A Meta-Analysis of Bidirectional Relations Between Parenting and Children’s Effortful Control

• In Event: Examining Relations Between Parental Behaviors and Children’s Self-Regulation with Diverse Samples and Methodologies

Abstract

Effortful control (EC), the regulatory component of temperament, has been found to be predicted by, and to predict, various types of parenting behaviors (Eisenberg et al., 2016). However, existing findings on the magnitude and direction of the relations between parenting behaviors and children’s EC are not conclusive. Thus, to obtain more comprehensive findings from both published and unpublished studies and from diverse populations, we conducted a meta-analysis of the existing literature focusing on the direction of effects and the magnitude of the longitudinal relations between parenting behavior and children’s EC. Specifically, two research questions were addressed: 1) What were the magnitudes of prediction from parenting behaviors to later children’s EC (parent effect), and of prediction from children’s EC to later parenting behaviors (child effect)? 2) If heterogeneity existed among parent and child effects, was the variance explained by a) publication status, and b) other moderators, such as sample characteristics, type of EC or parenting behavior, and aspects of study design?
After searching different databases (e.g., PsycINFO, PubMed, ProQuest Dissertation & Thesis Global) and selecting studies based on inclusion/exclusion criteria (i.e., at least one longitudinal path between parenting behavior and children’s EC with either direction of the prediction, children’s age as 10 years or younger), we included 2611 effect sizes from 290 studies in the current meta-analysis. Four coders coded the effect sizes, sample sizes, and information on study moderators based on a well-developed codebook. Reliabilities between coders were high (kappas = .76-1).
We conducted multi-level meta-analyses for parent and child effects separately. Three levels (i.e., effect size, study, dataset) were included for parent effects and two levels (i.e., effect size, study) were included for child effects because of the high overlap between the study and the dataset levels. We found small to moderate effect sizes for both the overall parent effect, r = .17, 95% CI [.16, .18], Q(1942) = 8450.14, p < .001, σ21 = .002, σ22 = .005, I2 = 75.50%, and the overall child effect, r = .15, 95% CI [.13, .17], Q(667) = 1802.77, p < .001, τ2 = .005, I2 = 63.66%, which was significantly smaller than the overall parent effect, β = -.02, 95% CI [-.03, .-002], t(184) = -2.33, p = .02. Tests of publication bias showed that for both parent and child effects, effect sizes were significantly larger in published articles than unpublished articles (e.g., dissertations, manuscripts). Further, heterogeneity existed among both the parent and the child effects. Type of parenting behaviors, type of children's EC, method similarity of parenting and EC, and consistency of informants moderated parent effects (see Table 1). Child age, time gap between assessments of parenting and EC, family structure (i.e., one vs. two parent household), the measurement of EC, method similarity of parenting and EC, and consistency of informants moderated the magnitude of child effects (see Table 2). Overall, our findings emphasized the bidirectional nature and heterogeneity of the relation between parenting and EC. We provided theoretical and methodological implications and future directions based on the findings.

Authors

• Xiaoye Xu, University of Southern California
  Presenting Author

• Tracy L Spinrad, Arizona State University

• Nancy Eisenberg, Arizona State University

• José M. Causadias, Arizona State University

• Kathryn Lemery-Chalfant, Arizona State University