034 .032 Black (race) .057 .263 Male (gender) .050 .648 Single (relationship) .017 .221 Veteran -.047 .444 Parent -.008 .385 Living with family members .071 .717 Education level .035 .094 Age group .001 .019 Years in Atlanta -.022 .006 Last month income .074 .046 Unsheltered sleeping locations -.139* .393 Number of disabilities -.270*** .176 Adjusted R2 .006 .089 SE 3.024 2.895 Dependent variable: Self-Efficacy *** p<.001 ** p<.01 * p<.05 154 Table 8.12: Multiple Regression for Relationship between Affective Commitment to the Homeless Identity and Self-Efficacy (H13B) Variable Model 1 Model 2 Beta SE Beta SE Affective commitment to homeless identity -.011 .034 .012 .033 Black (race) .062 .261 Male (gender) .058 .647 Single (relationship) .007 .218 Veteran -.049 .436 Parent -.008 .378 Living with family members .059 .714 Education level .043 .092 Age group -.007 .018 Years in Atlanta -.022 .005 Last month income .076 .045 Unsheltered sleeping locations -.152 .379 Number of disabilities -.275*** .172 Adjusted R2 -.003 .092 SE 3.017 2.869 Dependent variable: Self-Efficacy *** p<.001 ** p<.01 * p<.05 155 Table 8.13: Multiple Regression of the Relationship between Salience of the Homeless Identity and Self-Efficacy (H13C) Variable Model 1 Model 2 Beta SE Beta SE Salience of homeless identity -.152* .172 -.078 .170 Black (race) .059 .259 Male (gender) .053 .637 Single (relationship) .010 .217 Veteran -.035 .441 Parent -.004 .381 Living with family members .061 .704 Education level .029 .092 Age group -.009 .018 Years in Atlanta -.005 .005 Last month income .076 .045 Unsheltered sleeping locations -.153* .380 Number of disabilities -.271*** .174 Adjusted R2 .020 .106 SE 2.982 2.849 Dependent variable: Self-Efficacy *** p<.001 ** p<.01 * p<.05 156 Table 8.14: Multiple Regression of the Relationship between Centrality of the Homeless Identity and Self-Efficacy (H13D) Variable Model 1 Model 2 Beta SE Beta SE Centrality of homeless identity -.150* .154 -.137* .150 Black (race) .036 .280 Male (gender) .019 .708 Single (relationship) -.002 .227 Veteran -.065 .451 Parent -.007 .400 Living with family members .079 .773 Education level .076 .096 Age group .014 .019 Years in Atlanta -.058 .006 Last month income .080 .047 Unsheltered sleeping locations -.170** .396 Number of disabilities -.278*** .178 Adjusted R2 .019 .127 SE 3.046 2.873 Dependent variable: Self-Efficacy *** p<.001 ** p<.01 * p<.05 157 Hypothesis 14A reads: the longer people are homeless, the lower the self-efficacy. Hypothesis 14B states: the more times people are homeless, the lower the self-efficacy. With respect to hypotheses 14 (A-B), Tables 8.15 and 8.16 demonstrate a correlation between self-efficacy and total length of time homeless (beta = -.227; p<.01) and number of different times homeless (beta = -.147; p<.05) over the past three years in model one only. The relationships are in a negative direction. Therefore, the longer that people are in the homeless role and the more times they are homeless, the lower their self-efficacy. In addition, Table 8.15 model two suggests that living with family members (beta = -.128; p<.05), unsheltered sleeping locations (beta = -.158; p<.05) and number of disabilities (beta = -.260; p<.001) predict self-efficacy. For Table 8.16 model two, only unsheltered sleeping locations (beta = -.161; p<.01) and number of disabilities (beta = -.264; p<.001) have an effect on the dependent variable. Therefore, people living with family members while homeless have a higher self-efficacy than individuals who are homeless and alone. The more likely homeless people are to sleep on the streets rather than in sheltered locations, the lower the self-efficacy. Number of disabilities shows the strongest effect of all the variables towards self-efficacy for both tables. This relationship is also in a negative direction, meaning that the more disabilities homeless people possess, the lower the self-efficacy. An interesting finding is that unsheltered sleeping locations and disabling conditions predict self-efficacy in all the hypotheses that investigate self-efficacy. SUMMARY With hypotheses 7A-B, both total length of time and number of different times homeless over the past three years have an effect on self-esteem; however, only in model one. The results 158 Table 8.15: Multiple Regression of the Relationship between Total Length of Time Homeless and Self-Efficacy (H14A) Variable Model 1 Model 2 Beta SE Beta SE Total length of homelessness (past 3 years) -.227** .014 -.117 .015 Black (race) .053 .257 Male (gender) .024 .645 Single (relationship) .004 .222 Veteran -.026 .445 Parent -.016 .380 Living with family members .128* .760 Education level .046 .091 Age group -.020 .019 Years in Atlanta .020 .006 Last month income .063 .046 Unsheltered sleeping locations -.158* .386 Number of disabilities -.260*** .175 Adjusted R2 .048 .125 SE 2.928 2.808 Dependent variable: Self-Efficacy *** p<.001 ** p<.01 * p<.05 159 Table 8.16: Multiple Regression of the Relationship between Number of Different Times Homeless and Self-Efficacy (H14B) Variable Model 1 Model 2 Beta SE Beta SE Different times homeless (past 3 years) -.147* .131 -.078 .132 Black (race) .057 .269 Male (gender) .045 .650 Single (relationship) -.002 .219 Veteran -.043 .441 Parent -.009 .382 Living with family members .109 .737 Education level .048 .092 Age group -.010 .018 Years in Atlanta -.026 .005 Last month income .077 .045 Unsheltered sleeping locations -.161** .378 Number of disabilities -.264*** .178 Adjusted R2 .018 .109 SE 3.001 2.859 Dependent variable: Self-Efficacy *** p<.001 ** p<.01 * p<.05 160 for hypothesis eight do confirm that accepting the homeless identity does predict self-esteem but only in model one. For hypotheses 9A-D, there is just partial support because centrality of the homeless identity is the only identity theory variable to correlate with self-esteem. Finally, family support demonstrates a significant relationship to self-esteem in model one only (H10). The findings for self-efficacy are similar to the results for self-esteem. Family support also demonstrates a significant relationship to self-efficacy but across both models (H11). Accepting the homeless identity also predicts self-efficacy but for models one and two (H12). For hypotheses 13A-D, there is also only partial support. However, along with centrality of the homeless identity correlating with self-efficacy, so does homeless identity salience. As with self-esteem, both total length of time and number of different times homeless over the last three years have an effect on self-efficacy in model one only (H14A-B). 161 CHAPTER 9 RESULTS FOR HOMELESS IDENTITY MEANINGS AND BEHAVIORS The regression analysis results for homeless identity meanings and behaviors are discussed in this chapter. Homeless identity meanings are analyzed on aspects of the independent variables: homeless identity behaviors, identity theory, time homeless, self-esteem and self-efficacy. On the other hand, homeless identity behaviors are used to examine identity theory, self-esteem and self-efficacy. HOMELESS IDENTITY MEANINGS Relationship between Homeless Identity Behaviors and Meanings Identities are, in part, based on the shared meanings (Burke 1980; Burke and Tully 1977) and behavioral expectations attached to social positions or roles (Stryker and Burke 2000). Identities are stable when there is a match between the meanings involved with occupying the role and the role behaviors that are performed while interacting with others (Burke 1980; Burke and Reitzes 1981). Therefore, hypotheses 15 (A-G) examines the correlation between homeless identity behaviors and meanings. The dependent variable is homeless identity meanings while homeless identity behaviors are the independent variables. Descriptive adjective pairs such as hardworking/lazy, capable/incapable, kind/mean, motivated/unmotivated, friendly/unfriendly and independent/dependent are used to calculate homeless meanings. The identity behaviors of homeless people are measured on several dimensions: safe place to sleep, knowing which people 162 to trust, knowing the best meal sites, helping other homeless people, telling stories of past successes, staying sober and sharing information with other homeless persons. Tables 9.1 to 9.7 present the findings of the multivariate regressions. In regards to hypotheses fifteen (A-G), I find that having a safe place to sleep, knowing who to trust, helping other homeless people and sharing information with others who are homeless demonstrate a correlation with homeless identity meanings for both models (see Tables 9.1, 9.2, 9.4 and 9.7). Staying sober is only significant in model one, while knowing which places serve the best meals and telling stories of past successes have an effect on the homeless identity meanings in model two only as seen in Tables 9.3, 9.5 and 9.6. Thus, the more important the homeless identity behaviors to the respondents, the more likely people will be to describe their homeless identity meanings in positive terms. Model two shows that, after the control variables are added, unsheltered sleeping locations and number of disabilities are good predictors of homeless identity meanings across all the tables. The relationships are in a negative direction which means that individuals sleeping outdoors are more likely to describe themselves in negative terms as opposed to persons staying in sheltered places. Additionally, people with multiple disabilities are more likely to describe themselves in negative terms in relation to individuals with no disabling conditions. A third control variable to correlate with homeless identity meanings is race but only in Table 9.3. The positive relationship indicates that people of black race are more likely to describe themselves in positive terms than others of another race. Hypotheses 15 (A-G) states that the more important behaviors are to the homeless individuals, the more likely people are to describe their homeless identity meanings in negative terms. However, this study finds the opposite. Instead, the results show that the more important 163 Table 9.1: Multiple Regression of the Relationship between Safe Place and Homeless Identity Meanings (H15A) Variable Model 1 Model 2 Beta SE Beta SE Safe place .173** .526 .159** .521 Black (race) .114 .367 Male (gender) .010 .889 Single (relationship) .048 .311 Veteran -.052 .598 Parent -.027 .531 Living with family members -.038 1.051 Education level -.028 .130 Age group -.055 .025 Years in Atlanta -.031 .007 Last month income .012 .063 Unsheltered Sleeping Locations -.149* .535 Number of disabilities -.199** .240 Adjusted R2 .026 .074 SE 3.957 3.858 Dependent variable: Homeless identity meanings *** p<.001 ** p<.01 * p<.05 164 Table 9.2: Multiple Regression of the Relationship between Person to Trust and Homeless Identity Meanings (H15B) Variable Model 1 Model 2 Beta SE Beta SE Person to trust .173** .373 .164** .375 Black (race) .114 .366 Male (gender) .007 .888 Single (relationship) .041 .312 Veteran -.046 .598 Parent -.044 .528 Living with family members -.032 1.054 Education level -.016 .130 Age group -.042 .026 Years in Atlanta -.019 .007 Last month income .007 .063 Unsheltered Sleeping Locations -.146* .535 Number of disabilities -.208*** .240 Adjusted R2 .026 .075 SE 3.957 3.856 Dependent variable: Homeless identity meanings *** p<.001 ** p<.01 * p<.05 165 Table 9.3: Multiple Regression of the Relationship between Best Meal Sites and Homeless Identity Meanings (H15C) Variable Model 1 Model 2 Beta SE Beta SE Best meal site .097 .260 .130* .261 Black (race) .128* .373 Male (gender) .014 .897 Single (relationship) .055 .312 Veteran -.056 .602 Parent -.044 .531 Living with family members -.045 1.054 Education level -.019 .131 Age group -.063 .026 Years in Atlanta -.029 .007 Last month income -.004 .064 Unsheltered Sleeping Locations -.157* .536 Number of disabilities -.205*** .242 Adjusted R2 .006 .065 SE 3.999 3.878 Dependent variable: Homeless identity meanings *** p<.001 ** p<.01 * p<.05 166 Table 9.4: Multiple Regression of the Relationship between Helping Others and Homeless Identity Meanings (H15D) Variable Model 1 Model 2 Beta SE Beta SE Helping others .172** .311 .202*** .309 Black (race) .122 .364 Male (gender) .009 .881 Single (relationship) .058 .307 Veteran -.031 .596 Parent -.049 .524 Living with family members -.055 1.040 Education level -.022 .129 Age group -.050 .025 Years in Atlanta -.021 .007 Last month income .007 .062 Unsheltered Sleeping Locations -.151* .530 Number of disabilities -.226*** .240 Adjusted R2 .026 .090 SE 3.959 3.827 Dependent variable: Homeless identity meanings *** p<.001 ** p<.01 * p<.05 167 Table 9.5: Multiple Regression of the Relationship between Telling Stories and Homeless Identity Meanings (H15E) Variable Model 1 Model 2 Beta SE Beta SE Telling stories .115 .264 .177** .265 Black (race) .118 .366 Male (gender) .003 .885 Single (relationship) .050 .310 Veteran -.030 .600 Parent -.055 .527 Living with family members -.059 1.047 Education level -.030 .129 Age group -.058 .025 Years in Atlanta -.046 .007 Last month income .015 .063 Unsheltered Sleeping Locations -.178** .535 Number of disabilities -.219*** .241 Adjusted R2 .009 .079 SE 3.991 3.849 Dependent variable: Homeless identity meanings *** p<.001 ** p<.01 * p<.05 168 Table 9.6: Multiple Regression of the Relationship between Staying Sober and Homeless Identity Meanings (H15F) Variable Model 1 Model 2 Beta SE Beta SE Staying sober .174** .317 .107 .333 Black (race) .103 .369 Male (gender) .004 .896 Single (relationship) .058 .313 Veteran -.028 .615 Parent -.043 .533 Living with family members -.057 1.059 Education level -.046 .131 Age group -.066 .026 Years in Atlanta -.024 .007 Last month income .011 .064 Unsheltered Sleeping Locations -.145* .543 Number of disabilities -.179** .246 Adjusted R2 .026 .059 SE 3.957 3.890 Dependent