Table 1 Selected characteristics of the included studies
Related Figures (7)
Calculations for the meta-analysis were performed based on procedures outlined by Lipsey and Wilson (2001). Random-effects models and the method of moments were k number of effect sizes included, r effect size (weighted mean correlation coefficient), Z test for significance of r, 95%-CI lower and upper limits of 95% confidence interval, Q test for homogeneity of effect sizes. LS longitudinal study, SE;/SE> self-esteem at first/second time of measurement, Aparenting/ASE change in parenting style/self-esteem over time *n < 0.05, **p <0.01, ***p <0.001 esteem varied between the operationalization of permis- siveness by Baumrind (1966) versus Maccoby and Martin (1983) (Q(1) =4.21, p<0.05). While the former studies found no significant association of permissiveness with self- esteem (r = .00), the latter studies found a significant, albeit very small, positive association (r= 0.07, Table 2). With regard to research question 3b, we found no empirical support for t he suggestion that cross-sectional associations would be significantly stronger than the long- itudinal associations of parenting styles at first assessment with self-esteem at follow-up. The 95%-Cls of both kinds of mean effect sizes overl apped (Table 1), and the Q-tests found no significant moderator effects of the cross-sectional versus longitudinal study design (Q(1) varied between 0.00 (p<0.96; neglectful parenting) and 2.94 (p=0.09; authoritative parenting)). Thus, we combined the results on cross-sectional and longitudinal bivariate effect sizes in the following moderator analyses. With regard to research question 3c, we found that the size of associations of parenting styles with child self- esteem were similar for mothers and fathers (Table 3). For answering research question 3d, weighted regression Table 3 Moderating effect of study characteristics on bivariate associations of parenting styles with self-esteem in children and adolescents k number of effect sizes included, r effect size (weighted mean correlation coefficient), Z test for significance of r, 95%-CI lower and upper limit: of 95% confidence interval, Q test for homogeneity of effect sizes *n < 0.05, **p <0.01, ***p <0 .001 knumber of studies, B/? non-/standardized regression coefficient, t test for significance Table 4 Moderator effects of child age on the association between parenting styles and self-esteem
