A Cross-Cultural Quantitative Review of Research on Cross-Informant Correspondence in Reports of Child and Adolescent Mental Health

Child and adolescent mental health clients display mental health concerns in various ways, depending on the context (e.g., home vs. school). Any one client may display concerns in one or more contexts, and any two clients may display concerns in different contexts (e.g., home but not school vs. school but not home). Assessments that do not account for these individual differences may potentially “miss” identifying clients’ concerns. Thus, best practices in clinical assessments involve collecting multiple informants’ reports, and more specifically, reports from informants who observe behavior in multiple, relevant contexts.

Recent theoretical work highlights that low cross-informant correspondence reflects the idea that (a) informants vary in the contexts in which they observe child and adolescent behavior and (b) children and adolescents vary in the contexts in which they express behaviors measured in clinical assessments (De Los Reyes et al., 2013). If true, these ideas can inform strategies for using levels of cross-informant correspondence to interpret assessments of diagnostic status, intervention response, and context-specific symptom expressions (Dirks et al., 2012).

Should practitioners and researchers expect low levels of correspondence regardless of the cultural context? To address this question, we report findings of a meta-analysis of over 400 studies (1989-2016) that reported cross-informant correspondence estimates. In addition to coding correspondence (i.e., using the Pearson r metric), we coded the country of origin. We also coded levels of cultural “tightness” (i.e., strength of norms, tolerance for violations of norms) based on national estimates reported in recent work using the Tightness-Looseness Scale (Gelfand et al., 2011). We coded studies of cross-informant correspondence from over 30 countries. Despite the cross-cultural variation in research sites, we observed remarkably stable levels of cross-informant correspondence (mean r=.28; 95% Confidence Interval: [.25, .31]). Within this narrow band of variation in correspondence, we observed significant relations between correspondence and cultural tightness. Specifically, the greater the degree of “cultural tightness” of countries (i.e., strong norms and low tolerance for violating norms), the greater the level of cross-informant correspondence, b = 0.38, Z = 8.33, p<.001.

These robust findings regarding cross-informant correspondence research may inform the development of concepts for using and interpreting the outcomes of multi-informant assessments of child and adolescent mental health. Thus, findings have important implications for interpreting the reproducibility and replicability of preventive interventions, as well as for informing modifications to interventions.