variable: Homeless identity meanings *** p<.001 ** p<.01 * p<.05 169 Table 9.7: Multiple Regression of the Relationship between Sharing Information and Homeless Identity Meanings (H15G) Variable Model 1 Model 2 Beta SE Beta SE Share information .232*** .280 .216*** .276 Black (race) .108 .362 Male (gender) .017 .879 Single (relationship) .068 .306 Veteran -.055 .591 Parent -.042 .522 Living with family members -.044 1.036 Education level -.015 .129 Age group -.058 .025 Years in Atlanta -.022 .007 Last month income .006 .062 Unsheltered Sleeping Locations -.156* .527 Number of disabilities -.181** .238 Adjusted R2 .054 .097 SE 3.909 3.812 Dependent variable: Homeless identity meanings *** p<.001 ** p<.01 * p<.05 170 activities are to the homeless individuals, the more likely they will be to discuss themselves in positive descriptors. Relationship between Commitment, Salience and Centrality of the Homeless Identity and Homeless Identity Meanings Hypotheses 16 (A-D) investigate identity factors that influence homeless identity meanings. The dependent variable is homeless identity meanings while the independent variables are (interactive and affective) commitment, salience and centrality of the homeless identity. Thus, the greater the homeless (interactive and affective) commitment, salience and centrality, the more likely the respondents will be to describe the homeless identity meanings in negative terms. The results of the analysis are shown in Tables 9.8 to 9.11. Of the four hypotheses, only affective commitment to the homeless identity (H16B) predicts homeless identity meanings in model one (beta = .160; p<.01) and two (beta = .181; p<.01). The relationship of interactive commitment to the homeless identity with homeless identity meanings is in a positive direction, indicating that the stronger the ties to other homeless people, the more likely they will be to describe themselves in positive terms. This is in the opposite direction as predicted. Across all the hypotheses, only two control variables, unsheltered sleeping locations and number of disabilities, demonstrate a significant relationship with homeless identity meanings (see Tables 9.8-9.11). The relationship between the two control variables and the dependent variable is in a negative direction. This means that individuals sleeping outdoors are more likely to describe themselves in negative terms than people staying in sheltered locations. Additionally, 171 Table 9.8: Multiple Regression of the Relationship between Interactive Commitment to the Homeless Identity and Homeless Identity Meanings (H16A) Variable Model 1 Model 2 Beta SE Beta SE Interactive commitment to homeless identity -.092 .042 -.027 .044 Black (race) .104 .370 Male (gender) .001 .899 Single (relationship) .057 .314 Veteran -.057 .605 Parent -.034 .532 Living with family members -.052 1.028 Education level -.023 .131 Age group -.056 .026 Years in Atlanta -.021 .008 Last month income .002 .063 Unsheltered sleeping locations -.160* .547 Number of disabilities -.181** .241 Adjusted R2 .005 .043 SE 3.998 3.920 Dependent variable: Homeless identity meanings *** p<.001 ** p<.01 * p<.05 172 Table 9.9: Multiple Regression of the Relationship between Affective Commitment to the Homeless Identity and Homeless Identity Meanings (H16B) Variable Model 1 Model 2 Beta SE Beta SE Affective commitment to homeless identity .160** .046 .181** .045 Black (race) .105 .363 Male (gender) .006 .890 Single (relationship) .038 .307 Veteran -.047 .592 Parent -.028 .518 Living with family members -.069 1.017 Education level .000 .128 Age group -.073 .025 Years in Atlanta -.025 .007 Last month income -.004 .062 Unsheltered sleeping locations -.185** .527 Number of disabilities -.183** .234 Adjusted R2 .022 .077 SE 3.946 3.832 Dependent variable: Homeless identity meanings *** p<.001 ** p<.01 * p<.05 173 Table 9.10: Multiple Regression of the Relationship between Salience of the Homeless Identity and Homeless Identity Meanings (H16C) Variable Model 1 Model 2 Beta SE Beta SE Salience of homeless identity .008 .237 .074 .239 Black (race) .102 .365 Male (gender) -.003 .884 Single (relationship) .059 .309 Veteran -.056 .607 Parent -.034 .528 Living with family members -.050 1.012 Education level -.025 .129 Age group -.061 .025 Years in Atlanta -.033 .007 Last month income .003 .063 Unsheltered sleeping locations -.197** .534 Number of disabilities -.202** .239 Adjusted R2 -.004 .061 SE 3.984 3.854 Dependent variable: Homeless identity meanings *** p<.001 ** p<.01 * p<.05 174 Table 9.11: Multiple Regression of the Relationship between Centrality of the Homeless Identity and Homeless Identity Meanings (H16D) Variable Model 1 Model 2 Beta SE Beta SE Centrality of homeless identity .035 .212 .046 .212 Black (race) .087 .400 Male (gender) .034 .992 Single (relationship) .069 .325 Veteran -.053 .629 Parent -.021 .560 Living with family members -.083 1.122 Education level .000 .136 Age group -.055 .027 Years in Atlanta -.002 .008 Last month income -.007 .067 Unsheltered sleeping locations -.171* .563 Number of disabilities -.182** .248 Adjusted R2 -.003 .041 SE 4.035 3.946 Dependent variable: Homeless identity meanings *** p<.001 ** p<.01 * p<.05 175 people with multiple disabilities are more likely to describe themselves in negative terms in relation to individuals with no disabling conditions. In conclusion, the findings do not support hypotheses 16 (A-D). Affective commitment to the homeless identity does have an effect on homeless identity meanings but in the opposite direction predicted, meaning that the stronger the ties with other homeless people, the better for thinking of the self in positive terms. Neither interactive commitment, salience nor centrality of the homeless identity have a correlation with homeless identity meanings. Relationship between Time Homeless and Homeless Identity Meanings Chronically homeless persons are more likely to speak negatively about themselves than the newly homeless (Boydell et al. 2000). Thus, hypotheses 17 (A-B) examines the degree to which individuals describe themselves based on the time spent in the homeless role. Homeless identity meanings is the dependent variable while the independent variables are total length of homeless and number of different times homeless over the past three years. Tables 9.12 and 9.13 show the results of hypotheses 17 (A-B). Total length of time homeless (beta = -.210; p<.001) and number of different time homeless (beta = -.132; p<.05) over the past three years do predict homeless identity meanings but only for model one. As expected, the relationship is in a negative direction. Therefore, people who are chronically homeless are more likely to speak about themselves using negative adjectives such as being unmotivated rather than individuals who are newly homeless. This finding supports hypotheses seventeen (A-B) and previous researchers. In model two, unsheltered sleeping locations and number of disabilities show an effect on homeless identity meanings for both Tables 9.12 and 9.13. Therefore, individuals sleeping on the 176 Table 9.12: Multiple Regression of the Relationship between Total Length of Time Homeless and Homeless Identity Meanings (H17A) Variable Model 1 Model 2 Beta SE Beta SE Total length of homelessness (past 3 years) -.210*** .019 -.128 .020 Black (race) .097 .367 Male (gender) -.012 .901 Single (relationship) .064 .319 Veteran -.040 .615 Parent -.031 .537 Living with family members -.062 1.105 Education level .008 .130 Age group -.073 .026 Years in Atlanta .019 .008 Last month income .018 .064 Unsheltered sleeping locations -.165* .549 Number of disabilities -.179** .242 Adjusted R2 .044 .074 SE 3.926 3.855 Dependent variable: Homeless identity meanings *** p<.001 ** p<.01 * p<.05 177 Table 9.13: Multiple Regression of the Relationship between Number of Different Times Homeless and Homeless Identity Meanings (H17B) Variable Model 1 Model 2 Beta SE Beta SE Different times homeless (past 3 years) -.132* .180 -.077 .187 Black (race) .115 .382 Male (gender) -.002 .900 Single (relationship) .040 .314 Veteran -.042 .611 Parent -.015 .535 Living with family members -.042 1.060 Education level -.009 .130 Age group -.064 .025 Years in Atlanta -.021 .008 Last month income .003 .063 Unsheltered sleeping locations -.164** .534 Number of disabilities -.181** .246 Adjusted R2 .014 .052 SE 3.972 3.893 Dependent variable: Homeless identity meanings *** p<.001 ** p<.01 * p<.05 178 streets are more likely to describe themselves in a negative way as opposed to persons staying in sheltered locations. People with multiple disabilities are more likely to describe themselves in negative terms in relation to individuals without disabling conditions. Theses variable relationships exist no matter the length of time and number of times homeless over the past three years. Number of disabilities, according to the beta, has the strongest relationship with identity meaning than any other variable in model two for both tables. Self-Esteem and Self-Efficacy predict Homeless Identity Meanings The following hypotheses test the prediction that self-esteem and self-efficacy will have an effect on homeless identity meanings due to the meanings being measured with descriptive adjectives such as hardworking and friendly. Hypothesis 18A reads: The lower the self-esteem, the more likely the respondents will be to describe their homeless identity meanings in negative terms. Hypothesis 18B states: The lower the self-efficacy,the more likely the respondents will be to describe their homeless identity meanings in negative terms. The results displayed in Tables 9.14 and 9.15 are consistent with hypotheses 18 (A-B). For hypothesis 18A, the significant relationship between self-esteem and homeless identity meanings is found across both model one (beta = .273; p<.001) and model two (beta = .214; p<.001). Additionally, self-efficacy shows a correlation with homeless identity meanings in model one (beta = .292; p<.001) and two (beta = .233; p<.001) for hypothesis 18B. These relationships are in positive directions. Therefore, homeless people with high self-esteem and self-efficacy are more likely to describe themselves in positive terms such as hardworking and capable than those with low self-esteem and self-efficacy. 179 Table 9.14: Multiple Regression for Relationship between Self-Esteem and Homeless Identity Meanings (H18A) Variable Model 1 Model 2 Beta SE Beta SE Self-esteem .273*** .047 .214*** .051 Black (race) .089 .362 Gender (male) .011 .878 Single (relationship) .057 .304 Veteran -.052 .590 Parent -.039 .518 Living with family members -.036 1.005 Education level -.052 .129 Age group -.059 .025 Years in Atlanta -.037 .007 Last month income -.011 .062 Unsheltered sleeping locations -.145* .525 Number of disabilities -.118 .246 Adjusted R2 .071 .086 SE 3.858 3.826 Dependent variable: Homeless identity meanings *** p<.001 ** p<.01 * p<.05 180 Table 9.15: Multiple Regression of the Relationship between Self-Efficacy and Homeless Identity Meanings (H18B) Variable Model 1 Model 2 Beta SE Beta SE Self-efficacy .292*** .080 .233*** .084 Black (race) .097 .359 Male (gender) -.011 .873 Single (relationship) .052 .302 Veteran -.048 .587 Parent -.032 .514 Living with family members -.065 .998 Education level -.026 .126 Age group -.060 .025 Years in Atlanta -.018 .007 Last month income -.010 .061 Unsheltered sleeping locations -.140* .523 Number of disabilities -.125* .240 Adjusted R2 .082 .097 SE 3.836 3.804 Dependent variable: Homeless identity meanings *** p<.001 ** p<.01 * p<.05 181 For model two, unsheltered sleeping locations is also a predictor of homeless identity meanings for both Tables 9.14 (beta = .145; p<.001) and 9.15 (beta = .140; p<.05). This means that individuals sleeping in unsheltered places are more likely to think of themselves negatively as opposed to people in sheltered housing. Number of disabilities provides a third significant relationship for homeless identity meanings but only in Table 68. People with multiple disabilities are more likely to describe themselves in negative terms in relation to individuals with no disabling conditions. According to the betas, the variables with the strongest relationships to homeless identity meaning are self-esteem in Table 9.14 and self-efficacy in Table 9.15 for model one. HOMELESS IDENTITY BEHAVIORS Homeless Identity Behaviors predict Commitment, Salience and Centrality of the Homeless Identity Hypotheses 19 (A-D) examine factors that influence how homeless identity behaviors effect identity outcomes. The first hypothesis (H19A) states: the more important the homeless identity behaviors are to individuals, the greater the interactive commitment to the homeless identity. Hypothesis 19B reads: the more important the homeless identity behaviors are to individuals, the greater the affective commitment to the homeless identity. Hypothesis 19C states: the more important the homeless identity behaviors are to individuals, the greater the salience of the homeless identity. Hypothesis 19D reads: the more important the homeless identity behaviors are to individuals, the greater the centrality of the homeless identity. For hypothesis 19A, knowing which sites serve the best tasting meals only has an effect on the dependent variable in model one (beta = .120; p<.05) according to Table 9.16. On the other 182 Table 9.16: Multiple Regression of the Relationship between Best Meal Site and Interactive Commitment to the Homeless Identity (H19A-1) Variable Model 1 Model 2 Beta SE Beta SE Best meal site .120* .375 .104 .380 Black (race) .021 .529 Male (gender) .005 1.288 Single (relationship) -.107 .438 Veteran .041 .878 Parent .009 .763 Living with family members .058 1.451 Education level -.133* .187 Age group .161* .036 Years in Atlanta .119 .011 Last month income -.054 .091 Unsheltered sleeping locations .198** .769 Number of