4 For better, for worse, or both? 133 Concurrent Effects (Exploratory) As exploration, we tested whether the main findings (i.e., lagged effects) would replicate with concurrent parenting associations, produced by ML-AR(1) models including adolescent-reported parenting as a time-varying covariate (see Table F1 in Appendix F). Different than the main findings, all hypothesized average parenting effects were found with the concurrent models (H1a-d confirmed; see Table F2 in Appendix F). Moreover, similar as the main findings, all parenting associations showed meaningful betweenfamily variance (H2 confirmed), although effect heterogeneity was smaller. Regarding the responsivity patterns, we found similar patterns for the concurrent associations as for the lagged effects (H3 confirmed; for full overview, see Table F3), including no unsusceptible adolescents. Nevertheless, based on concurrent effects, the majority of the sample (61% vs. 26% with lagged effects) was classified as differential susceptible and a small percentage as adverse sensitive (2%) and vantage sensitive (1%). Notably, 93% of adolescents who had a differentially susceptible pattern with lagged effects had a similar pattern with concurrent effects (for a detailed comparison, see Table F4). However, 63 of the 156 adolescents who had differentially susceptible pattern with concurrent models showed an unexpected responsivity pattern (i.e., opposing parenting effects) with lagged models. Moreover, fewer adolescents showed opposing parenting effects (36% vs. 66% with lagged effects). Furthermore, again H4 was not confirmed, as differentially susceptible adolescents did not differ in trait environmental sensitivity compared to all others, W = 8358, p = .150. Additional Sensitivity Analyses (Exploratory) To further assess the robustness of the findings concerning the responsivity patterns, we additionally conducted sensitivity analyses. We explored to what extent: (a) the classification was influenced by the effect size cut-off by raising the cut-off to .10 (see results in Table G1 in Appendix G); (b) the classification was influenced by participants who had five or less observations (n = 23; see Table G2); and (c) the estimation of individual effect sizes was influenced by the inclusion of participants who had no over-time variation (n = 72; see Table G3). Although group sizes slightly varied across analyses, we found the same predicted and unpredicted responsivity patterns as in the main analyses, in which differential susceptible (20% to 28%), “opposing effect of parental support” (24% to 30%), and unperceptive (25% to 28%) were again the three largest subgroups. Hence, we conclude that the main findings were robust across the abovementioned methodological factors as the results remained in line with our main hypothesis that different responsivity patterns coexist.
RkJQdWJsaXNoZXIy MTk4NDMw