An Examination of Cross-Racial Comparability of Mother-Child Interaction among African American and Anglo American Families

Article excerpt

This study examined the cross-racial comparability of maternal quality of assistance and supportive presence coded from a video protocol using data from the Infant Health and Development Program for low-birth-weight, premature 30month-olds and their mothers. Evidence of equivalence of measures is necessary before comparisons can be made across groups. Multiple-group mean and covariance structures analysis was used to demonstrate the invariance of the measures and make comparisons for Anglo American and African American treatment and comparison groups of dyads. Comparisons across groups indicated similar variances and correlation between child and maternal behavior. Differences were found between the mean scores, with Anglo American treatment families scoring the highest.

Key Words: cross-cultural, measurement comparability, mother-child interaction.

Social scientists are showing a growing awareness of the need to examine cross-cultural (including racial and ethnic) generalizability of family processes and interactions (Berry, 1989; Harkness & Super, 1995; Rohner & Cournoyer, 1994). In particular, developmentally oriented researchers are interested in investigating the universal aspects of the relation between developmental outcomes and specific parenting practices. To conduct this work, quantitative investigation must rely on the use of instruments known to be comparable across populations. There is a serious gap in the understanding of the equivalence of parenting measures among subgroups, however (Holden & Edwards, 1989; McGuire & Earls, 1993).

There are several reasons researchers and practitioners working with populations in the United States are interested in the impact of cultural differences on family processes such as parenting behavior and parent-child interactions. First, aspects of parenting are thought to have direct impact on children's growth and development (Barnard, Bee, & Hammond, 1984; Bradley, Mundfrom, Whiteside, Casey, & Barrett, 1989). Second, for intervention programs to effectively support families, it is imperative that programs be structured to consider the influence of the families' culture on parenting behavior and interactions with children. Finally, the growing ethnic and racial diversity in the United States means that programs directed toward improving family processes will increasingly encounter families of diverse cultural backgrounds.

Minority mothers in U.S. populations frequently score lower than White mothers on measures of parenting behavior, which has led some researchers to question the comparability of these measures across cultural groups (e.g. Berlin, Brooks-Gunn, Spiker, & Zaslow; 1995; Bradley et al., 1994; Sugland et al., 1995). In their examination of maternal behavior, Berlin and her colleagues questioned the differences found between African American and Anglo American families in the prediction of child language and behavior from constructs. For example, maternal supportive presence was predictive of children's receptive language abilities for African American children but not for Anglo American children. Because the comparability of the meaning of the observational coding system between African American and Anglo American samples is not known, Berlin and her colleagues concluded that it was "possible that the measures of parenting ... are not equivalent across the two racial groups" (p. 681). Sugland et al., in their investigation of the comparability of a measure of parenting behavior, found that "dimensions of parenting are not equally important in explaining child outcomes for different racial/ethnic subgroups" (p. 632). Although concerns about the comparability of the measurement of parenting behavior prompted their investigation, the techniques they used are not thought to provide sufficient evidence of comparability (Meredith, 1993). Again, the possibility that the measures of parenting across the racial groups were not equivalent cannot be ruled out. …