HOSTILITY: INDIVIDUAL DIFFERENCES IN COGNITIVE PROCESSES by Weronika E. Sroczynski Hon. B.Sc, University of Toronto, 2006 THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN PSYCHOLOGY THE UNIVERSITY OF NORTHERN BRITISH COLUMBIA December 2010 © Weronika E. Sroczynski, 2010 1*1 Library and Archives Canada Published Heritage Branch 395 Wellington Street Ottawa ON K1A 0N4 Canada Bibliotheque et Archives Canada Direction du Patrimoine de I'edition 395, rue Wellington Ottawa ON K1A 0N4 Canada Your file Votre reference ISBN: 978-0-494-75118-3 Our file Notre reference ISBN: 978-0-494-75118-3 NOTICE: AVIS: The author has granted a non- exclusive license allowing Library and Archives Canada to reproduce, publish, archive, preserve, conserve, communicate to the public by telecommunication or on the Internet, loan, distribute and sell theses worldwide, for commercial or non- commercial purposes, in microform, paper, electronic and/or any other formats. L'auteur a accorde une licence non exclusive permettant a la Bibliotheque et Archives Canada de reproduire, publier, archiver, sauvegarder, conserver, transmettre au public par telecommunication ou par I'lnternet, preter, distribuer et vendre des theses partout dans le monde, a des fins commerciales ou autres, sur support microforme, papier, electronique et/ou autres formats. The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. L'auteur conserve la propriete du droit d'auteur et des droits moraux qui protege cette these. Ni la these ni des extraits substantiels de celle-ci ne doivent etre im primes ou autrement reproduits sans son autorisation. In compliance with the Canadian Privacy Act some supporting forms may have been removed from this thesis. Conformement a la loi canadienne sur la protection de la vie privee, quelques formulaires secondaires ont ete enleves de cette these. While these forms may be included in the document page count, their removal does not represent any loss of content from the thesis. Bien que ces formulaires aient inclus dans la pagination, il n'y aura aucun contenu manquant. • • I Canada ii Abstract High hostile individuals appear to interpret social information as more hostile than it really is, which is thought to be deleterious to their health. The present investigation sought to determine whether activating hostile thought processes affects impression formation and behavior differently in high versus low hostile individuals. Participants were classified as high hostile or low hostile based on their Cook-Medley Hostility Scale scores. Participants were randomly assigned to either a hostile or a neutral semantic priming condition. Low hostile women formed more negative impressions of a hypothetical person when primed with hostile words than neutral ones. No differences were observed for high hostile individuals. Hostile priming also caused low hostile women to exhibit more hostile behaviors than high hostile women. These findings suggest that high and low hostile individuals process hostile information differently and that high hostile individuals do not merely possess a generalized hostile-other bias. ill TABLE OF CONTENTS Abstract ii Table of Contents iii List of Tables vi List of Figures vii List of Appendices viii Acknowledgements ix Introduction 1 Defining Hostility 2 Hostility and Heart Disease 3 Mechanisms by which hostility confers risk 4 The Moderating Role of Sex 5 Social Cognitive Dimensions 6 Social cognitive mechanisms of hostility 7 Individual differences in cognitive processing 8 Priming: Activating knowledge structures 9 Influencing perception and behavior 10 Direction of priming effects 11 The Present Study 12 Word Rating Study 14 Method 14 Participants 14 Apparatus and Materials 15 IV Word selection 15 Rating booklet 15 Procedure 16 Analysis and Results 16 Priming Study 18 Method 20 Participants 20 Apparatus and Materials 21 Cook-Medley Hostility Scale 21 Computer and software 21 Stimulus words 21 Vignettes 21 Anger interview 22 Video camera 22 Procedure 22 Priming task 23 Impression formation task 23 Recall memory test 24 Recognition memory test 24 Anger interview 24 Design and Data Reduction 25 Analysis and Results 26 Impression formation 27 Recall memory Recognition memory 31 Reaction time 32 Discussion 33 Impression formation 33 Recall memory 36 Recognition memory 36 Reaction time 36 Video Rating Study 36 Method 37 Participants 37 Apparatus and Materials 37 Videos 37 Rating booklets 38 Procedure 38 Design and Data Reduction 38 Analysis and Results 39 Discussion 41 General Discussion 42 Limitations 47 Future Directions 48 References 50 Appendices 56 List of Tables Table 1. Factor loadings, communalities (h ), and percent of variance for principal factors extraction and direct oblimin rotation on the 12 trait descriptors. Table 2. Reliabilities (intraclass correlations) of decoders' ratings of encoders' behavioral dimensions. vii List of Figures Figure 1. Mean ratings on the Nice factor as a function of hostility status for men and 30 women. Figure 2. Women's mean ratings on the Nasty factor as a function of hostility status 31 for two priming conditions. Figure 3. Men's mean ratings on the Nasty factor as a function of hostility status for 32 two priming conditions. Figure 4. Observed hostility as a function of priming condition for women. 40 Figure 5. Observed hostility as a function of priming condition for men. 41 V l l l List of Appendices Appendix A List of hostile words for the word rating study 56 Appendix B List of neutral words for the word rating study 57 Appendix C Hostility rating scale 58 Appendix D Valence rating scale 59 Appendix E Imageability rating scale 60 Appendix F Information sheet, word rating study 61 Appendix G Consent form, word rating study 63 Appendix H Debriefing handout, word rating study 64 Appendix I List of hostile primes and test words for the priming task 65 Appendix J List of neutral primes and test words for the priming task 66 Appendix K Hostile paragraph for the impression formation task 67 Appendix L Anger interview protocol 68 Appendix M Information sheet, priming study 69 Appendix N Consent form, priming study 71 Appendix O Debriefing handout, priming study 72 Appendix P Video rating scale for the video rating study 73 Appendix Q Information sheet, video rating study 74 Appendix R Consent form, video rating study 75 ix Acknowledgements I would like to thank my supervisor, Dr. Ken Prkachin. This thesis would not have been possible without his enthusiasm, insight, and good humor. As well, I thank my committee members, Dr. Glenda Prkachin, Dr. Glen Schmidt, and Dr. Paul Siakaluk, for their feedback and guidance. 1 would also like to thank the Office of Graduate Programs at UNBC and the Social Sciences and Humanities Research Council of Canada (SSHRC) for awarding me with a Spring- Summer Research Project Award in May 2008. This funding allowed me to complete the word rating study. 