Cross-Informant Agreement between Parents and Adolescents in 20 Cultures

As early as the Isle of Wight study (1964), it was reported that parents and children demonstrated only modest agreement about children’s problems. A meta-analysis by Achenbach et al. (1987) reported a mean correlation of only .25 for parent-youth agreement. Many cross-informant studies have been conducted in the U.S. and abroad, but these studies have varied in the informants and instruments used, scales assessed, and statistical methods utilized. Multicultural cross-informant agreement has not been heretofore investigated using consistent methods across data sets. The present study was a multicultural investigation of cross-informant agreement between parents’ reports on the Child Behavior Checklist (CBCL; Achenbach & Rescorla, 2001) and adolescents’ reports on the Youth Self-Report (YSR; Achenbach & Rescorla, 2001). The study’s focus was the consistency of cross-informant agreement patterns across 20 cultures, including an examination of informant effects on mean scale scores, mean scale correlations, mean item and dyadic item scores, and categorical agreement. Informant had a significant effect on all mean scores, with adolescents scoring higher than parents in all 20 cultures (effect sizes of 16% of variance on Internalizing Problems, 17% on Externalizing Problems, and 22% on Total Problems’ scores). Mean YSR-CBCL discrepancy did vary significantly across culture, but culture x informant effect sizes were generally small. Parent-adolescent correlations averaged .41 on Internalizing, .44 on Externalizing, and .43 on Total Problems. Mean item correlations were high across cultures, averaging.82-.84 across broad-band scales with little variability. Dyadic Q correlations were much lower than mean item Q correlations, indicating that there was great variation across dyads within each culture in their item score agreement. Agreement statistics concerning the adolescent’s deviant status indicated that informant agreement was consistently high across cultures when the adolescent was deemed healthy, with specificity ranging from 86% to 87% across broad-band scales. When one informant indicated the adolescent had deviant status, there was much lower agreement and wider variation between cultures. Sensitivity ranged from 40% to 42%across aggregate scales. Overall, there was a great deal of cross-cultural consistency in patterns of parent-adolescent cross-informant agreement using five different statistical methods of analysis. Although the 20 cultures studies varied greatly in geographical, historical, economic, political, religious, and ethnic characteristics, they were very similar with respect to agreement between adolescents and their parents regarding the adolescent’s behavioral and emotional problems