Multi-group comparisons of gender for relationships among the variables
Multi-group analysis was employed to assess the model’s fit across different groups. A suitable model fit would suggest that sample attribute variables had moderating effects, leading to subsequent group difference analysis. To explore potential variable variations based on demographic characteristics, nested models were constructed using the grade and gender of children as group variables, utilizing M-plus. In line with the full model, M1, equivalent models were created using multiple group comparisons within the SEM framework. Firstly, an unconstrained M2 (morphological equivalence) model was developed for gender. Subsequently, fully constrained models, M3 (loadings equivalence), were derived from M2. As presented in Table 5, the results revealed that two models (M2 and M3) exhibited acceptable fit indices. However, upon comparing the models, it was evident that there was no significant difference in χ² between M2 and M3. This suggested that the correlation between the variables remained consistent across genders.
<Insert Table 5 HERE>
Given the constancy of inter-variable interactions across genders, further analysis was conducted to examine the patterns of these paths across genders. Notably, boys’ closeness TCR did not significantly predict SR (see Figure 2). Similarly, girls’ conflict TCR did not exhibit significant effects on SE (see Figure 3). In the remaining pathways, no significant gender differences were observed. In particular, SR emerged as the strongest predictor, with predictions of mathematics and vocabulary reaching high values of 0.496 and 0.515 for boys, and 0.591 and 0.472 for girls, respectively. Additionally, closeness TCR demonstrated positive predictive effects on SE for both boys and girls, with slightly stronger predictive values for girls than for boys (0.314 > 0.228). The predictions of SE for mathematics and vocabulary did not exhibit gender difference. However, the predictive effect on vocabulary was somewhat greater for boys than for girls (0.306 > 0.277).
<Insert Figure 2 HERE>
<Insert Figure 3 HERE>