Prosocial behavior tendency and responsibility-taking tendency among Chinese college students: the “broken chain” within a specific slice of the CASEL framework

The CASEL framework posits that relationship skills influence responsible decision-making through self-awareness and self-management, while Gross’s process model suggests that emotional awareness (EA) enables cognitive reappraisal (CR). Integrating these two frameworks suggests a sequential pathway (relationship skills → EA → CR → responsible decision-making), yet whether this pathway holds in non-Western, collectivist educational settings remains untested. Here, we deliberately test a specific slice of the CASEL framework–operationalizing the predictor as prosocial behavior tendency (PBT) rather than peer relationships, and the outcome as responsibility-taking tendency (RTT) rather than full responsible decision-making. Chinese private college students ( N = 516, 56.4% medical majors, M = 19.29, SD = 0.90, 68.2% female) completed measures of PBT, EA, CR, and RTT. We tested the sequential mediation model (PBT → EA → CR → RTT) using structural equation modeling with bias-corrected bootstrap CIs (5,000 resamples), and compared medical versus non-medical students via multi-group analysis. Scalar invariance was supported (Δχ 2 = 18.36, Δdf = 22, p = 0.684; ΔCFI < 0.01), validating the multi-group comparisons. The model showed mixed discriminant validity (HTMT acceptable, but Fornell-Larcker failed for EA-CR) and acceptable fit (CFI = 0.907, TLI = 0.896, RMSEA = 0.061). In the full-sample four-factor model, PBT directly predicted RTT (β = 0.619, 95% CI [0.452, 0.785]), but none of the indirect paths were significant (all BC CIs included zero), and CR → RTT was also non-significant (β = 0.123, p = 0.212). Regression invariance was supported (Δχ 2 = 9.80, Δdf = 6, p = 0.133), indicating no significant moderation by academic discipline. Contrary to predictions, the sequential mediation was unsupported when EA and CR were treated separately. When EA and CR were combined into a single emotion competence factor, a significant indirect effect emerged (B = 0.264, 95% CI [0.169, 0.359]), a pattern consistent with the second-order model. This indicates that EA and CR share substantial variance and function as an integrated emotion competence rather than as independent sequential steps. These findings delimit the generalizability of the sequential assumption within this specific CASEL operationalization, and underscore the outcome-domain specificity of emotion regulation. Our conclusions are strictly limited to prosocial behavior tendency and responsibility-taking tendency, and should not be generalized to the full CASEL construct.