disabilities .032 .351 Adjusted R2 .011 .071 SE 5.866 5.685 Dependent variable: Interactive commitment to the homeless identity *** p<.001 ** p<.01 * p<.05 183 hand, helping other people who are homeless, telling stories of past successes, staying sober and sharing information with other homeless individuals predict the interactive commitment to the homeless identity in both models (see Tables 9.17-9.20). Thus, the more important finding a meal site that serves good tasting food, helping others, telling stories and sharing information are to homeless individuals, the higher the number of friends who are also homeless. The previous behaviors have a positive relationship with the dependent variable while staying sober shows a negative relationship, indicating that the more important staying sober is to homeless persons, the fewer friends they will have who are also homeless. The homeless identity behaviors, finding a safe place to sleep and knowing which persons to trust, have no correlation with the interactive commitment to the homeless identity. For model two across all five tables (9.16-9.20), education level, age and unsheltered sleeping locations are predictors of the interactive commitment to the homeless identity. Additionally, years in Atlanta has an effect on the dependent variable in Table 9.17. Age, unsheltered sleeping locations and years in Atlanta have a positive relationship with the dependent variable while the relationship with education level is in a negative direction. Therefore, people with a high school education or less, older individuals, those sleeping outdoors and persons living in Atlanta for considerable length of time have a greater interactive commitment to the homeless identity than college educated people, younger individuals, persons staying in sheltered locations and those living in Atlanta for only a short period of time. Helping other homeless people, telling stories of past successes and sharing information with other homeless individuals are the homeless identity behaviors that show a significant relationship with affective commitment to the homeless identity across both models for hypothesis 19B (see Tables 9.21-9.23). This means that the more important helping others, 184 Table 9.17: Multiple Regression of the Relationship between Helping Others and Interactive Commitment to the Homeless Identity (H19A-2) Variable Model 1 Model 2 Beta SE Beta SE Helping others .225*** .440 .227*** .440 Black (race) .029 .511 Male (gender) .007 1.252 Single (relationship) -.108 .427 Veteran .067 .861 Parent .005 .747 Living with family members .045 1.420 Education level -.137* .182 Age group .177** .036 Years in Atlanta .124* .011 Last month income -.041 .089 Unsheltered sleeping locations .201*** .752 Number of disabilities .004 .347 Adjusted R2 .047 .111 SE 5.757 5.561 Dependent variable: Interactive commitment to the homeless identity *** p<.001 ** p<.01 * p<.05 185 Table 9.18: Multiple Regression of the Relationship between Telling Stories and Interactive Commitment to the Homeless Identity (H19A-3) Variable Model 1 Model 2 Beta SE Beta SE Telling stories .234*** .370 .206*** .380 Black (race) .017 .512 Male (gender) .002 1.258 Single (relationship) -.114 .431 Veteran .069 .868 Parent -.004 .752 Living with family members .035 1.433 Education level -.141* .183 Age group .166** .036 Years in Atlanta .096 .011 Last month income -.038 .090 Unsheltered sleeping locations .173** .760 Number of disabilities .014 .347 Adjusted R2 .051 .100 SE 5.744 5.594 Dependent variable: Interactive commitment to the homeless identity *** p<.001 ** p<.01 * p<.05 186 Table 9.19: Multiple Regression of the Relationship between Staying Sober and Interactive Commitment to the Homeless Identity (H19A-4) Variable Model 1 Model 2 Beta SE Beta SE Staying sober -.194*** .444 -.160** .459 Black (race) .006 .515 Male (gender) -.013 1.268 Single (relationship) -.087 .435 Veteran .017 .885 Parent .000 .759 Living with family members .063 1.441 Education level -.132* .185 Age group .173** .036 Years in Atlanta .117 .011 Last month income -.054 .090 Unsheltered sleeping locations .171** .771 Number of disabilities .013 .352 Adjusted R2 .034 .084 SE 5.796 5.643 Dependent variable: Interactive commitment to the homeless identity *** p<.001 ** p<.01 * p<.05 187 Table 9.20: Multiple Regression of the Relationship between Share Information and Interactive Commitment to the Homeless Identity (H19A-5) Variable Model 1 Model 2 Beta SE Beta SE Share information .191** .400 .185** .390 Black (race) .009 .512 Male (gender) .008 1.264 Single (relationship) -.097 .431 Veteran .039 .866 Parent .003 .753 Living with family members .063 1.432 Education level -.134* .183 Age group .161* .036 Years in Atlanta .121 .011 Last month income -.047 .090 Unsheltered sleeping locations .195** .758 Number of disabilities .048 .346 Adjusted R2 .033 .096 SE 5.799 5.608 Dependent variable: Interactive commitment to the homeless identity *** p<.001 ** p<.01 * p<.05 188 Table 9.21: Multiple Regression of the Relationship between Helping Others and Affective Commitment to the Homeless Identity (H19B-1) Variable Model 1 Model 2 Beta SE Beta SE Helping others .368*** .379 .380*** .393 Black (race) .034 .459 Male (gender) .043 1.130 Single (relationship) .058 .384 Veteran .000 .771 Parent -.021 .666 Living with family members .008 1.284 Education level -.092 .162 Age group .074 .032 Years in Atlanta .034 .009 Last month income .026 .080 Unsheltered sleeping locations .130* .673 Number of disabilities -.069 .309 Adjusted R2 .132 .125 SE 4.958 4.978 Dependent variable: Affective commitment to the homeless identity *** p<.001 ** p<.01 * p<.05 189 Table 9.22: Multiple Regression of the Relationship between Telling Stories and Affective Commitment to the Homeless Identity (H19B-2) Variable Model 1 Model 2 Beta SE Beta SE Telling stories .391*** .316 .395*** .335 Black (race) .018 .455 Male (gender) .035 1.124 Single (relationship) .047 .383 Veteran .009 .769 Parent -.039 .664 Living with family members -.010 1.282 Education level -.097 .162 Age group .057 .032 Years in Atlanta -.020 .010 Last month income -.034 .079 Unsheltered sleeping locations .079 .674 Number of disabilities -.055 .306 Adjusted R2 .149 .132 SE 4.908 4.956 Dependent variable: Affective commitment to the homeless identity *** p<.001 ** p<.01 * p<.05 190 Table 9.23: Multiple Regression of the Relationship between Share Information and Affective Commitment to the Homeless Identity (H19B-3) Variable Model 1 Model 2 Beta SE Beta SE Share information .261*** .354 .262*** .359 Black (race) -.003 .474 Male (gender) .041 1.176 Single (relationship) .076 .400 Veteran -.045 .799 Parent -.023 .693 Living with family members .036 1.335 Education level -.092 .169 Age group .047 .033 Years in Atlanta .027 .010 Last month income .017 .083 Unsheltered sleeping locations .119 .700 Number of disabilities .001 .318 Adjusted R2 .065 .055 SE 5.146 5.174 Dependent variable: Affective commitment to the homeless identity *** p<.001 ** p<.01 * p<.05 191 telling stories and sharing information is to homeless people, the greater the strength of attachment to the homeless identity. Finding a safe place to sleep, knowing who to trust, knowing which meal sites serve best tasting food and staying sober are the independent variables that do not have a correlation with the dependent variable. In model two for Tables 9.22 and 9.23, only the homeless identity behaviors demonstrate a significant relationship with the dependent variable. Only in Table 9.21 does a homeless identity behavior and a control variable, unsheltered sleeping locations, both predict affective commitment to the homeless identity. Therefore, individuals sleeping in unsheltered locations have a greater affective commitment to the homeless identity compared to those staying in sheltered facilities. With hypothesis 19C, helping others who are homeless, telling stories of past successes and sharing information are the independent variables that demonstrate a relationship with salience of the homeless identity across both models (see Tables 9.24, 9.25 and 9.27). Thus, the more important helping others, telling stories and sharing information are to people who are homeless, the greater the salience of the homeless identity. Staying sober is a predictor of the dependent variable in model one only (beta = -.161; p<.001) according to Table 9.26. This relationship is in a negative direction, indicating that the more important staying sober is to homeless individuals, the less likely they are to invoke the homeless identity in different situations. Sleeping in unsheltered locations is a control variable that correlates with salience of the homeless identity in Tables 9.24, 9.26 and 9.27. The control variable, veterans, is only significant with the dependent variable for Table 9.25 while number of disabilities only has an effect in Table 9.27. Therefore, people who sleep outside, veterans and those with disabilities 192 Table 9.24: Multiple Regression of the Relationship between Helping Others and Salience of the Homeless Identity (H19C-1) Variable Model 1 Model 2 Beta SE Beta SE Helping others .128* .080 .134* .081 Black (race) .091 .094 Male (gender) .034 .229 Single (relationship) -.084 .079 Veteran .119 .160 Parent .108 .138 Living with family members -.049 .261 Education level -.038 .033 Age group .065 .007 Years in Atlanta .089 .002 Last month income -.071 .016 Unsheltered sleeping locations .145* .139 Number of disabilities .099 .064 Adjusted R2 .013 .054 SE 1.041 1.019 Dependent variable: Salience of the homeless identity *** p<.001 ** p<.01 * p<.05 193 Table 9.25: Multiple Regression of the Relationship between Telling Stories and Salience of the Homeless Identity (H19C-2) Variable Model 1 Model 2 Beta SE Beta SE Telling stories .228* .066 .228* .068 Black (race) .093 .092 Male (gender) .036 .225 Single (relationship) -.096 .077 Veteran .133* .158 Parent .099 .136 Living with family members -.064 .257 Education level -.037 .033 Age group .062 .006 Years in Atlanta .058 .002 Last month income -.064 .016 Unsheltered sleeping locations .121 .137 Number of disabilities .094 .062 Adjusted R2 .048 .086 SE 1.022 1.001 Dependent variable: Salience of the homeless identity *** p<.001 ** p<.01 * p<.05 194 Table 9.26: Multiple Regression of the Relationship between Staying Sober and Salience of the Homeless Identity (H19C-3) Variable Model 1 Model 2 Beta SE Beta SE Staying sober -.161** .081 -.104 .085 Black (race) .078 .093 Male (gender) .022 .230 Single (relationship) -.071 .079 Veteran .089 .162 Parent .107 .138 Living with family members -.038 .262 Education level -.035 .034 Age group .064 .007 Years in Atlanta .085 .002 Last month income -.079 .016 Unsheltered sleeping locations .125* .141 Number of disabilities .104 .064 Adjusted R2 .022 .047 SE 1.036 1.023 Dependent variable: Salience of the homeless identity *** p<.001 ** p<.01 * p<.05 195 Table 9.27: Multiple Regression of the Relationship between Share Information and Salience of the Homeless Identity (H19C-4) Variable Model 1 Model 2 Beta SE Beta SE Share information .171** .071 .173** .070 Black (race) .081 .092 Male (gender) .041 .228 Single (relationship) -.078 .078 Veteran .099 .158 Parent .106 .137 Living with family members -.036 .259 Education level -.033 .033 Age group .055 .007 Years in Atlanta .087 .002 Last month income -.076 .016 Unsheltered sleeping locations .143* .138 Number of disabilities .127* .063 Adjusted R2 .029 .113 SE 1.034 1.012 Dependent variable: Salience of the homeless identity *** p<.001 ** p<.01 * p<.05 196 experience a greater salience of the homeless identity than individuals staying in sheltered locations, non-veterans and persons without disabling conditions. Whereas several of the homeless identity behaviors predict commitment and salience to the homeless identity, none of the homeless identity behaviors have an effect on centrality of the homeless identity. This is probably due to the homeless identity not ranking high on the centrality hierarchy. Since there is no significant relationship, no tables are shown for this particular hypothesis. In conclusion, these results show limited support for hypotheses 19 (A-D). Specifically, centrality of the homeless identity has no significant relationships with any of the homeless identity behaviors. Of the control variables, unsheltered sleeping locations predicts the dependent variable in all but three of the tables (Tables 9.22, 9.23 and 9.25). Relationship between Staying Sober and Self-Esteem and Self-Efficacy Hypotheses twenty (A-B) examine the relationship between a specific homeless identity behavior, staying sober, and self-esteem and self-efficacy. Hypothesis 20A states: the more important staying sober is to individuals, the higher the self-esteem. Hypothesis 20B reads: the more important staying sober is to individuals, the higher the self-efficacy. The results of the analysis are presented in Tables 9.28 and 9.29. Staying sober does predict self- esteem in model one (beta = .297; p<.001) and model two (beta = .226; p<.001). Additionally, staying sober has an effect on self-efficacy in models one (beta = .233; p<.001) and two (beta = .158; p<.01). Of the control variables, education level and number of disabilities have a significant relationship with the dependent variables, self-esteem and self-efficacy for both tables. Thus, the 197 Table 9.28: Multiple Regression of the Relationship between Staying Sober and Self-esteem (H20A) Variable Model 1 Model 2 Beta SE Beta SE Staying sober .297*** .373 .226*** .372 Black (race) .081 .422 Male (gender) -.027 1.038 Single (relationship) -.013 .355 Veteran .014 .723 Parent .030 .619 Living with family members -.075 1.180 Education level .126* .151 Age group -.020 .030 Years in Atlanta .058 .009 Last month income .091 .074 Unsheltered sleeping locations -.073 .630 Number of