The effect of race on kinship support: A meta-analysis

Researchers have used three models--cultural deviance, cultural equivalence, and cultural variance--to conceptualize the African American family. Using the topic of kinship support, this paper assesses, using meta-analytic statistical techniques, the extent to which past studies find a relationship between race and four kinship support outcomes (support given, support received, perceived support, and coresidence), and the extent of support for the three models. This analysis is based on 109 independent samples contained within 47 journal articles published between 1965-1993. Indexes, abstracts, computerized databases, annotated bibliographies, and review articles from gerontology, health, psychology, social work, and sociology were used to find potential articles. Two major categories of variables are included in the analysis. Study design variables include: year of publication, year sample was collected, discipline, type of sample selection process used, sample size, relationship of support giver or receiver to the sample respondents, and the extent to which the studies include control variables. Sample characteristic variables include: residence, marital status, age, income, education, and sex. It is found initially that African Americans are significantly more likely to give support to kin, receive support from kin, and coreside with kin. With the exception of coresidence, however, these differences are quite small. It is also found that both study design and sample characteristics have an impact on the race effect. Thus, there are certain conditions under which this race effect favoring African Americans tends to exist. This is quite evident for support given and support received. The race effect for coresidence favoring African Americans remains fairly strong even after controlling for the effects of third variables. Finally, a close examination of the data reveals some support for all three models. Data from support given and support received provide some support for the cultural deviance model. Data from perceived support, support given, and support received provide some support for the cultural equivalence model, and data from coresidence provide support for the cultural variance model