1 Researchers have long suspected that psychological factors are implicated in the development of coronary heart disease (Allred & Smith, 1991). This suspicion has inspired a vast amount of research attempting to elucidate the behavioral antecedents of heart disease. Over the past 30 years hostility has emerged as a predictor of deleterious cardiovascular health outcomes. Specifically, hostility has been implicated in the severity of coronary atherosclerosis (Byers, 1991; MacDougall, Dembroski, Dimsdale, & Hackett, 1985; Matthews, Owens, Edmundowicz, Lee, & Kuller, 2006; Williams, Haney, Lee, Kong, Blumenthal, & Whalen, 1980) as well as heart disease incidence (Gonnie, Hans, Van Lenthe, Kempen, Van Eijk, & Mackenbach, 2009; Matthews, Glass, Rosenman, & Bortner, 1977) and mortality (Barefoot, Dahlstrom, & Williams, 1983; Shekelle, Gale, Ostfeld, & Paul, 1983). Despite its association with poor cardiovascular health, exactly how hostility may confer risk of heart disease has remained elusive. Various mechanisms have been proposed. For instance, according to the psychophysiological model a general reactivity mechanism is responsible for the connection between hostility and heart disease (Smith & Christensen, 1992). Specifically, this model presumes that hostile individuals chronically experience heightened cardiovascular reactivity (e.g., elevated blood pressure) and that this general reactivity is what predisposes them to develop heart disease. This model appears viable but has not been consistently supported in the literature. Alternatively, Smith's (1992) transactional model proposes that psychological and social factors mediate the relationship between hostility and heart disease. Smith's model presumes that hostile individuals possess a "unique cognitive style" (p. 401) that creates stressful social interactions which, in turn, elicit greater physiological reactivity (e.g., elevated blood pressure) that contributes to illness (Allred & Smith, 1991). If hostile cognitions and behaviors mediate the relationship between physiology and illness, then a 2 clear theoretical understanding of how hostile thought processes and behaviors function is necessary before interventions aimed at reducing the psychological factors that exacerbate the disease process can be implemented successfully. Accordingly, the purpose of the present study was to explore individual differences in the cognitive processes mediating hostile behavior. Specifically, the present study addressed whether activating hostile thought processes would affect impression formation or behavior differently in high versus low hostile individuals. Defining Hostility Since the emergence of the study of hostility in the context of coronary heart disease research, the construct has been defined using different theoretical notions. Consequently, the term "hostility" is not a uniform psychological construct throughout the literature. Related constructs such as anger, cynicism, and aggression are frequently used interchangeably with hostility (Barefoot, 1992; Megargee, 1985; Spielberger et al., 1985). This becomes problematic when studies of hostility are incorrectly assumed to refer to and measure the same construct. In order to avoid confusion and miscommunication, Barefoot (1992) proposed using a comprehensive definition in which hostility is viewed as a multifaceted construct with cognitive, affective, and behavioral components. The cognitive component of hostility consists of negative attitudes or beliefs about others and the world (Barefoot, 1992). This includes cynicism, a set of negative attitudes or beliefs about human nature in general, as well as hostile attributions, beliefs that the self is the target of the antagonistic behavior of others (Barefoot, Dodge, Peterson, Dahlstrom, & Williams, 1989). The affective component of hostility is comprised of emotional states such as anger, annoyance, disgust, contempt, resentment, envy, and jealousy (Barefoot, 1992; Smith, 1992). Lastly, the behavioral component of hostility includes acts such as verbal and physical aggression (Barefoot, 1992) as well as facial expressions such as fewer non- 3 Duchenne smiles (Prkachin & Silverman, 2002). The definition of hostility used in this study is consistent with Barefoot's framework. Hostility and Heart Disease Coronary heart disease is the leading cause of death in Canada. In 2005, heart disease accounted for just over 30% of all deaths in both men and women (Statistics Canada, 2009). Heart disease is also the most costly disease in Canada; heart disease was responsible for about $22 billion in expenditures in Canada in 2005 (Heart and Stroke Foundation of Canada, 2010). Not only does heart disease negatively affect our health care system through its associated costs, it has substantial long-term mental, physical, emotional, and social consequences for those it affects. Consequently, a considerable amount of research has been devoted to investigating the processes and risk factors that contribute to the development of heart disease. Cardiologists Friedman and Rosenman (1959, cited in Smith & Leon, 1992) reportedly noticed that some heart patients were more likely than non-patients to exhibit a specific pattern of behavior characterized by competitive achievement orientation, time urgency, and easily- aroused anger or hostility. This pattern became referred to as Type A behavior. Though research on Type A behavior flourished for some time, null findings cast a shadow of doubt over this construct as the key psychosocial determinant of heart disease. Consequently, attempts at isolating the exact components of Type A that lead to heart disease ensued. For instance, Dembroski, MacDougall, Williams, Haney, and Blumenthal (1985) demonstrated through component scoring of the Type A Structured Interview, the gold standard of techniques for measuring Type A behavior, that there was no relationship between global Type A and disease severity, but that hostility was significantly related to coronary atherosclerosis severity. Such findings prompted researchers to contend that perhaps hostility was the "toxic" component of the 4 Type A construct that is responsible for its association with heart disease (Siegman & Dembroski, 1989). Further studies have revealed that hostility predicts severity of coronary atherosclerosis (Byers, 1991; MacDougall et al., 1985; Matthews et al., 2006; Williams et al., 1980) as well as heart disease incidence (Gonnie et al., 2009; Matthews et a l , 1977) and mortality (Barefoot et al., 1983; Shekelle et al., 1983). Moreover, the most recent meta-analytic review supports the contention that hostility is an independent predictor of heart disease (Miller, Smith, Turner, Guijarro, & Hallet, 1996). Mechanisms by which hostility confers risk. Because hostility is implicated in the development of heart disease, how it confers risk is of specific concern. One influential account is the psychophysiological reactivity model (Smith & Christensen, 1992). This model proposes that hostile individuals experience elevated cardiovascular and neuroendocrine responses to stress and that this heightened reactivity damages the heart predisposing the development of heart disease. Investigations of the relationship between individual differences in hostility and physiological reactivity, however, have revealed mixed results. While some studies have revealed a relationship between blood pressure reactivity and scores on various measures of hostility (e.g., Dembroski, MacDougall, Shields, Petitto, & Lushene, 1978; Diamond, Schneiderman, Schwartz, Smith, Voip, & Pasin, 1984; Jorgensen & Houston, 1986; Steptoe, Melville, & Ross, 1984), others have obtained negative findings (e.g., Hastrup, Kraemer, Hotchkiss, & Johnson, 1986; Manuck, Proietti, Rader, & Polefrone, 1985). Thus, a general reactivity mechanism has not been consistently supported. Alternatively, it has been proposed that the physiological reactivity experienced by hostile individuals does not generalize to all stressors, but is instead specific to social sources of stress (Smith & Leon, 1992). Indeed, studies reveal that scores on the Cook-Medley Hostility 5 Scale are associated with blood pressure reactivity to social interactions such as high-conflict role-playing (Hardy & Smith, 1988), controversial debate (Smith & Allred, 1989), and harassment during a word-identification task (Suarez & Williams, 1989), but unrelated to cardiovascular responses to standard laboratory stressors such as a cold pressor and mental arithmetic (Sallis, Johnson, Trevorrow, Kaplan, & Horvell, 1987), mental subtraction and timed Stroop tasks (Smith and Houston, 1987), and cognitive stressors (Kamarck, Manuck, & Jennings, 1990). Evidence that social stressors differentially affect hostile versus non-hostile individuals suggests that the effect of hostility on heart disease is mediated by social factors, consistent with Smith's (1992) transactional model. Smith proposed that behavior confers risk of heart disease because interpersonal hostility provokes reciprocal hostility, which in turn increases intrapersonal conflict (which can then elicit cardiovascular reactivity) and social isolation leading to an elevated risk of heart disease. The Moderating Role of Sex The way that men and women experience and express