disabilities -.295*** .287 Adjusted R2 .088 .184 SE 4.895 4.620 Dependent variable: Self-esteem *** p<.001 ** p<.01 * p<.05 198 Table 9.29: Multiple Regression of the Relationship between Staying Sober and Self-efficacy (H20B) Variable Model 1 Model 2 Beta SE Beta SE Staying sober .233*** .225 .158** .228 Black (race) .058 .258 Male (gender) .063 .636 Single (relationship) .001 .218 Veteran -.021 .443 Parent -.013 .379 Living with family members .085 .723 Education level .006* .093 Age group -.002 .018 Years in Atlanta -.023 .005 Last month income .093 .045 Unsheltered sleeping locations -.111 .386 Number of disabilities -.275*** .176 Adjusted R2 .051 .128 SE 2.956 2.832 Dependent variable: Self-efficacy *** p<.001 ** p<.01 * p<.05 199 higher the education level, the higher the self-esteem and self-efficacy. Number of disabilities has a negative relationship with the dependent variables. Therefore, the more disabling conditions that homeless people have, the lower the self-esteem and self-efficacy. SUMMARY In conclusion, hypotheses 15 (A-G) show a match between the homeless identity meanings and behaviors: safe place, person to trust, best meal sites, helping others, telling stories, staying sober and sharing information. Affective commitment to the homeless identity demonstrates a significant relationship with homeless identity meanings across both models (H16B) while interactive commitment, salience and centrality of the homeless identity do not predict the dependent variable. For hypotheses 17 (A-B), total length of time and number of different times homeless over the past three years predict homeless identity meanings but only in model one. Both self-esteem (H18A) and self-efficacy (H18B) have an effect on homeless identity meanings. The final hypotheses investigate homeless identity behaviors as the independent variables. In relation to identity theory, several homeless identity behaviors are able to predict interactive commitment (best meal sites , helping others, telling stories, staying sober and sharing information), affective commitment (helping others, telling stories and sharing information) and salience (helping others, telling stories, staying sober and sharing information) of the homeless identity. However, no homeless identity behaviors show an effect on centrality of the homeless identity. With hypotheses 20 (A-B), staying sober is a good predictor of both self-esteem and self-efficacy. 200 CHAPTER 10 CONCLUSION AND IMPLICATIONS My goal for the dissertation has been to answer the following questions: While homeless, do individuals adopt a homeless identity? Is a homeless identity the primary or most salient identity for people while in the situation? How committed are people to the homeless identity during their homelessness? Do people think of themselves in positive or negative terms while homeless? What activities are important for individuals who are homeless? How does being homeless affect people’s self-worth? What attempts do individuals make to exit the homeless situation? This chapter will discuss the findings and implications of the study. CONTRIBUTIONS OF THIS STUDY The dissertation has been innovative and significant in several ways. First, whereas past research investigated homeless identity, the meaning of being homeless and the self-worth of homeless people by conducting ethnographies, this study tested these variables utilizing a standardized questionnaire. A survey provides the ability to test for validity, reliability and statistically significant relationships among the multiple variables. Second, in this study I built on the previous field research of Snow and Anderson (1994). Whereas Snow and Anderson (1994) predominately discussed the construction of a homeless identity in terms of whether people embrace it or not, I examined the homeless identity theoretically (and empirically) using Stryker’s structural identity theory: identity salience, centrality and commitment (Stryker and Serpe 1994). Additionally, I applied a new approach to 201 measuring identity salience with a survey question based on conversational topics with others in various situations such as with family and at service provider agencies. This measurement is different than past research in which students were asked about meeting people for the first time and what they would tell others about themselves (Stryker and Serpe 1994). Third, most self-esteem research has focused on assessing traditional role identities. The premiere self-esteem scale, developed by Rosenberg (1979), was tested using high school students. This study instead focuses on the implications of a negative or stigmatized identity on the self-worth of people. Finally, this project expands on the common pathology model of homelessness, which focuses on fixing the problems of individuals, and applies a new identity centered approach to policies and program development. This new method highlights and evaluates other identities that homeless people possess such as parent and worker, instead of only focusing on the stigmatized homeless identity. This new approach also focuses on the positive meanings, activities and self-worth of people who are homeless by emphasizing their strong survival skills, ability to provide emotional support to other homeless people and their mid to high level of self-esteem. ACCEPTING THE HOMELESS IDENTITY People without their own permanent homes who stay in unsheltered locations, emergency shelters or transitional housing are often placed by others into an established category based on a social identity of homelessness. The homeless situation and corresponding identity is not seen as fitting the traditional American cultural norms. Therefore, people in the situation are labeled as stigmatized and seen as reduced in value (Goffman 1963). 202 According to Snow and Anderson (1992; 1993), to combat this negative label, homeless people discuss themselves with others at a personal level. The verbal claiming of a personal identity by homeless individuals occurs because they cannot identify themselves through other material means such as housing or vehicles. Emphasizing the personal identity is done to show that the homeless identity is inconsistent from their desired self-concept. The distancing of themselves from the homeless identity and the focus instead on their personal identity allows individuals to preserve a measure of their self-worth (Snow and Anderson 1992; 1993). Whereas some homeless people distance themselves from the homeless identity, others come to embrace the homeless identity and to discuss their homeless status more often in conversations. Thus for these individuals, their social and personal identities are more congruent. This implies an acceptance of the social identity and its traits, which according to Snow and Anderson (1992) tends to occur over time. Supporting Snow and Anderson (1992), this research does find that the longer and more often that people are homeless, the more likely they will be to accept the homeless identity. A majority (61 percent) of the study respondents do identify as homeless when directly asked. Besides times homeless, the study also discovered two other factors that affect accepting the homeless identity: people sleeping on the streets and those with multiple disabilities. Service agency personnel would do well to understand that homeless people have identity and self-concept needs as well as needs for food and shelter. It is essential for agency staff to be aware of the implications of identifying as homeless for people and how acceptance of the role plays into their situation. Perhaps accepting the homeless role, along with its values, expectations and social relationships, is a matter of survival for people. Survival while homeless means learning and mastering the skills of finding food, shelter and additional assistance such as 203 medical care. It also means preserving the self during this difficult situation. The homeless identity may become a part of their self-concept during this time in order to cope. Accepting and incorporating the homeless identity while maintaining an overall positive self-concept is a process that takes time and work. Thus, it is a struggle for people to protect and maintain their self-concept as well as a struggle to survive while homeless. IDENTITY THEORY Salience and Centrality of the Homeless Identity People occupy multiple roles in society and thus possess as many identities or selves as they do roles (James 1890). Whereas Snow and Anderson (1992) focus on the homeless role as a social identity, Stryker (1989) would see being homeless as a role identity. For Stryker, multiple role identities of the self are organized into a hierarchal structure. This hierarchy is based on identity salience, the probability of consistently invoking an identity in different situations (Stryker and Serpe 1994). For this study, the homeless identity demonstrates the greatest salience for the respondents (38 percent). Identity centrality relates to how important the role identity is to individuals, and often represents the desirability from a personal point of view as to the preferred identity. Among respondents, only nine chose the homeless identity as their most central identity. In fact, less than a quarter (21 percent) of the respondents rank the homeless identity as one of their five most important identities. The homeless identity is only picked eighth overall for identity centrality, meaning the homeless identity is not highly ranked. Instead, the majority of the respondents (78 percent) indicate that the family identity is one of the top five highest in importance and is chosen most often as the primary central identity. 204 As previously mentioned, when asked directly if they identify as homeless, the majority of respondents (61 percent) indicate that they do. This means that most homeless people accept the homeless identity, invoking it most often in different situation but not placing it high in the identity centrality hierarchy. This is probably due to homelessness being a stigmatizing identity. On the other hand, the most important identity, the family identity, is a positive one. However, the family identity ranks only fifth in salience overall. According to McCall and Simons (1978), less prominent identities can get activated in situations due to pressure to follow the cultural norms or from others. Perhaps, a factor for invoking the homeless identity over others is the need to survive in the situation. At provider agencies, obtaining services such as a free meal or place to sleep is only possible by invoking the homeless status. It is also helpful for individuals to have homeless friends who can provide information on where to get services and for safety. Thus, people are invoking their situational self to receive assistance from service providers and other homeless people. In these situations, they are not invoking their ideal self, how individuals like to see themselves given what is important to them, which is based off the prominence hierarchy of identities (McCall and Simons 1978). For example, some unaccompanied homeless people may hold being good parents as their most important identity. Unfortunately, they are unable to perform the parent role adequately because in this situation they have no phone to call their children, no transportation to visit or no home to have the children over for dinner. Others may not want to contact their children because they do not want to be seen as a burden on them or are embarrassed to have their children see them in this situation. Typically, families are homeless a shorter length of time than individuals (Parker 2009). Perhaps it is beneficial that heads of homeless households are seen as possessing 205 the positive central identity of mother or father to counter balance the stigmatizing salient identity of being homeless. The significance of the homeless identity salience can also be seen by the length of time spent in the homeless role. The current length of time homeless has an effect on invoking the identity more frequently across various situations but has no predictive value as a prominent identity. This means that even if people are homeless for years, they will not see the homeless identity as an important identity in comprising their self-concept. During this study, other factors have been found that influence homeless identity salience including sleeping on the streets, increasing number of disabilities and being a veteran. Usually when homeless people first apply for services they are asked a series of questions around basic demographics such as age, race, disabling condition and income. Perhaps, service providers could also ask questions related to positive and central identities such as parental status, friend relationships and job skills. Then the case worker could ask about their children, friendships or even job opportunities, instead of just focusing on the negative aspects of the homeless situation when homeless people come in for assistance. Interactive and Affective Commitment to the Homeless Identity Overall, the majority of homeless respondents (53 percent) demonstrate a low affective commitment to the homeless identity. This type of commitment is qualitative based on the strength of ties to other homeless people. The quantitative dimension of identity commitment is interactive, based on the number of friends who share the same role identity, and it is also low. Almost half (46 percent) of the respondents have either no homeless friends or one to two friends who are homeless. Thus, commitment to the homeless identity, along with homeless identity 206 centrality, is low for individuals whereas salience to the homeless identity is high. This is probably due to the homeless identity being thought of as negative. This study does find, however, that there are a few factors which