Chapter 4 124 adolescent psychological functioning, we assumed stationarity. Subsequently, we estimated six ML-VAR(1) models with Mplus 8.5 (Muthén & Muthén, 2020), combining 2 parenting variables (psychological control/support) with 3 types of adolescent outcomes (self-esteem/depressive symptoms/anxiety symptoms). At the within-family level, we estimated the concurrent and bidirectional lagged effects as well as the autoregressive effects. At the between-family level, we estimated the variance around the within-family effects (i.e., random effects) and the associations between all random effects and with the random intercepts. To account for unequal time intervals between measurements due to missing data, we set TINTERVAL to 1. Moreover, to account for convergence issues, we simplified two out of six models by removing the between-family associations between the random lagged and autoregressive effects. Still the model with parental psychological control and adolescent anxiety symptoms did not converge, which left us five models to test our hypotheses. An overview of the model specifications and settings for each final interpreted ML-VAR(1) model can be found in Table A1 in Appendix A. Inference Criteria and Hypothesis Testing The hypothesized average parenting effects (H1) were derived from fixed within-family lagged effects from parenting to adolescent psychological functioning (significant when Bayesian credible intervals did not include zero). Subsequently, between-family variance around these average within-family lagged effects (H2) was investigated. To investigate which theoretical responsivity patterns would emerge in the sample (H3), we summarized how the five standardized within-family lagged effects combine within an individual adolescent (using STDYX standardization and using the R package “Mplus Automation”; Hallquist & Wiley, 2018). Individual effect sizes were interpreted based on a smallest effect size of interest of .05 (SESOI; Beyens et al., 2021; Lakens et al., 2018), which can be considered a small to medium lagged within-family effect according to recent guidelines (Adachi & Willoughby, 2015; Orth et al., 2022). Hence, we considered effect sizes smaller than .05 as null effects (-.05 > β < .05), effects with a size of β ≥ .05 as positive effects, and effects with a size of β ≤ -.05 as negative effects. Table 1 shows an overview of the inference criteria per responsivity pattern. Finally, to test H4, we compared the subgroups on their mean scores of the HSC, using a two-sided alpha of .05. Deviations From Preregistration We followed our preregistered plan in almost each step, with the following exceptions. In contrast to our preregistered plan, we included participants who had no over-time variance
RkJQdWJsaXNoZXIy MTk4NDMw