hostility and aggression differs (Knight, Guthrie, Page, & Fabes, 2002). Men are generally more aggressive and impulsive (Struber, Luck, & Roth, 2008) whereas women are more suppressive and pessimistic with their hostile feelings (Stoppard, 2000). Men and women also manifest different behaviors in threatening social situations. MacLaren, Best, and Bigney (2010) found that women were more likely to orient away from threats whereas men were more likely to choose a confrontational style of reaction. Recent findings also suggest that sex has a moderating effect on the relationship between hostility and cardiovascular function. Hughes and Stoney (2000) found that men show less parasympathetic withdrawal during psychological stressors than women. Ruiz, Uchino, and Smith (2006) replicated this finding. Weng, Lin, and Jiang (2010) found that sex moderated the 6 relationship between multidimensional hostility and health outcomes. Specifically, expressive hostility (e.g., external manifestations like verbal and physical aggression) exacerbated psychosomatic symptoms (e.g., fatigue, headache, weakness, etc) in females but buffered it in males, whereas suppressive hostility (e.g., holding anger inside) exacerbated psychosomatic symptoms in males. Thus, sex is an important variable to consider when investigating hostility. Because sex influences how individuals experience and express hostility, it is likely that sex also impacts the way that hostility affects social relationships and health outcomes. Social Cognitive Dimensions Underlying all social cognition theories is the idea that people hold cognitive representations of their social environment in memory and access those representations to process and interpret social information (Huesmann, 1998). Borrowing from information processing theories, human cognitive processes are likened to computer processes where a sequence of cognitive processes lead to the output of a particular behavior. Social behavior operates as a function of biologically determined structures (hardware), cognitive processes (software), memory databases that hold social and other information (knowledge structures), and environmental inputs (cues) perceived by the individual (Huesmann, 1998). Within this framework, knowledge structures are represented as a network of nodes interconnected by links (McNamara, 2005). Nodes can represent semantic constructs (i.e., knowledge about the meaning of words and concepts) or episodic constructs (i.e., knowledge about events, people, or objects). Links connecting nodes together represent the relationship between those nodes (e.g., the link between the node representing knowledge of cats is linked to the node for dogs by their relationship as mammals). New information can be encoded, or integrated into existing memory structures and existing information can be retrieved from memory stores via the activation of 7 associated nodes. Larger macro knowledge structures referred to as schemas (Huesmann, 1998) contain substantial information organized to facilitate comprehension about a concept, its attributes, and its relationship to other concepts. Different kinds of schemas hold different kinds of information. Self-schemas organize information about the self; other-schemas organize information about other people; event-schemas organize information about events, and so forth. Social cognitive mechanism of hostility. Two influential social information-processing models have been proposed to explain how cognitive processes mediate aggressive behavior, one developed by Huesmann and his colleagues, and the other developed by Dodge and his colleagues (Huesmann, 1998). Even though these models have mainly focused on aggressive behavior of children, a review of the proposed mechanisms mediating aggression is pertinent because similar processes likely mediate aggression in adults. Huesmann (1998) combines Crick and Dodge's (1994) reformulated social information processing model of aggression with his own initial model (Huesmann, 1988) to highlight the core processes common to both models. Both models presume that an individual faced with a social situation uses his or her preexisting set of biologically determined capabilities and cognitive database of social knowledge to evaluate and interpret situational cues, and to choose and enact an appropriate behavioral response. For example, a child that is faced with a peer who is attempting to initiate rough-and-tumble play first forms a mental representation of the social situation. The child does this by selectively attending to internal and external (situational) cues, encoding those cues, and interpreting them based on past experiences and previously acquired social knowledge. Social knowledge obtained through previous experience is stored in the cognitive database in the form of schemas. Schemas are recalled from memory and used to interpret and understand the present social situation (e.g., information about peers). 8 Interpretations of social cues can be influenced by causal inferences, which are attributions about the cause of an event or the intent of a peer (e.g., Jack wants to play with me, Jack wants to hurt me). After a situation is interpreted, possible behavioral responses are reviewed based on appropriateness of the response, self-efficacy for enacting the response, and desired outcome. A behavior is then selected and enacted (e.g., play with Jack, fight with Jack). Individual differences in cognitive processing. Despite the implied common underlying structure and function of mental processes, variations in behavior and cognition are possible because of individual differences in the interpretation of social cues, the content of knowledge structures, the accessibility of constructs, and the evaluation of behavioral responses. Because a given social situation likely contains many external cues, which cues are attended to and how they are interpreted varies from person to person. Because hostile individuals possess a bias toward interpreting ambiguous cues as more hostile than their non-hostile counterparts (Guyll & Madon, 2003), they are likely to interpret social situations as more hostile than they really are. Moreover, evidence suggests that hostile individuals selectively attend to hostile cues and ignore non-hostile cues. For instance, Allied and Smith (1991) found that hostile individuals recalled more hostile adjectives following hostile social interaction than their non-hostile counterparts did. Though social information processing models presume that all individuals represent information in knowledge structures (schemas), the content of these structures varies based on their unique past experiences. Habra (2006) found that high hostile individuals processed social support-related information differently than low hostile individuals. High hostile participants evaluated offers of social support more negatively after their social support schemas were cued than after their hostility schemas were cued, whereas low hostile showed the opposite effect. Hostile individuals are presumed to have encoded in memory a greater number of hostile 9 schemas or to possess more extensive and better elaborated hostile scripts than non-hostile individuals. Consequently, hostile schemas or scripts should be more readily activated and retrieved by hostile individuals. In addition, schemas that are frequently activated may no longer require cues to be retrieved and therefore become chronically accessible (Caprara & Cervone, 2000). Because behavioral scripts are evaluated before being enacted, the evaluation can differ between hostile and non-hostile individuals. Hostile individuals likely possess normative beliefs that condone more hostility than non-hostile individuals do. Thus, even if similar aggressive or hostile scripts are activated in hostile and non-hostile individuals, non-hostile individuals may deem them less socially acceptable and be less likely to enact them than hostile individuals. Priming: Activating knowledge structures. A line of research that informs the present is research on priming. Priming involves the