influence commitment to the homeless identity resulting in a high attachment to the homeless role. Factors that correlate with an increasing number of friends who are also homeless include having a high school diploma or less, increasing age and living in Atlanta for a longer length of time. Additionally, sleeping outdoors increases both the number of homeless friends and the strength of those friendships. According to Stryker (Stryker and Serpe 1994), salience and centrality are functions of people’s commitment to an identity. This study finds that interactive commitment of the homeless identity does predict homeless identity salience. Thus, an increasing number of friends who are homeless results in an increasing frequency of invoking homelessness in different situations. On the other hand, affective commitment to the homeless identity does not correlate with homeless identity salience and neither interactive nor affective commitment to the homeless have a significant relationship with homeless identity centrality. Thus, interactive commitment to the homeless identity predicts homeless identity salience which predicts length of time in role. Further research needs to occur to understand the implications for the self-concept when there is a mismatch between the identity salience hierarchy (Stryker and Serpe 1994) and the identity centrality or prominence hierarchy (McCall and Simons 1978). The study indicates that a stigmatized identity results in high salience but low centrality of the homeless identity. With homeless identity commitment, a greater attachment to the identity occurs with an increased length of time in the role. 207 For example, heavy drug users would probably not choose the addict identity as one of their top five important identities. However, I would believe that they invoke the addict identity quite frequently across situations such as when purchasing the drugs, hanging out with friends who are addicted and in encounters with police officers. As with being homeless, the problem is that the stigmatizing identity is not just the physical addiction to the drug but also the incorporation of the negative identity, along with its meanings and behaviors, into the self-concept. Additionally, the longer people heavily use drugs, the more they will be attached to the addict role by having an increasing number of friends who are also addicts with the ties to those friend friends becoming stronger over time. That is why drug rehabilitation is not just about the physical addiction to the drug but also involves intensive counseling. HOMELESS IDENTITY MEANINGS Identities are based on the shared multiple meanings attached to the roles. The meanings of identities are based on bipolar adjective pairs that occur along several dimensions (Burke 1980; Burke and Tully 1977). For the homeless identity, the stereotypical adjective pairs include lazy (hardworking-lazy), incapable (capable-incapable), dishonest (honest-dishonest), mean (kind-mean), unmotivated (motivated-unmotivated), unfriendly (friendly-unfriendly) and dependent (independent-dependent). Among the dimensional pairs, various people will pick different adjectives for that same identity. On average, the majority of respondents (97 percent) thought positively about themselves, that they are hardworking, capable, honest, kind, motivated, friendly and independent. A possible reason for this choice is that the stereotypical adjective pairs are often chosen by housed people in comparison to individuals who are homeless. However, homeless people are more than likely 208 comparing themselves to others who are homeless. This comparison is based on homelessness as a social identity (Tajfel and Turner 1986). Accordingly, the longer that people are homeless, the more likely they will be to stop making social comparisons with other groups such as domiciled persons and instead start comparing themselves to other people who are also homeless such as those who are mentally ill (Farrington and Robinson 1999). For example, Snow and Anderson (1992) would suggest that among the homeless having a positive identity of being a trustworthy friend who protect their homeless friends in times of trouble and willingly share their limited resources when they can is a good thing. In fact, this study finds that the stronger the ties to other people who are also homeless, the more positive the homeless identity is seen as. Perhaps, this homeless group views themselves as capable because they are able to stay safe and find food and shelter, often traveling long distances to get assistance, every single day for months or even years. This also indicates that they are hardworking, just not in the conventional sense of a job. On the other hand, there are a few factors that influence people in describing themselves more negatively among the adjective pair choices. Across all analysis with homeless identity meanings, when people are sleeping in unsheltered locations and for persons with multiple disabilities, they think of themselves negatively. In addition, this study does find that the longer and more often people are homeless, the more likely they will be to discuss themselves negatively. Thus, as service providers assist those who are chronically homeless, they need to realize that these individuals are facing not just physical obstacles but also mental obstacles. They may think of themselves as lazy, incapable, dishonest, unkind, unmotivated, unfriendly and dependent. 209 For example, the very institutions that aim to help people get off the streets could be perpetuating the idea that homeless individuals are dependent on institutions to survive. HOMELESS IDENTITY BEHAVIORS These shared meanings link the identity to its corresponding behaviors. By determining the meanings of the identity, we can thus predict behaviors that accompany the identity (Burke 1980; Burke and Reitzes 1981; Burke and Tully 1977). Several behaviors attached to the homeless identity include finding a safe place to sleep, finding meal sites that serve decent food, knowing which persons to trust, staying sober, sharing information and helping other people who are homeless, and telling past success stories of their life when housed. On average, all of these activities are seen as important for the respondents in this study. In fact, all the behaviors show a significant relationship with the homeless identity meanings. This relationship is probably reciprocal so that the more important the behaviors are to the homeless individuals, the more likely they will be to discuss the homeless identity meanings in positive terms. Additionally, the more likely homeless people will be to discuss the homeless identity meanings in positive terms, the more important these behaviors are to them. For people who are homeless, their days and activities are not totally unstructured. Similarly to housed people, homeless persons still have routines to follow. For example, individuals sleeping at emergency shelters will wake around 5 a.m. to gather their personal belongings and get ready for the day. They will have breakfast before leaving the shelter. After that, some will head to appointments such as medical while others may go to the public libraries or to the parks to spend time during the day (Reitzes et al. 2011). Lines for evening beds tend to form outside the shelters in the early afternoon because the check in time is typically 5 p.m. 210 Upon entering, people check in, eat dinner and get ready for bed. This bed gives individuals a safe place to sleep for the night. As can be seen by the previous description of day’s activities, these patterns of behaviors are not only determined by the homeless individuals but are also shaped by the service providers. Factors that influence use of services include geographic location of the agencies and time when assistance is offered (Snow and Anderson 1993). For example, on Sundays there are two churches that serve sit down meals during the mid-morning. One is located downtown while the other is located in midtown. According to the 2007 local Tri-J homeless survey (Massey, Runkle and Parker 2007), the majority of the respondents (76 percent) learn about services from other homeless people. Besides sharing information on available services, homeless individuals also provide emotional support to each other (Massey, Runkle and Parker 2007). Trust is built among the homeless population through sharing information and providing support, from emotional support to safety. As part of the verbal construction of the self while homeless, people will discuss their past lives with other individuals who are homeless. These references to the past self are discussed in positive terms. Often these stories are focused around accomplishments of their children, sexual exploits and financial embellishments (Snow and Anderson 1993). Almost half (41 percent) of the study respondents indicate that heavy drinking and drug use are problems for them. According to Snow and Anderson (1993), use of drugs and alcohol increases with lengthening time spent on the streets. This substance use occurs for several reasons: street life subculture, out of boredom and to escape the hardships of life on the streets. The increasing importance of some of these behaviors (sharing information, helping others and telling stories of past successes) predict an increasing number of homeless friends, 211 strengthening attachment to others who are homeless and an increasing probability of invoking homelessness across different situations. Additionally, the more important getting drunk or high is for people, the more homeless friends they will have and the more they will discuss being homeless in different situations. Finally, the increasing importance of knowing which places serve the best free meals will result in more friends who are homeless. Just as the meanings for the homeless identity encompass several dimensions, so do the homeless identity behaviors. A couple of these behaviors are done to survive while homeless including finding a safe place to sleep and finding meal sites that serve decent food. Additional behaviors are related to finding support from others who are also in the situation: knowing which persons to trust, sharing information and helping other people who are homeless. Finally, other activities such as staying sober and telling stories about past success are done to maintain a self-concept. SELF-ESTEEM AND SELF-EFFICACY The dissertation finds that not all homeless people have low self-esteem. In fact, the majority of respondents (91 percent) demonstrate a self-esteem score in the high to middle range. Self-esteem seems to vary for people in the homeless situation depending on several factors. For people who are homeless, low self-esteem is the outcome of being homeless longer and more often, having little to no family support, possessing a high school diploma or less, sleeping on the streets, experiencing multiple disabilities, accepting the homeless identity and having that identity be the most important. This low self-esteem can affect the quality of daily experiences and overall could result in a lower satisfaction for life. Along the same lines as self-esteem, most respondents (89 percent) show a self-efficacy level in the high to middle range. As with self- 212 esteem, there are several variables that impact low self-efficacy: unaccompanied, homeless longer and more often, low to no family support, high school diploma or less, accepting the homeless identity, invoking the homeless identity frequently, having the homeless identity be most central, sleeping in unsheltered locations and possessing multiple disabilities. This finding therefore reflects the idea that self-esteem and self-efficacy are not stable, constant traits across situations but instead vary based on many factors. According to Rosenberg (1979), individuals with low self-esteem view themselves negatively, feeling that they are unworthy and inadequate. Not understanding their own value, people with low self-esteem are not satisfied with who they are and are not satisfied with their life. They tend to blame themselves for their poor performance and have difficulty with understanding the larger social structural implications regarding their plight. Similarly to self-esteem, people with low self-efficacy will view themselves as incapable. It is through the consequences of their actions, or lack of actions, that they develop a belief whereby they do not feel competent in accomplishing their goals. Often this is based on the lack of their access to resources, autonomy and control (Gecas and Schwalbe 1983). I believe that even if motivated to increase one’s level of self-esteem or self-efficacy, it is a difficult process, especially for individuals who are homeless. Most often increasing one’s self-esteem is accomplished by seeing a counselor and delving into one’s past, which is a lengthy process. Instead for homeless people, the focus needs to be on improving the factors that affect the outcomes of self-esteem and self-efficacy. Thus, service providers could focus on increasing self-esteem by getting people who sleep outside to stay in sheltered locations. Once staying at facilities it is important for staff to treat people with 213 dignity and respect. Individuals need to be treated based on their personal identities and not lumped together as one homogenous group (Miller and Keys 2001). One way to increase self-efficacy is to get people into drug or alcohol rehabilitation programs. Another avenue for improving self-efficacy is to encourage those without a high school diploma to get their GED. Of course, these are long range goals. Instead, the focus needs to be on creating short-term realistic goals that are accomplishable by being broken down into specific achievable tasks. For example, with obtaining the GED, short term goals will include