Developmental Variation in Racial/Ethnic Disparities in Depressive Symptoms: An Individual Participant Data Meta-Analysis

• In Event: Leveraging Innovative Methods in Meta-Analysis to Advance Developmental Theory

Abstract

A growing body of research across various scientific disciplines (e.g., medicine, clinical psychology) has documented racial/ethnic disparities in the development of depressive symptoms. The magnitude of these disparities across groups in the U.S. is not well understood. The cultural similarities hypothesis proposes that Whites and minorities are more similar than different in levels of depressive symptoms, and these differences are equal or smaller in magnitude than differences between- and within-minority groups (Causadias et al., 2018). In contrast, the cultural differences hypothesis states that differences between Whites and minorities in levels of depressive symptoms are larger in magnitude than differences between- (e.g., African Americans, Latinos) and within- (e.g., Latinos: Mexican Americans, Cuban Americans) minority groups the opposite. In the present study, we conducted a structural equation modeling (SEM) based meta-analysis, which combines the strengths of meta-analysis and SEM (Cheung, 2015; Viswesvaran & Ones, 1995), to understand cultural differences and similarities in the development of depressive symptoms.

We conducted a systematic review and meta-analysis of individual participant data (IPD) from nationally representative studies in the U.S., to: 1) Estimate the overall average difference of depressive symptoms between Whites and minorities, as well as between- and within- minority groups; 2) Determine if study and methodological moderators account for these differences; 3) Test the cultural differences and similarities hypotheses. We included studies from the Inter-university Consortium for Political and Social Research (ICPSR) that contained a measure of depression/depressive symptoms, adequate sample size for two or more ethnic groups, and self-reported race/ethnicity. We identified a total of 73 datasets on 27 nationally representative survey series (N = 2,116,853).

We employed a three-level random-effects models and moderation analyses of study (e.g., age, sex, education, income, occupation, and socioeconomic status) and methodological characteristics (e.g., measurement reliability, depression measure) to examine potential sources of heterogeneity in effect size estimates (Figure 1). We meta-analyzed the IPD using a two-step analysis by standardizing group differences (Cohen’s d) in the first step and then aggregating across studies in step two using SEM. The advantages of this approach compared to traditional meta-analytic methods are that IPD allows examination of within and between group comparisons, that multilevel modeling supports extracting multiple comparisons from the same study, and that full information maximum likelihood estimation can be specified using SEM for missing moderators.

Our findings are consistent with the cultural similarities hypothesis. The average absolute difference between White and minority participants was d = 0.09, 95% CI[0.07,0.11], while the average absolute difference between minority participants was d = 0.07, 95% CI[0.06,0.09] and within minority participants was d = 0.10, 95% CI[0.06,0.15]. The mean age of the sample moderated the absolute difference between Whites and minorities (B = .001, p = .006) and between minorities (B = .001, p = .031), suggesting that the average absolute difference was larger in magnitude in samples with higher mean ages. The present study will contribute to advancing current knowledge of utilizing SEM based meta-analysis from a developmental perspective.

Authors

• Karina Cahill, Arizona State University
  Presenting Author

• Kevin Michael Korous, GDIT
  Non-Presenting Author

• Longfeng Li, Pennsylvania State University, University Park
  Non-Presenting Author

• Eiko Fried, Leiden University
  Non-Presenting Author

• José M. Causadias, Arizona State University
  Non-Presenting Author