presentation of category-relevant information in order to activate specific category representations or schemas in memory making them more readily accessible. Referring back to the network model of memory, access to a specific concept (e.g., cat) can be facilitated by previous exposure to a related concept (e.g., dog). According to the spreading activation model, exposure to a concept (i.e., priming) activates its internal representation allowing it to be retrieved from memory. The activation of one concept spreads to related concepts because of their meaningful connections, facilitating their retrieval. For example, concepts such as dog, cat, horse, and cow might each be represented in memory, connected by their relationship(s) to each other (e.g., they are all mammals, they all have four legs). The activation of the concept dog spreads to the related concept cat, which subsequently becomes more readily accessible because it is already partially activated (McNamara, 2005). Priming can be accomplished in a variety of ways. A classic research paradigm that employs priming is the lexical decision task in which participants are tasked with classifying 10 stimuli as either words or non-words. Meyer and Schvaneveldt (1971) found that participants correctly identified stimuli pairs as words significantly faster when they were semantically related (e.g., BREAD-BUTTER) than when they were not (e.g., BREAD-DOCTOR). Accordingly, this type of priming became referred to as semantic priming. In semantic priming the stimulus to which participants respond (e.g., to indicate whether or not the stimulus is a word) is the target and the preceding stimulus is the prime (McNamara, 2005). Priming has also been used to study attention and automatic information processing. A common priming paradigm uses a sentence construction task, in which participants are asked to construct a grammatically correct four- or five-word sentence from a group of five or six words as quickly as possible. For example, the group of words "he breaks it arm her" can be used to construct the sentence "he breaks her arm". This type of task is used to surreptitiously prime constructs of interest in order to influence subsequent ostensibly unrelated tasks without participants' explicit awareness. For example, Srull and Wyer (1979) used a sentence construction task to activate concepts associated with either hostility or kindness in order to influence a subsequent impression formation task where a target person was evaluated along several personality dimensions. Another common paradigm is a word presentation task in which prime words are presented briefly (e.g., for 50-100 ms) on a computer screen to participants across several trials to surreptitiously influence subsequent tasks. For example, Bargh and Pietromonaco (1982) presented hostility relevant words subliminally or below the threshold of conscious awareness to influence participants' evaluations of target individuals. Influencing perception and behavior. There has been a considerable amount of research in the area of automatic information processing which has demonstrated that it is possible to influence both perception and behavior via priming. Research reveals that priming 11 can influence social perception immediately following exposure to primes. For example, Higgins, Rholes, and Jones (1977) found that participants judged a hypothetical person described as ready to cross the Atlantic in a sailboat more negatively after being primed with the word "reckless" than with the word "adventurous". Likewise, Srull and Wyer (1979) found that the more participants were exposed to words semantically related to "hostility" during a sentence construction task, the more their subsequent ratings of a hypothetical person named Donald were consistent with the primed trait. Bargh and Pietromonaco (1982) found similar results by presenting hostility relevant words subliminally or below the threshold of conscious awareness. They found that the more participants were exposed to hostile words, the more negative their evaluations of Donald were. In addition to influencing impression formation, priming has been shown to affect actual behavior. For example, Bargh, Chen, and Burrows (1996) used a sentence construction task to prime participants with one of three constructs: rude, polite, or neutral (e.g., using words like disturb versus patiently versus watches, respectively). The authors found that participants in the rude priming condition interrupted the experimenter significantly faster than participants in the polite and neutral priming conditions and that a greater proportion of those in the rude condition interrupted the experimenter than those in the neutral and polite conditions. In their second study, Bargh et al. found that participants exposed to words related to the construct of "old" walked slower when leaving the experiment than those participants who were exposed to words related to the construct of "young". Direction of priming effects. The literature outlines different ways that priming can influence impression formation (DeCoster & Claypool, 2004). Priming can result in assimilation, where impressions are biased toward primes (Higgins et al., 1977; Srull & Wyer, 1979), or 12 contrast, where impressions are biased away from primes. Two different types of contrast effects are presumed to exist. First, anchoring occurs when extreme primes (e.g., Hitler as a hostile stimulus) are used as standards of comparison in impression formation tasks, causing participants to evaluate the hypothetical target as possessing less of the primed trait. Second, correction occurs when participants become aware that primed traits have influenced their impressions and consciously correct their evaluations away from the primed trait. Evidence reveals that personality can also affect the direction of priming effects. Priming can have differential effects depending on whether the individual is low or high on a given trait. For example, Maier, Bemer, and Hau (2007) found that, when primed with hostile words, low trait anxious individuals evaluated others negatively, whereas high trait anxious individuals (i.e., those who are most sensitive to hostility) evaluated others positively. Habra (2006) found that the effect of priming is also influenced by trait hostility. Specifically, she found that hostile individuals process social support-related information differently than non-hostile individuals, such that hostile participants evaluated helpful responses in stressful situations more negatively when primed with social support related words than with hostile words, whereas non-hostile participants showed the opposite effect. The Present Study There were three independent variables: sex, hostility, and condition. Sex had two levels, male and female; hostility had two levels, low and high; and condition had two levels, neutral and hostile. The present study employed a priming paradigm where primes were either semantically related (e.g., loathe, hate, despise) or unrelated (e.g., originate, scan, devise) to the concept of hostility. Priming participants with words semantically related to hostility was expected to 13 activate the construct of hostility in memory and thereby bias their subsequent perception during an impression formation task and behaviour during an interview toward a more hostile direction. Because the content of schemas differs from person to person, it was presumed that the content of hostile schemas held by high hostile individuals would differ from that of low hostile individuals and that activating these schemas would differentially affect the perceptions and behaviors of these two groups. A main effect of hostility was expected such that, relative to their low hostile counterparts, high hostile individuals would give more negative evaluations of others because they generally tend to view others more negatively. A main effect of priming condition was also expected such that, overall, the activation of hostile schemas would lead to more negative evaluations of others than the activation of neutral schemas. Of