completing homework assignments and passing the tests. As individuals accomplish and celebrate the short term goals, they begin to see themselves as casual agents. This encourages people to feel that they are capable and thus improve their feelings of self-efficacy. ATTEMPTS TO LEAVE THE HOMELESS SITUATION When people first become homeless, they are more than likely thinking that this will be a temporary situation. They hope to quickly get into housing again. Unfortunately, that is not always the case and individuals can be homeless for years. There are two main factors that contribute to an increased length of time homeless: sleeping on the streets and multiple disabilities. In fact, of all of the control variables, possessing multiple disabilities and sleeping in unsheltered locations appear to influence almost all of the dependent variables. Thus outcomes of people with multiple disabilities also include being homeless longer, more likely to accept the homeless identity, more likely to invoke the being homeless in different situations, more likely to think negatively about themselves, lower self-esteem and lower self-efficacy. In addition to these outcomes, individuals sleeping on the streets 214 will also have an increased number of friends who are homeless with the ties to those friendships being stronger. These findings indicate how crucial it is for people who are living on the streets to be placed into some type of housing. An interesting finding is that it does not matter what type of housing, whether emergency shelter or transitional housing. It just matters that people have a place to sleep indoors while homeless. Another important factor is for communities to address the needs of people who have disabling conditions such as mental health issues and substance abuse problems with specialized programs. Besides sleeping outdoors and having multiple disabilities, two other factors are found in the study to influence length of time homeless. The older people are and the longer that they have lived in Atlanta increases the length of time that individuals are in the situation and thus the homeless role. Therefore, service provider agencies need to be aware that older people may have a more difficult time getting housing. Anecdotally service providers have stated that an issue for the community is homeless people who have recently moved from out of town. That maybe so, but it appears that also of great concern are long term residents of the city who have become homeless. For homeless people to obtain housing, two factors are most important: income and support systems. Across all three hypotheses that investigate exiting homelessness, income had the most impact on people making attempts to find permanent housing such as looking at apartments or filling out applications and no longer be homeless. This makes sense in that money is needed to obtain and maintain housing. Another factor is the need for support from family and friends. Interestingly, having the encouragement of friends who are also homeless can be beneficial for leaving homelessness, 215 probably because this might be the only support that some homeless people have from others who are close to them. Additionally, the support of domiciled family members for their loved ones who are homeless is also crucial in obtaining housing. These relatives could also be instrumental in providing continual support after their homeless family members have become housed. According to Ebaugh (1988), to exit the homeless identity requires disengagement and disidentification from the role, followed by an adjustment, adaptation and reestablishment of an identity in a new role. Typically there are four stages to role exiting: first, doubts as to the ability to no longer be homeless, next is the seeking and weighting of role alternatives such as being domiciled, then there are the turning points when the final decision and act is made to find housing, and finally, what happens after exiting the homeless role in terms of establishing an ex-role identity. Of most concern are the properties of role exiting. One variable for exiting the role is the length of time that the role-exit process takes (Ebaugh 1988). For those who quickly find housing after becoming homeless, then this is not much of an issue. But as the duration of homelessness increases, then it can become harder for people to feel hope that they will successfully find permanent housing and no longer be homeless. Another important element of role exiting is the degree of control. The exiter does not operate in isolation but is instead often dependent on institutions and other people to facilitate the process (Ebaugh 1988). Unfortunately, there are often difficult barriers that people have to overcome in order to get assistance while homeless. For example, few transitional housing programs in the community take people who have a serious mental illness such as schizophrenia. 216 A third factor is the ability to return to a previous role (Ebaugh 1988). For example, during this current economic downtown, it becomes more difficult for many to find employment and return to the worker role. Trying to accomplish this goal of role exiting on their own is a fourth concern (Ebaugh 1988). For unaccompanied persons it is typically a solo process of no longer being homeless. A final element of role exiting for homeless people is the sequence (Ebaugh 1988). For individuals sleeping on the streets, it first starts with finding transitional housing. Then, it is obtaining employment, followed be finding transportation to get to and work clothes for the new job. A sufficient amount of money must be saved to get a permanent place to live. Once the housing is located, then it will take more effort to furnish it. Of course, this is only a sequence list of the big steps required to find housing and no longer be homeless. These goals do not include smaller tasks such as obtaining the birth certificate in order to get a current Georgia identification card. LIMITATIONS AND FUTURE RESEARCH One limitation of the dissertation is that it is a cross-sectional study, taking place only over a short period of time. A recommended next step for homeless identity research includes a longitudinal study which allows for process and change to be studied over an extended period of time. Additionally, a longitudinal study will generate a stronger prediction of the causal direction between the variables (Singleton and Straits 1999). Transitional housing programs let people stay up to two years as they move from homelessness into permanent housing. Therefore, persons could be surveyed within a short time that they become homeless and enter the program, in the middle of the program and just before 217 leaving the program so that the homeless identity can be understood over time. Once people move into permanent housing, the surveys could continue by investigating how people transition from a homeless negative identity into a housed positive identity. Along with the identity surveys, evaluation research could be conducted regarding the self-esteem and self-efficacy of homeless people. Again, this could occur in stages as people enter and exit the transitional housing program and upon moving into permanent housing. Would this future study support that self-esteem and self-efficacy are stable aspects of the self or would levels increase and decrease over time while in the situation? Another possibility for future identity research is to focus on homeless families. Whereas this study is predominately comprised of unaccompanied individuals, other research could concentrate on adults accompanied with their children. Primarily for this community, homeless families are headed by African American single mothers who have never been married. This family structure is typically associated with high rates of poverty. More than likely the female adult head of household would indicate that being a parent is their most important identity, especially if they are the sole provider. Where the difference could lie between this study and with research on homeless families would be with identity salience. Would the parents also indicate that the homeless identity is the most salient or would the parent identity be invoked more often across different situations such as getting the children ready for school, when asking for assistance at agencies and at dinner? Additionally, what is the commitment of these parents to the homeless identity? How does being homeless impact their self-esteem and self-efficacy? Beyond homeless identity research, I am also interested in investigating other negative identities such as obesity and addiction. These are both stigmatizing identities that a number of 218 people face in our society. They struggle not only with their physical health but also with constructing and maintaining a positive self-concept. I believe that the theories and concepts used in this study can also be applied to examining and understanding the issues related to possessing these negative identities as part of the self. SUMMARY At first people do not readily want to accept a negative or stigmatizing identity. Individuals will fight for the identity not to become part of their self-concept because the situation and corresponding identity at first will be seen as temporary. However, the longer that people are in the situation, the more likely they will be to take on the identity, embracing the role, associations and institutions attached. They will incorporate the identity into their self-concept in order for them to manage as the situation becomes more permanent. According to the findings of the study, a new negative identity will probably be invoked frequently in different situations even though it may not be a central identity nor will it be a strong commitment to the identity. Over time, it appears that the attachment to the identity will increase. However, the importance of the identity to the self-concept will not. This study finds that just because people are homeless does not mean that they think badly of themselves. In fact, they may be working harder than others to live life on the streets and thus think of themselves as survivors and not victims. To cope while homeless, their activities are important on several levels: getting their basic physical needs met, getting the emotional support that they need and maintaining their self-concept while homeless. Self-esteem is the evaluation of overall personal worth, while self-efficacy is the assessment of one’s competence level. Whereas most researchers focus on self-esteem as the 219 main self-worth issue facing people who are homeless, it does appear that both issues are of concern. For most homeless people, they are able to maintain a high to middle range level for both. However, it is important to address the factors such as possessing only a high school diploma or less that do cause a lowered self-esteem and self-efficacy level for people while homeless. Exiting the homeless role is a process with multiple factors. That’s why saying to homeless individuals that they just need to get a job doesn’t work or to just quit drinking or using drugs. When the stigmatized identity has become a part of the self-concept, then the process of no longer embracing that negative identity is difficult and requires many steps. Finally, for communities addressing the issue of homelessness, it is of utmost importance to tackle the two main factors, seeping outdoors and multiple disabilities, that affect almost all of the outcomes for this study: time homeless, accepting the homeless identity, salience and commitment to the homeless identity, homeless identity meanings, self-esteem and self-efficacy. Additionally, it is crucial for communities to understand that the most important factor for homeless people to make attempts to find permanent housing is income, no matter the other factors. Thus, to truly address significant issues of homelessness, communities need to focus their efforts on getting unaccompanied street people into some type of housing, providing treatment services for individuals with multiple disabilities and providing job programs so that people can earn an income. By communities addressing these major areas of concern, they can have an impact on improving the lives of people who are suffering while in this situation and on reducing the length and number of times that people are homeless. 220 REFERENCES Adams, Bert and R.A. Sydie. 2002. Contemporary Sociological Theory. Thousand Oaks, CA: Pine Forge Press. Agresti, Alan and Barbara Finlay. 1997. Statistical Methods for the Social Sciences, 3rd Edition. Upper Saddle River, NJ: Prentice Hall. Ashley, Johnson and Pandey. 2005. The 2005 Metro Atlanta Tri-Jurisdictional Collaborative Homeless Census and Survey: Homeless Survey Report, Covering DeKalb County and Fulton County, Including the City of Atlanta, Georgia. Andrew Young School of Policy Studies (Georgia State University) and Pathways Community Network (Atlanta, Georgia). Bogue, Donald. 1963. Skid Row in American Cities. Chicago, IL: Community and Family Study Center, University of Chicago. Boydell, Katherine, Paula Goering and Tammy Morrell-Bellai. 2000. “Narratives of Identity: Re-presentation of Self in People who are Homeless.” Qualitative Health Research 10: 26-38. Brault, Matthew. 2008. Americans with Disabilities: 2005, Household Economic Studies. U.S. Department of Commerce, Economics and Statistics Administration, U.S. Census Bureau. (www.census.gov) Brown, Cheryl. 1998. Family or Work: Identity, Salience Battles for Working Men and Women. Ph.D. dissertation, Department of Sociology, Georgia State University, Atlanta. Burke, Peter. 1980. “The Self: Measurement Implications from a Symbolic Interactionist Perspective.” Social Psychology Quarterly 43: 18-29. 221 Burke, Peter and Donald C. Reitzes. 