Quantify informant effect on the relation between parenting and adolescent outcomes using meta-analyses

• In Event: Poster Session 11
  In Poster Session: PS 11 Topic - Methods, History, Theory

Abstract

BACKGROUND: Different informants often provide inconsistent information on the same study constructs (De Los Reyes et al., 2015; Hou et al., 2019). While there is an overwhelming concern about the effect of method bias on the association between two study constructs, there are debates regarding whether method bias inflates, deflates, or has no effect on estimates of the relationship between two constructs (Podsakoff et al., 2012). Prior studies are inconsistent regarding the extent of informant effect on the relationship between parenting and adolescent outcomes (Kawabata et al., 2011; Pinguart et al., 2017). The current study aims 1) to quantify the extent of informant effect on the association between parenting and adolescent outcomes and 2) to test whether informant effect varies across parenting dimensions, adolescent domains, and informant characteristics (i.e., age, gender, ethnicity, socioeconomic status, and family structure).

METHODS: We conducted a systematic literature search of academic journal articles and dissertations, abstract screening, full-text screening, and emailed authors when necessary data were not available in the paper. We included 214 articles that involved both parent- and adolescent-reported parenting at the same wave and provided sufficient data to calculate effect sizes of the relation between parenting and adolescent outcomes. Meta-regression and subgroup analyses were conducted using the “robumeta” R package.

RESULTS: The current study demonstrated a small yet statistically significant common-informant effect (b = .08, p < .001). The positive association between parenting and adolescent adjustment was stronger when they were reported by the same informants (r = .20, 95% CI [.18, .22]) than when they were reported by different informants (r = .12, 95% CI [.10, .13]), indicating an common-informant inflation of 167% (rsame.informant / rdiff.informant). The extent of common informant effect did not significantly vary across parenting dimensions and adolescent outcome domains as well as informant characteristics.

CONCLUSIONS: The findings highlight the need for controlling for method bias. Various possible approaches to controlling for method bias will be discussed; for example, one possible approach is to adjust results using the common-informant inflation estimated in the current study.

Authors

• Yang Hou, Florida State University
  Presenting Author

• Aprile D Benner, University of Texas at Austin
  Non-Presenting Author

• Su Yeong Kim, University of Texas at Austin
  Non-Presenting Author

• Shanting Chen, University of Florida
  Non-Presenting Author

• Shiri Spitz
  Non-Presenting Author