particular interest was the interaction between hostility and priming condition. It was expected that the effect of priming would depend on trait hostility. That is, it was expected that the difference between the ratings given by high hostile and low hostile individuals when primed with neutral words would differ significantly from the difference between the ratings given by high hostile and low hostile individuals when primed with hostile words. For low hostile individuals it was expected that the activation of hostile schemas would prompt more negative evaluations of others than the activation of neutral schemas. Two possible predictions were made about the results of the hostile priming and neutral conditions for high hostile individuals. First, if the activation of hostile schemas in high hostile individuals has an additive effect on the perception of others it was expected that hostile individuals would evaluate hypothetical persons more negatively when primed with hostile words than neutral words and that this difference would be significantly larger than that for low hostile individuals. Second, it is possible that the activation of hostile 14 schemas in high hostile individuals would lead them to make less negative evaluations of others than the activation of neutral schemas if they view hostility as more benign (or less malevolent) than their non-hostile counterparts. Because research has demonstrated that sex differences exist in the expression of hostility and the manifested cardiovascular response to psychological stressors, examining sex is important when investigating hostility. Thus, sex was included as an independent variable in the present study. No specific predictions were made about the pattern of results. Word Rating Study The purpose of the word rating study was to obtain the stimulus words used in the priming task. A list of 198 words was compiled, half of which were words thought to convey hostility (Appendix A) and the other half were words thought to be unrelated to hostility (Appendix B). Participants in the word rating study rated this list of words on hostility, valence, and imageability. These ratings were used to select 45 words that were rated high in hostility and low in valence (i.e., extremely negative) for the experimental condition and 45 words that were rated as low in hostility and neutral in valence (i.e., neither positive nor negative) for the control condition. In addition, the imageability ratings were used to ensure that the experimental and control words were matched on how easily they are to imagine. Using these ratings to select the stimulus words ensured the validity of the priming conditions and that any possible effects in the priming study could be attributed to the intended effects rather than differences in extraneous variables. Method Participants. Forty-five people were recruited using flyers posted around the University of Northern British Columbia. Participants were compensated with $10 cash. Six participants 15 were excluded from the analysis because they did not know over 20% of the words presented in the rating package and/or English was not their first language. Two participants were excluded because they decided to discontinue the experiment. One participant was excluded because he was obviously completing the ratings randomly. The overall sample was comprised of 36 people (20 women, 16 men), with a mean age of 27.24 (SD = 6.03). Two participants did not indicate their age. Apparatus and Materials. Word selection. The words selected for inclusion were derived from related priming studies in the literature such as those conducted by Srull and Wyer (1979) and Bargh and Pietromonaco (1982). Hostile words more closely related to physical aggression rather than hostility (e.g., beat, whip, punch, etc) were omitted in favor of words connoting covert forms of aggression (e.g., loathe, abhor, reproach, etc). Additional hostile and neutral words were selected using the thesaurus feature in Microsoft Word 2002. Specifically, synonyms of words like loathe were selected for the hostile word list and synonyms for seemingly neutral words like depict were selected for the neutral word list. The neutral and hostile words were also matched on CELEX word frequency (Baayen, Piepenbrock, & Van Rijn, 1993). The average CELEX word frequency was 12.67 (SD = 43.20) for hostile words and 13.00 (SD = 45.77) for neutral words. Rating booklet. The rating booklet was 33 pages and contained a list of 198 words which consisted of 99 words selected for their hostile connotation (e.g., MOCK, DEMEAN, YELL, etc; Appendix A) and 99 neutral words (e.g., TELL, DEVISE, GAZE, etc; Appendix B). The order of the hostile and neutral words was randomized. Three rating scales were used for each of the 198 words: hostility, valence, and imageability (Appendices C, D, and E, respectively). Hostility was rated on a 7-point scale ranging from 1 "not at all hostile" to 7 "very hostile". Valence was rated 16 on a 7-point scale ranging from 1 "extremely negative" to 7 "extremely positive". Lastly, imageability (how easily a word can be imagined) was rated on a 7-point scale ranging from 1 "not imageable" to 7 "highly imageable". Procedure The word rating study was approved by UNBC's Research Ethics Board. Participants entering the lab were greeted by the experimenter and were oriented to an experimental room and seated at a desk. An information sheet (Appendix F) with a description of the experiment was provided and the participant was given time to read through it. The content of the information sheet was reiterated verbally and participants were given the opportunity to ask questions. Participants then signed a consent form (Appendix G). Once ready to begin the experiment, participants were provided with the rating booklet and rating instructions. The presentation order of the ratings was counterbalanced to create three unique rating booklets. Accordingly, 15 participants completed the hostility ratings first, the valence ratings second, and the imageability ratings third; another 15 participants completed imageability first, hostility second, and valence third; and the last 15 participants rated valence first, imageability second, and hostility third. When the experiment was completed participants were fully debriefed verbally and given a copy of the debriefing sheet (Appendix H) that explained the purpose of the experiment. Participants were also given the opportunity to ask questions about the experiment. Analysis and Results Of the 99 hostile words rated by participants, the 45 words with the highest mean hostility ratings (M= 6.16, SD = 0.28) were selected for use in the subsequent priming study. This ensured that the words selected as hostile primes were empirically consistent with the construct of hostility. The valence ratings for the 45 selected hostile words were low (M= 1.69, 17 SD = 0.25) ensuring that these words were perceived as negative rather than positive and therefore also empirically consistent with the construct of hostility. Of the 99 neutral words rated by participants, 45 words with mean valence ratings closest to 4, meaning that they were 'neither negative nor positive', (M= 4.37, SD = 0.24) were selected for use in the subsequent priming study. This ensured that the selected words did not evoke highly negative or positive reactions. These words were also low on hostility (M= 2.82, SD = 0.27) to ensure that they were not confounded with the hostile words. An independent samples t test revealed that the words selected as hostile primes were significantly more hostile than those selected as neutral primes, /(58) = 44.79,/? < .001, SE = 0.08, 95% CI= 3.21-3.51, which confirmed that both hostile and neutral primes were consistent with the meaning they were intended to evoke. The mean hostility rating was 6.18 (SD = .30) for the hostile prime words and 2.82 (SD = .28) for the neutral prime words. Another independent samples t test revealed that the neutral primes were significantly more positive than the hostile primes, /(58) = 52.13,/? < .001, SE = 0.07, 95% CI= 