1981.”The Link Between Identity and Role Performance.” Social Psychology Quarterly 44: 83-92. Burke, Peter and Jan Stets. 2009. Identity Theory. New York, NY: Oxford University Press. Burke, Peter and Judy Tully. 1977. “The Measurement of Role/Identity.” Social Forces 55: 881-897. Cast, Alicia and Peter Burke. 2002. “A Theory of Self-Esteem.” Social Forces 80: 1041-1068. Christian, Julie and Dominic Abrams. 2003. “The Effects of Social Identification, Norms and Attitudes on Use of Outreach Services by Homeless People.” Journal of Community & Applied Social Psychology 13: 138-157. Cohen, Alex. 2001. “The Search for Meaning: Eventfulness in the Lives of Homeless Mentally Ill Persons in the Skid Row District of Los Angeles.” Culture, Medicine, and Psychiatry 25: 277-296. Coldwell, Craig and William Bender. 2007. “The Effectiveness of Assertive Community Treatment for Homeless Populations with Severe Mental Illness: A Meta-Analysis.” American Journal of Psychiatry 164: 393-399. Copeland, Larry and Charisse Jones. 2005. “Atlanta Puts Heat on Panhandlers.” USA Today August 15, 2005 Croker, Jennifer and Brenda Major. 1989. “Social Stigma and Self-Esteem: The Self-Protective Properties of Stigma.” Psychological Review 96: 608-630. Czaja, Ronald and Johnny Blair. 1996. Designing Surveys: A Guide to Decisions and Procedures. Thousand Oaks, CA: Pine Forge Press. Diblasio, Frederick and John Belcher. 1993. “Social Work Outreach to Homeless People and the Need to Address Issues of Sel-Esteem.” Health and Social Work 18 222 Ebaugh, Helen. 1988. Becoming an Ex: The Process of Role Exit. Chicago, IL: University of Chicago Press. Elliott, Marta and Lauren Krivo. 1991. “Structural Determinants of Homelessness in the United States.” Social Problems 38: 113-131. Farrington, Alice and W. Peter Robinson. 1999. “Homelessness and Strategies of Identity Maintenance: A Participant Observation Study.” Journal of Community & Applied Social Psychology 9: 175-194. Freidel, Frank. 2001. The Presidents of the United States of America. Washington, D.C.: White House Historical Association. Gecas, Viktor and Michael Schwalbe. 1983. “Beyond the Looking-Glass Self: Social Structure and Efficacy-Based Self Esteem.” Social Psychology Quarterly 46: 77-88. Gerstel, Naomi, Cynthia Bogard, J. Jeff McConnell and Michael Schwartz. 1996. “The Therapeutic Incarceration of Homeless Families.” Social Service Review December: 543-572. Goering, Paula, George Tolomiczenko, Tess Sheldon, Katherine Boydell and Donald Wasylenski. 2002. “Characteristics of Persons who are Homeless for the First Time.” Psychiatric Services 53: 1472-1474. Goffman, Erving. 1963. Stigma: Notes on the Management of Spoiled Identity. New York, NY: Simon & Schuster, Inc. Goldfinger, Stephen, Russell Schutt, George Tolomiczenko, Larry Seidman, Walter Penk, Winston Turner and Brina Caplan. 1999. “Housing Placement and Subsequent Days Homeless among Formerly Homeless Adults with Mental Illness.” Psychiatric Services 50: 674-679. 223 Hill, Ronald Paul. 2003. “Homelessness in the US: An Ethnographic Look at Consumption Strategies.” Journal of Community & Applied Social Psychology 13: 128-137. Hopper, K. and J. Hamburg. 1984. The Making of America’s Homeless: From Skid Row to New Poor, 1945-1984. New York, NY: Community Service Society. Hoyt, Danny R. and Nicholas Babchuk. 1983. “Adult Kinship Networks: The Selective Formation of Intimate Ties with Kin.” Social Forces 62: 84-101. James, William. 1890. Principles of Psychology. New York, NY: Holt. Johnson, Janet, Shena Ashley and Lakshmi Pandley. 2005. The 2005 Metro Atlanta Tri-Jurisdictional Collaborative Homeless Census and Survey: Homeless Survey Report, Covering DeKalb County and Fulton County, including the City of Atlanta, Georgia. Atlanta, GA: Pathways Community Network. (http://data.memberclicks. com/site/pat/2005_Tri-J_Homeless_Survey%20-%20final.pdf). Koegel, Paul, Elan Melamid and Audrey Burnam. 1995. “Childhood Risk Factors for Homelessness among Homeless Adults.” American Journal of Public Health 85: 1642-1649. Knoke, David, George Bohrnstedt and Alisa Potter Mee. 2002. Statistics for Social Data Analysis, fourth edition. Belmont, CA: Wadsworth/Thomas Learning. MacKnee, Chuck and Jennifer Mervyn. 2002. “Critical Incidents That Facilitate Homeless People’s Transition Off the Streets.” Journal of Social Distress and the Homeless 11: 293-306. Main, Thomas. 1998. “How to Think About Homelessness: Balancing Structural and Individual Causes.” Journal of Social Distress and the Homeless 7: 41-54. 224 MacKnee, Chuck and Jennifer Mervyn. 2002. “Critical Incidents That Facilitate Homeless People’s Transition Off the Streets.” Journal of Social Distress and the Homeless 11: 293-306. Maslow, Abraham. 1970. Motivation and Personality. New York, NY: Harper & Row. Massey, Jane, Elizabeth Runkle and Josie Parker. 2007. Experiencing Homelessness: Highlights from the 2007 Metro Atlanta Tri-Jurisdiction Survey. Atlanta, GA: Pathways Community Network. Mead, George. 1934. Mind, Self & Society. Chicago, IL: The University of Chicago Press. Miller, Alison and Christopher Keys. 2001. “Understanding Dignity in the Lives of Homeless Persons.” American Journal of Community Psychology 29: 331-354. Morrell-Bellai, Tammy, Paula Goering and Katherine Boydell. 2000. “Becoming and Remaining Homeless: A Qualitative Investigation.” Issues in Mental Health Nursing 21: 581-604. Morris, Joan. 1998. “Affiliation, Gender, and Parental Status among Homeless Persons.” Journal of Social Psychology 138: 241-250. National Institute of Mental Health. 2010. National Institute of Mental Health (NIMH), Transforming the Understanding and Treatment of Mental Illness Through Research. July 19, 2010. (http://www.nimh.nih.gov/statistics/index.shtml) Osborne, Randall. 2001. “ ‘I May Be Homeless, But I’m not Helpless’: The Costs and Benefits of Identifying with Homelessness.” Self and Identity 1: 43-52. Osgood, Charles. Focus on Meaning. Hague, Netherlands: Mouton. Parker, Josie. 2009. Experiencing Homelessness: 2009 Metro Atlanta Tri-Jurisdictional Collaborative Homeless Survey. Atlanta, GA: Pathways Community Network. 225 Parker, Josie, Nykia Greene-Young, Natasha Davenport and Autumn Austin. 2011. The 2011 Metro Atlanta Tri-Jurisdictional Collaborative Continuum of Care Homeless Census: City of Atlanta, Fulton County and DeKalb County. Atlanta, GA: Pathways Community Network. Reitzes, Donald C., Timothy J. Crimmins, Johanna Yarbrough and Josie Parker. 2011. “Social Support and Social Network Ties Among the Homeless in a Downtown Atlanta park.” Journal of Community Psychology 39: 274-291. Reitzes, Donald C. and Charles Jaret. 2009. "Currents in a Stream: College Student Identities and Ethnic Identities and Their Relationship with Self-Esteem, Efficacy, and Grade Point Average in an Urban University." Social Science Quarterly 90: 345-367. Rosenberg, Morris. 1979. Conceiving the Self. New York, NY: Basic Books. Rosenheck, Robert. 2000. “Cost-Effectiveness of Services for Mentally Ill Homeless People: The Application of Research to Policy and Practice.” American Journal of Psychiatry 157: 1563-1570. Rossi, Peter. 1989. Down and Out in America. Chicago, IL: University of Chicago Press. Rossi, Peter. 1990. “The Old Homeless and the New Homelessness in Historical Perspective.” American Psychologist 45: 954-959. Runkle, Elizabeth and Josie Parker. 2007. The 2007 Metro Atlanta Tri-Jurisdictional Collaborative Homeless Census Report, Covering the City of Atlanta, Fulton County and DeKalb County, Georgia. Atlanta, GA: Pathways Community Network. Ry, Meredith van. 1993. Homeless Families: Causes, Effects and Recommendations. New York, NY: Garland Publishing. 226 Singleton, Jr., Royce and Bruce Straits. 1999. Approaches to Social Research, Third Edition. New York, NY: Oxford University Press. Serpe, Richard. 1987. “Stability and Change in Self: A Structural Symbolic Interactionist Explanation.” Social Psychology Quarterly 50:44-55. Snow, David and Leon Anderson. 1987. “Identity Work among the Homeless: The Verbal Construction and Avowal of Personal Identities.” American Journal of Sociology 92: 1336-1371. Snow, David and Leon Anderson. 1993. Down on Their Luck. Berkeley, Ca: University of California Press. Snow, David, Leon Anderson, and Paul Koegel. 1994. “Distorting Tendencies in Research on the Homeless.” American Behavioral Scientist 37: 461-475. Snow, David and Michael Mulcahy. 2001. “Space, Politics, and the Survival Strategies of the Homeless.” American Behavioral Scientist 45: 149-169. Stets, Jan and Peter Burke. 2000. “Identity Theory and Social Identity Theory.” Social Psychology Quarterly 63: 224-238. Stryker, Sheldon. 1989. “Further Developments in Identity Theory: Singularity versus Multiplicity of Self.” Pp. 35-57 in Sociological Theories in Progress: New Formulations, edited by Joseph Berger, Morris Zeilditch Jr., and Bo Anderson. Newbury Park, CA: Sage. Stryker, Sheldon and Richard Serpe. 1994. “Identity Salience and Psychological Centrality: Equivalent, Overlapping, or Complementary Concepts?” Social Psychology Quarterly 57:16-35. Stryker, Sheldon and Peter Burke. 2000. “The Past, Present, and Future of Identity Theory.” Social Psychology Quarterly 63: 284-298. 227 Sudman, Seymour. 1976. Applied Sampling. New York, NY: Academic Press. Tajfel, Henri and John Turner. 1986. “The Social Identity Theory of Intergroup Behavior.” Pp. 7-24 in Psychology of Intergroup Relations, edited by Stephen Worchel and William Austin. Chicago, IL: Nelson-Hall Publishers. Tam, Tammy, Cheryl Zlotnick and Marjorie Robertson. 2003. “Longitudinal Perspective: Adverse Childhood Events, Substance Use, and Labor Force Participation among Homeless Adults.” The American Journal of Drug and Alcohol Abuse 29: 829-846. Thoits, Peggy. 1992. “Identity Structures and Psychological Well-being: Gender and Marital Status Comparisons.” Social Psychology Quarterly 55: 236-256. Thorton, Russell and Peter Nardi. 1975. “The Dynamics of Role Acquisition.” American Journal of Sociology 80: 870-885. Turner, Ralph. 1962. “Role-Taking: Process Versus Conformity.” Pp. 20-39 in Human Behavior and Social Processes: An Interactionist Approach, edited by Arnold Rose. New York, NY: Routledge. U.S. Census Bureau. 2005-2009. American Community Survey (ACS) 5-Year Estimates. American Fact Finder. Washington, D.C. (www.factfinder.census.gov) U.S. Census Bureau. 2010. Age Groups and Sex. 2010 Census Summary File 1. American Fact Finder. Washington, D.C. (www.factfinder2.census.gov) U.S. Department of Education. Subtitle B of title VII of the McKinney-Vento Homeless Education Assistance Improvement Act of 2001 (42 U.S.C. 1143 et seq.). U.S. Department of Health and Human Services. 2003. Secretary’s Work Group on Ending Chronic Homelessness. “Ending Chronic Homelessness: Strategies for Action.” (http://aspe.hhs.gov/hsp/homelessness/strategies03/ch.htm). 228 U.S. Department of Housing and Urban Development. 2007. “Defining Chronic Homelessness: A Technical Guide for HUD Programs.” Office of Community Planning and Development. Office of Special Needs Assistance Programs. (http://www.hudhre.info/ documents/DefiningChronicHomeless.pdf) U.S. Department of Health and Human Services. 2010. The HHS poverty guidelines for the remainder of 2010 (August 2010). Washington, D.C. (www.aspe.hhs.gov) U.S. Department of Housing and Urban Development. McKinney-Vento Homeless Assistance Act of 1986 (P.L. 100-77, sec 103(2)(1), 101 sat. 485 (1987). As amended by S. 896 The Homeless Emergency Assistance and Rapid Transition to Housing (HEARTH) Act of 2009. U.S. Department of Housing and Urban Development. Office of Community Planning and Development. 2009. “The 2008 Annual Homeless Assessment Report to Congress: A Summary of Findings.” (http://www.hudhre.info/documents/4th HomelessAssessmentReport.pdf). Vannoy-Hiller, Dana and William W. Philliber. 1989. Equal Partners. Newbury Park, Ca: Sage. Vogt, W. Paul. 1999. Dictionary of Statistics & Methodology: A Nontechnical Guide for the Social Sciences, Second Edition. Thousands Oak, CA: Sage Publication. Weinreb, Linda and Peter Rossi. 1995. “The American Homeless Family Shelter ‘System’.” Social Service Review March: 86-107. Wright, James, Beth Rubin, and Joel Devine. 1998. Beside the Golden Door: Policy, Politics, and the Homeless. New York, NY: Aldine De Gruyter. Zimet, G.D., N.W. Dahlem, S.G. Zimet and G.K. Farley. 1988. “The Multidimensional Scale of Perceived Social Support.” Journal of Personality Assessment 52: 30-41. 