2.55-2.82, which confirmed that both hostile and neutral primes were consistent with the valence they were intended to evoke. The mean valence rating was 4.35 (SD = .26) for the neutral prime words and 1.67 (SD = .28) for the hostile prime words. A third independent samples t test revealed that the hostile and neutral words selected as primes were not significantly different on imageability, /(58) = 1.94,/? = .24, SE = 0.23, 95% CI= -.18-.73. The mean imageability rating was 5.23 (SD = 0.71) for the hostile words and 4.95 (SD = 1.03) for the neutral words. Of the 45 hostile words selected for use in the priming study, 30 were randomly designated as hostile primes and the remaining 15 were designated as test words for the recognition memory test to be conducted after the priming task (Appendix I). Independent 18 samples t tests revealed that the 30 hostile primes did not differ significantly from the 15 hostile test words on ratings of hostility (M= 6.18 vs. 6.11; SD = 0.30 vs. 0.23), valence (M= 1.67 vs. 1.74; SD = 0.26 vs. 0.23), and imageability (M= 5.23 vs. 5.48; SD = 0.71 vs. 0.72). Likewise, of the 45 neutral words selected for use in the priming study, 30 were randomly designated as neutral primes and the remaining 15 as test words for the recognition memory test (Appendix J). Independent samples t tests revealed that the 30 neutral primes did not differ significantly from the 15 neutral test words on ratings of hostility (M= 2.82 vs. 2.80; SD = 0.28 vs. .27), valence (M= 4.35 vs. 4.41; SD = .26 vs. .18), and imageability (M= 4.95 vs. 5.02; SD = 1.03 vs. .69). An independent samples t test confirmed that the words selected as hostile test words were significantly more hostile than those selected as neutral test words, 7(28) = 36.29, p < .001, SE = 0.09, 95% CI= 3.12-3.50, consistent with the findings for the prime words. The mean hostility rating was 6.11 (SD = .23) for the hostile test words and 2.80 (SD = .27) for the neutral test words. Another independent samples 7 test revealed that the neutral test words were significantly more positive than the hostile test words, 7(28) = 35.14, p < .001, SE = 0.08, 95% CI = 2.51-2.83, consistent with the findings for the prime words. The mean valence rating was 4.41 (SD = .18) for the neutral test words and 1.74 (SD = .23) for the hostile test words. A third independent samples 7 test revealed that the hostile and neutral words selected as test words were not significantly different on imageability, 7(28) = 1.80, p = .08, SE = 0.26, 95% C/= -.06-.99. The mean imageability rating was 5.48 (SD = .72) for the hostile words and 5.02 (SD = .69) for the neutral words. Priming Study The purpose of the priming study was twofold. First, the priming study attempted to 19 replicate previous findings demonstrating that high hostile individuals form more negative impressions of others than low hostile individuals. Second, the priming study was conducted to determine whether the activation of hostile processes functions differently in high hostile versus low hostile individuals. Previous research shows that high hostile individuals form more pejorative impressions of others than their low hostile counterparts. Allred and Smith (1991) found that high hostile individuals rated a confederate as more hostile following social interaction with him than low hostile individuals. Likewise, Guyll and Madon (2003) found that high hostile individuals evaluated others more harshly and recalled less of the favorable information. Thus, it has been proposed that high hostile individuals possess a "hostile-other schema", which causes them to interpret social information as more hostile than it really is (Allred & Smith, 1991). If high hostile individuals do possess a hostile processing bias then they would be expected to form a more negative impression of a hypothetical person than low hostile individuals given the same information. Because the neutral priming words were unrelated to a central theme and carried little valence they were not expected to affect the subsequent person evaluation in any systematic way. Thus, exposure to the neutral priming words was expected to reveal differences between how high hostile and low hostile individuals process information when making snap judgments. Whereas the general activation of schemas is proposed to function similarly across individuals, the individual content and accessibility of those schemas is supposed to differ. Higgins, King, and Mavin (1982) found that how easily a particular construct comes to mind varies from person to person and that impressions of other people are influenced most by those constructs that are chronically accessible, or spring to mind most easily. Moreover, the more frequently a construct is activated in memory, the more accessible it becomes, requiring less 20 effort to retrieve it (Bargh & Pietromonaco, 1992). Thus, attempting to activate the same mental construct of hostility in two different individuals should yield different effects when one individual possesses a highly elaborated hostility schema while the other does not. Because high hostile individuals are supposed to possess a chronically accessible hostile-other schema, priming them with hostile cues would be expected to bring to mind hostile thought processes more easily than it would for low hostile individuals. Moreover, it has been suggested that the combined effect of prompting temporary access to a construct via priming and chronic accessibility are additive so that "people are most likely to use a construct if it is chronically accessible for them and if it is also primed" (Caprara & Cervone, 2000, p. 257). If this is true, then the activation of hostile schemas in high hostile individuals should have an additive effect on their perception of others. That is, high hostile individuals primed with hostile words would be expected to evaluate hypothetical persons most negatively, much more than low hostile individuals primed with high hostile words and more than high hostile individuals primed with neutral words. Alternatively, it is possible that the activation of hostile schemas in high hostile individuals would lead them to make less negative evaluations of others than the activation of neutral schemas if they view hostility as more benign (or less malevolent) than their non-hostile counterparts. Method Participants. A total of 75 undergraduate students at the UNBC participated in this study. Participants signed themselves up for the experiment online using the Psychology Research Participation System. Because this set of participants was sampled from the same university community as the participants in the word rating study, it is assumed that the ratings of the stimulus words provided by the first group were applicable to the second group. 21 Participants completed the Cook-Medley Hostility Scale as part of the online prescreening test. Participants were compensated with 1% course credit for their involvement in the experiment. The overall sample was comprised of 75 students (44 women, 31 men), with a mean age of 22.43 (SD = 4.94). Two participants were excluded from the analysis because their data were considered to be outliers based on two criteria. First, unlike all the other participants, their ratings on the Nice factor were higher than their ratings on the Nasty factor. This pattern was inconsistent with the hostile behavior of the target individual they were evaluating. Second, their standard scores were higher than 2.5. Apparatus and Materials. Cook-Medley Hostility Scale. The Cook-Medley Hostility Scale (Cook & Medley, 1954) is a 50-item, true-false self-report questionnaire derived from the Minnesota Multiphasic Personality Inventory (MMPI). Scores can range from 0 to 50. Participants were identified as either high or low in trait hostility using a median split of 18 (SD = 6.38). The mean hostility score for the low hostile group was 12.63 (SD = 4.00) and the mean hostility score for the high hostile group was 26.24 (SD = 5.18). Computer and software. The priming task was administered using an Intel© Pentium© D CPU with a 3.20 GHz processor and 3.00 GB of RAM. The monitor was 19 inches and set to a resolution of 1024 x 768 pixels. Participants' responses were registered using a regular computer keyboard. SuperLab 4.0.2 was the stimulus presentation software that was used for the priming and person evaluation tasks. Stimulus words. The stimulus words used in the priming study consisted of hostile and neutral primes and test words obtained in the word rating study (Appendices I and J). Vignettes. The hostile vignette is a 253 word paragraph describing a hypothetical 22 stimulus person, Daniel, engaging in several sequential hostile behaviors such as refusing to pay his rent until the landlord repaints his apartment and demanding his money back after buying a gadget at a store (Appendix K). Anger interview. The anger interview is a brief 5-10 minute interview in which the participant recalls and describes an interpersonal incident that made them angrier than they have ever been (Appendix L). Video camera. The interview was videotaped using a Sony DVCAM Digital Camcorder (Model no. DSR-PD 150) and Sony Mini DV Digital Video Cassettes. Procedure The priming study was approved by UNBC's Research Ethics Board. Participants entering the lab were greeted by the experimenter, oriented to an experimental room and seated at a desk. An information sheet (Appendix M) describing the experiment was provided and the participant was given time to read through it. The content of the information sheet was reiterated verbally and participants were given the opportunity to ask questions. Participants then signed a consent form (Appendix N). Once ready to begin the experiment, participants were oriented to the computer station where the first part of the experiment was conducted. Once seated at the computer participants were instructed to follow the prompts provided by the computer and direct any questions or concerns arising during the course of the experiment to the experimenter. When the first task was initiated by the participant, the experimenter took a seat outside the room at a desk within view of the participant. Participants completed a series of tasks on the computer which included a priming task, an impression formation task, a free recall memory test, and a recognition memory test. After completion of the computer tasks, participants were directed to another experimental room to conduct the videotaped anger interview with the experimenter. 23 Priming task. During the priming task, participants were seated at the computer station and instructed to maintain their gaze on a fixation cross (+) displayed in the middle of the screen. In the experimental condition the fixation cross was replaced by hostile words (e.g., "PERSECUTE", "LOATHE", "DESPISE", etc) whereas, in the control condition the fixation cross was replaced by neutral words (e.g., "ORIGINATE", "DEPICT", "INSPECT", etc). Participants were randomly assigned to either the control or experimental condition. Each word was displayed for three seconds and then replaced by a fixation cross displayed for half a second, then the next word, etc. Thirty unique words were presented once in each condition (Appendices F and G). Prior to the presentation of the words, participants were instructed to "really think about the meaning of each word as it is presented". A practice trial preceded the experimental trial where seven words, unrelated to the experimental words, were presented (i.e., PENCIL, PAPER, PEN, ERASER, CHAIR, TABLE, DESK). Impression formation task. The impression formation task was administered via the computer and began immediately after the priming task ended. During this task participants were prompted to read a paragraph describing a hypothetical stimulus person, "Daniel", engaging in hostile behaviors (Appendix K_). After reading the vignette, participants were prompted to evaluate Daniel by indicating the degree to which they considered each of twelve trait words (i.e., hostile, unfriendly, aggressive, kind, considerate, thoughtful, boring, conceited, narrow- minded, dependable, sympathetic, intelligent) to describe him on a 7-point scale where 1 indicated "strongly disagree" and 7 indicated "strongly agree". Participants made their ratings using the 1 to 7 numeric keys on the keyboard. The presentation order of the 12 trait words was randomized for each participant. This task was the same for both the experimental and control conditions. This vignette was taken from Srull and Wyer (1979). The paragraph was constructed 24 from participant ratings of a large pool of individual behaviors along a scale ranging from 0 ('not at all hostile') to 10 ('extremely hostile'). Five ambiguous behaviors (i.e., those rated lower than the mean rating of any behavior identified as hostile and higher than any behavior identified as non-hostile) were randomly selected to create the Donald vignette. Recall memory test. Next participants completed a free recall memory test. They were instructed to type as many words from the priming task as they could remember in two minutes. Typed words were recorded by the computer. This task was the same for both the experimental and control conditions. Recognition memory test. Next participants completed a recognition memory test. Forty-five words were presented in random order one at a time. Thirty of these words were previously presented during the priming task and 15 of the words were new words not previously presented but related semantically to the priming words used during the priming task (Appendices F and G). Each word was presented in the middle of the computer screen one at a time. For each word, participants used the keyboard to indicate whether the word was previously presented or not (i.e., they pressed 1 for "no, I did not see this word earlier" and 7 for "yes, I did see this word earlier"). Participants in the neutral priming condition received neutral words and participants in the hostile priming condition received hostile words. Anger interview. The participant was seated in the experimental room in a chair facing the experimenter, who was seated in a chair in front of the participant. The video camera was positioned just behind and to the left of the experimenter. The experimenter conducted an abbreviated Anger Interview (Appendix L) during which the experimenter asked the participant to describe an interpersonal experience where they felt the most anger they have ever felt. The participant was instructed to describe the incident. After the participant was finished describing 25 the incident, the experimenter instructed the participant to further imagine the incident, focusing his or her attention to sensory cues associated with the experience. At that point the experimenter elicited further descriptions by asking several specific questions (e.g., "What was it that made you so angry?", "How did you react?"). The interview was videotaped for use in the next experiment. When the experiment was completed, participants were fully debriefed verbally and given a copy of the debriefing sheet (Appendix O) that explained the purpose of the experiment. Participants were also given the opportunity to ask questions about the experiment. Design and Data Reduction Participants were classified as either high hostile or low hostile based on their scores on the Cook-Medley Hostility Scale (CMHS). Participants who scored higher than the median score of 18 on the CMHS were classified as "high hostile" whereas participants who scored 18 or lower were classified as "low hostile." The priming study was a 2 (Sex: male, female) x 2 (Condition: hostile, neutral) x 2 (Hostility: low, high) between-subjects factorial design. The dependent variables measured impression formation, recall memory, recognition memory, and reaction time during the recognition memory task. To measure participants' impressions of the hypothetical target person they read about after the priming manipulation, a global hostility rating was derived by taking an average of six of the 12 characteristics the target person was evaluated on (i.e., aggressive, hostile, unfriendly, kind, considerate, and