229 APPENDICES CONSENT FORM TITLE City, Self and Identity: Processes and Outcomes PRINCIPAL INVESTIGATOR Dr. Donald C. Reitzes STUDENT INVESTIGATOR Josie Parker PURPOSE You are invited to take part in a research study. The goal of the study is to look into the maintenance and changes of roles and identities among people who use shelters, meal sites and other outreach services. You are being asked to take part because of your experience with these services. The survey will take 20-30 minutes. About 400 people will be asked to take part in the study. PROCEDURES If you decide to take part, you will fill out a nine page survey. If you have any questions or problems while filling out the survey, the student researcher, Josie Parker, or an assistant will be on hand to help. A cracker snack or breakfast bar is being given as a gift for your time and effort in filling out the survey. The survey is being carried out at meal sites, service agencies and shelters from June to September, 2008. RISKS AND BENEFITS In this study, you will not have any more risks than you would in a normal day of life. Taking part in this study may not help you personally. In general, I do hope to gain a better grasp of how roles and identities are used to survive the struggles of daily life. Your input may lead to information that could help understand what it means to be in this situation and ways to address it. VOLUNTARY PARTICIPATION AND WITHDRAWAL Taking part in this study is your choice. You do not have to be in this study. If you decide to be in the study and change your mind, you have the right to stop at any time. You may skip any question while filling out the survey. You will be given a gift for taking part in the study even if you skip questions or stop in the middle of the study. CONFIDENTIALITY All information gathered from this study will remain private. Your name or information will not be given to anyone. A survey number rather than your name will be kept on survey records. The surveys will be kept in a storage unit in a locked room. Only the student researcher will have direct access to the information you provide but there may be times when Josie Parker will review the data with her advisor, Donald C. Reitzes. Your initials and other facts that might point 230 to you will not appear when this study is presented or the results published. The results will be summarized and reported in group form. You will not be identified personally. CONTACT PERSONS If you have any questions about this study, you may contact Josie Parker at (770) 851-9032 or Dr. Donald C. Reitzes at (404) 413-6506. If you have questions or concerns about your rights as a participant in this study, you may contact Susan Vogtner in the Office of Research Integrity, Georgia State University, at (404) 413-3513. COPY OF CONSENT FORM We will give you a copy of this consent form to keep. You will indicate your consent to volunteer for this research by completing a survey. 231 SURVEY PART I 1. What is your race or ethnicity? (Check only one box.) Black or African American Asian or Pacific Islander White or Caucasian (non-Hispanic) Other single race: ___________________________ Hispanic or Latino/a Other multi-racial: __________________________ 2. What is your gender? (Check only one box.) 3. What is your relationship status? (Check only one box.) Single and/or dating Separated or divorced In a committed relationship but not married Widowed Married Other: __________________ 4. Have you ever served in the military? (Check only one box.) Yes No 5. Are you a parent? (Check only one box.) Yes No 6. Are you currently living with any family members such as a spouse, child(ren), parent, sibling, etc.? (Check only one box.) Yes No Female Male Transgender 232 7. What is the highest level of education you have completed? (Check only one box.) 11th grade or less (no high school diploma) High school diploma or GED Some college (no degree) Technical or vocational degree 2 year or 4 year college degree (AA / AS / BA/ BS degree) Some graduate school (no degree) MA, MS, Ph.D. or professional degree (MD, JD, etc.) 8. What are your initials? (Write answer on lines.) First letter of first name: ______ First letter of last name: _____________ 9. What is your date of birth? (Write answer on lines.) Month: __________ Day: ____________ Year: ______________ 10. How long have you lived in Atlanta? (Check only one box.) Less than 1 month 5 – 9 years 1 – 6 months 10 – 15 years 7 months – 1 year More than 15 years 1 year – 4 years Do not live in Atlanta 11. How much money did you make last month? (Check only one box.) $0 $751 - $1000 $1 - $250 $1001 - $1500 $251 - $500 Over $1501 $501 - $750 233 PART II 1. The following list provides several identities and groups that are important to a person. Please choose the 5 identities or groups that you believe are most important to you and rank them below. For example, if you believe that being a parent is most important to you, then you would write 1 on the line beside parent. If being a friend is second most important to you, then write 2 on the line beside friend. If there is an identity or group not listed that fits better, please write it on the other line. (Write the number of importance on the line in front of the identity.) _____ Being a family member (parent / son /daughter / brother / sister, etc.) _____ Relationship status (married, Dating, Single, etc.) _____ Being a friend _____ Being part of a racial or ethnic group _____ Age group _____ Gender (male / female) _____ Sexual orientation _____ Being homeless _____ Educational level _____ Job / occupation / worker _____ Being a student _____ Being a veteran _____ Being an alcoholic / addict _____ Being mentally healthy or ill _____ Being physically healthy or disabled _____ Religion or religious affiliation _____ Other identity (Please describe): ____________________________________ 234 The following questions ask about certain situations where you would encounter other people. From the list below, please choose what you would talk about most with the other people in that particular setting. For example, when at a job, you might talk to other people about your job most frequently because you are at work and about your children second most frequently because you are a proud parent. If there is a subject not listed that fits better, please write it on the frequently discuss line. If no subject fits, write N/A. • Family members or issues • Homeless issues • Relationship issues • Job / employment / worker issues • Friendship (other friends) • Student issues • Race or ethnic issues • Veteran issues • Issues of being a certain age • Issues of being an alcoholic / addict • Gender (masculine / feminine) • Mental state • Sexual orientation • Physical health or disability • Educational issues • Weight issues • Sports • Religion or religious issues 2. When hanging out with friends, what do you talk about most often? (Please choose from the list above.) Most frequently discuss: ______________________________________________ Second most frequently discuss: ________________________________________ 3. When spending time with family, what do you talk about most often? (Please choose from the list above.) Most frequently discuss: _______________________________________________ Second most frequently discuss: ________________________________________ 4. When at a service provider agency, what do you talk about most often? (Please choose from the list above.) Most frequently discuss: ______________________________________________ Second most frequently discuss: ________________________________________ 5. When at the doctor’s office or hospital, what do you talk about most often? (Please choose from the list above.) Most frequently discuss: _______________________________________________ Second most frequently discuss: _________________________________________ 6. When at a shelter, what do you talk about most often? (Please choose from the list above.) Most frequently discuss: ______________________________________________ Second most frequently discuss: _______________________________________ 235 PART III For the following statements, decide if you strongly agree, agree, disagree or strongly disagree. (Check only one box for each statement.) Strongly Agree Agree Disagree Strongly Disagree 1. I feel that I’m a person of worth, at least on an equal plane with others. 2. I feel that I have a number of good qualities. 3. All in all, I am inclined to feel that I am a failure. 4. I am able to do things as well as most other people. 5. I feel I do not have much to be proud of. 6. I take a positive attitude toward myself. 7. On the whole, I am satisfied with myself. 8. I wish I could have more respect for myself. 9. I certainly feel useless at times. 10. At times I think I am no good at all. 11. I can pretty well control things that happen to me. 12. I often feel helpless in dealing with the problems of life. 13. I can do just about anything I really set me mind to. 14. Sometimes I feel that I’m being pushed around. 15. When trying to learn something new, I soon give up if I am initially successful. 16. What happens to me in the future mostly depends on me. 236 PART IV 1. Where have you usually slept at night in the past month? (Check only one box.) Emergency shelter Anywhere I can (outside, emergency shelter, etc.) Transitional (longer term) shelter House or apartment of friend or family member Treatment program Hotel or motel room Outdoors or on the street Own apartment, room or house Abandoned Building Other (Describe): In a car or other vehicle _________________________ 2. Do you identify as a homeless person? (Check only one box.) Strongly identify Not sure if identify Identify Do not identify 3. What are the reasons for you becoming homeless? (Check all that apply.) Unemployed ( no job) due to being fired or quitting Unemployed (no job) due to being laid off No available jobs for skill or educational level Not making enough money (on job or with SSI/benefits) to pay bills Unable to pay rent or mortgage Government cutbacks (SSI, welfare benefits, disability, etc.) Family violence Physical illness or medical problem Divorce or separation Mental illness Death in family Physically disabled Can’t stay with family or friends Relocation from out of town Lost public housing or Section 8 Alcohol or drug use Lost housing due to non-economic reasons (ex. evicted for lease violations) Hurricane Katrina No available public housing or Section 8 Personal choice Can’t find affordable housing Other (Describe): ___________________ Bad luck Don’t know / no answer 237 4. How many different times have you been homeless during the last 3 years, including this time? (Check only one box.) 1 time 4 times 2 times 5 times or more 3 times Don’t know / no answer 5. For this current time of homelessness, how long have you been homeless since your last permanent housing? (Check only one box.) Less than 1 month 2 years 1 - 3 months 3 years 4 - 6 months 4 years 7 - 11 months 5 years or more 1 year Don’t know / no answer 6. Over the past three years, what has been the total length of time that you have been homeless? (Check only one box.) Less than 1 month 1 year 1 - 3 months 2 years 4 - 6 months 3 years 7 - 11 months Don’t know / no answer 7. Please read the list below and decide if any of the items are currently an issue for you. (Check all that apply.) Alcohol abuse Mental illness Drug abuse Depression Physical disability Anxiety Chronic health problem None of these are a problem HIV / AIDS 238 PART V For the following statements, decide if you strongly agree, agree, disagree or strongly disagree. (Check only one box for each statement.) 11. How many of your friends are homeless? (Check only one box.) 0 6-10 1 - 2 11 -15 3 - 5 16 or more Strongly Agree Agree Disagree Strongly Disagree 1. I don’t feel connected to others who are homeless. 2. My homeless friends are a real source of comfort to me. 3. I enjoy and value the social ties and contacts that I’ve made as a homeless person. 4. I haven’t made good friends with others who are homeless. 5. I’m happy when I’m with friends who are also homeless. 6. I get the emotional help and support I need from my friends who are homeless. 7. I count on my homeless friends when I get sick. 8. I can’t depend on my homeless friends when things go wrong. 9. Most other homeless people don’t treat me well. 10. People who are homeless understand me better than most other people do. 239 PART VI Please read each set of descriptions. For each set, decide which word best describes you. (Check only one box for each set of words.) 1-7. As a homeless person, I am Think about yourself as a person who is homeless. How important are the following activities to you? (Check only one box for each activity.) Very honest Honest Neutral Dishonest Very dishonest Very friendly Friendly Neutral Not friendly Very unfriendly Very kind Kind Neutral Mean Very mean Very capable Capable Neutral Incapable Very incapable Very independent Independent Neutral Dependent Very dependent Very hardworking Hard-working Neutral Lazy Very lazy Very motivated Motivated Neutral Not motivated Very unmotivated Very Important Important Not so important Not at all important 8. Finding a safe place to sleep 9. Knowing which person to trust 10. Knowing which meal sites serve the best tasting food 11. Helping other homeless people 12. Telling stories about past successes 13. Getting drunk or high 14. Sharing information with other homeless people 240 PART VII 1. What do you need to do to no longer be homeless? (Check all that apply.) Job that pays a good livable wage Job that consistently is 40 hours a week Job that pays better than present job Maintain job Find stable, affordable housing Get GA identification or Driver’s license Section 8 or public housing Work on medical needs NoGo back to school for further education (degree) or GED Assistance from God or Higher Power Drug or alcohol treatment Counseling or case management Transportation assistance Mental health treatment GSave money Get SSI benefits Legal assistance Run of good luck Change in attitude Other: ________________________ 2. In the past month, how many attempts (i.e. filling out an application, viewing an apartment or house, etc.) have you made to get permanent housing such as an apartment or house? (Check only one box.) 0 2 4 1 3 5 or more 3. Would you feel guilty if you were able to move into a home or apartment and your homeless friends couldn’t? (Check only one box.) Yes No I have no homeless friends 241 For the following statements regarding your family, decide if you strongly agree, agree, disagree or strongly disagree. (Check only one box for each statement.) PART VIII 1. Is there anything else that you would like to share about your experiences? (Write answer below.) Thank you very much for taking your valuable time and effort to complete this survey. Have a great day. Strongly Agree Agree Disagree Strongly Disagree 4. I don’t get the emotional help and support I need from my family. 5. I count on my family when I get sick. 6. I count on my family when things go wrong. 7. My family is not a real source of comfort. 8. My family helps out when I am in financial need.

Setting

Abstract

Extracted data

Setting

Abstract

Extracted data

Related items

Related items