thoughtful, where the last three characteristics were reverse scored). Recall memory was measured by the number of words correctly identified during the two-minute memory test. Two dependent variables were derived from the recognition memory test. First, the discrimination index (A") gave participants' ability to discriminate between old and new items on the recognition memory task. Values closer to 1 indicate good 26 discrimination performance, values closer to 0 indicate reverse discriminability (i.e., misidentifying old words as new and new words as old), and a value of 0.5 indicates poor discrimination or chance performance. The discrimination index was calculated from a pair of hit (H) and false alarm (FA) rates using Snodgrass and Corwin's (1988) equations. When H > FA, A' = 0.5 + [(H - FA)(1 + H -FA)]/[(4H(1 - FA)] and when FA > H, A' = 0.5 - [(FA - H)(l +FA- H)]/[4FA(1 - H)]. Hit rate gave the percent of old words correctly identified as old ones in the recognition memory test and was calculated using Snodgrass and Corwin's (1988) correction formula where 0.5 was added to each frequency and divided by N + \, where N was the number of old items presented (30 words). False alarm rate gave the percent of new words incorrectly identified as old ones in the recognition memory test and was also calculated using Snodgrass and Corwin's (1988) correction formula where 0.5 was added to each frequency and divided by Af+1, where N was the number of new items presented (15 words). Second, the bias index (B") gave participants' general response bias and was calculated using Snodgrass and Corwin's (1988) equations using the same pair of hit (H) and false alarm (FA) rates for each participant. When H > FA, B" = [H(l - H) - FA(1 - FA)]/[H(1 - H) + FA(1 - FA)] and when FA > H, B" = [FA(1 - FA) - H(l - H)]/[FA(1 -FA) + H(l - H)]. Lastly, three reaction time (RT) measures were obtained. Average RT for new words (i.e., test words) gave the average RT for responding to the 15 new words presented in the recognition memory task. Average RT for old words (i.e., prime words) gave the average RT for responding to the 30 old words presented in the recognition memory task. And RT difference gave the average difference in RT between the old and new words and was calculated by subtracting the average RT for old words from the average RT for new words. Analysis and Results 27 Impression formation. A 2 (Sex: male, female) x 2 (Condition: hostile, neutral) * 2 (Hostility: low, high) factorial analysis of variance (ANOVA) was conducted to determine whether sex, condition, or hostility had an effect on the global hostility rating and whether any interactions between these variables affected the global hostility rating. None of the main effects or interactions were statistically significant. Because there was no empirical basis for grouping six of the 12 trait descriptors (i.e., aggressive, hostile, unfriendly, kind, considerate, and thoughtful) to create the composite global hostility rating variable and because none of the independent variables had an effect on it, factor analysis was conducted to reduce the original 12 trait descriptors to a smaller number of empirically derived composite variables or factors. To determine the number of factors to extract, two criteria were considered. First, the absolute magnitude of the eigenvalues of the factors indicated that four factors had an eigenvalue above 1. Secondly, the scree test indicated that two factors should be rotated since the plot started to level off at the third factor. Based on these criteria a range of two to four possible factors was chosen. Direct Oblimin rotation was chosen because the factors were expected to be highly correlated with each other and therefore likely to be non-orthogonal (i.e., oblique). Since the rotated factor solution using two factors was most interpretable, it was decided that there were probably two factors underlying the trait measures used. Loadings of variables on factors, communalities, and percents of variance are shown in Table 1. The communalities for the items aggressive, considerate, dependable, hostile, narrow- minded, sympathetic, thoughtful, and unfriendly were high and those for the items boring, conceited, intelligent, and kind were low. The items considerate, dependable, kind, sympathetic, 28 and thoughtful and were most associated with factor 1 and the items aggressive, hostile, narrow- minded, and unfriendly were most associated with factor 2. Based on the content of these sets of Table 1 Factor loadings, communalities (h ) , and percent of variance for principal factors extraction and direct oblimin rotation on the 12 trait descriptors. Item F\ F2 h _ _ .64 .37 .63 .42 -.19 .59 .04 .27 -.03 .45 .39 .19 .29 .19 .01 .45 .10 .12 .72 .49 .08 .51 15.54 Note. Factor labels are: Nice (Fi) and Nasty (F2). Loadings < .40 are in bold. items, the two factors were named Nice (considerate, dependable, kind, sympathetic, and thoughtful) and Nasty (aggressive, hostile, narrow-minded, and unfriendly). These two factors accounted for 23.04% and 15.54% of the variance of the 12 variables, respectively. A 2 (Sex: male, female) x 2 (Condition: hostile, neutral) x 2 (Hostility: low, high) factorial ANOVA was conducted to determine whether sex, condition, or hostility had an effect on the Nice factor and whether any interactions between these variables affected the Nice factor. The ANOVA revealed that there were no significant main effects for sex, condition, or hostility. There was a significant, indicating that mean ratings on the Nice factor varied between men and women as a function of hostility. The interaction was broken down in two ways. First hostility at each level of sex was investigated. The simple effect of hostility at male approached statistical Aggressive Hostile Unfriendly Considerate Kind Thoughtful Boring Conceited Dependable Intelligent Narrowminded Sympathetic % of variance -.00 -.14 -.17 .75 .54 .64 .17 -.37 .67 .38 .14 .73 23.04 29 significance, F(l, 65) = 3.82,p = .055, MSE = .61, r\2 = .12. Thus low hostile males rated Daniel as significantly less nice than high hostile males. The simple effect of hostility at female was not significant. Thus low hostile females did not significantly differ from high hostile females in their ratings of Daniel on the nice factor. The significant interaction was broken down a second way, where sex at each level of hostility was investigated. The simple effect of sex at low hostility was significant, F( l , 65) = 7.86,/? < .01, MSE= .61, r\2 = .18. This means that, on average, low hostile males rated Daniel as significantly less nice than low hostile females. The simple effect of sex at high hostility was not significant. This means that, on average, high hostile males did not significantly differ from high hostile females in their ratings of Daniel on the Nice factor. No other interactions were statistically significant. The means and standard errors are displayed in Figure 1 below. A 2 (Sex: male, female) x 2 (Condition: hostile, neutral) x 2 (Hostility: low, high) factorial ANOVA was conducted to determine whether sex, condition, or hostility or any interactions between these variables had an effect on the Nasty factor. There were no statistically significant main effects or two-way interactions. There was a significant sex x condition x hostility interaction, F(l , 65) = 4.88, p = .031, r\2 = .07. To follow up this three-way interaction, a 2 (Sex: male, female) x 2 (Hostility: low, high) ANOVA was conducted at each level of sex. At the first level of sex, male, the condition x hostility interaction was not significant. This two- way interaction was not followed up further. At the second level of sex, female, the condition x hostility was significant, F(l, 65) = 3.89, MSE = .82, p < .05, i f = .09. To follow-up this two-way interaction, a one-way ANOVA was conducted at each level of hostility for sex at female. For low hostile individuals, the effect of condition approached 30 3.5 o +^ u

Psychology

Abstract

Extracted data

Psychology

Abstract

Extracted data

Related items

Related items