Multivariate modeling of organizational dynamics and faculty well-being in higher education

Leadership styles are increasingly recognised as strategic organizational resources for faculty well-being, yet the evidence remains fragmented: prior studies have examined leadership, workplace harassment-systematic, recurrent negative acts that constrain the target's capacity for self-defence-and psychological resources in isolation, and rarely within a single explanatory model. Whether these processes operate equivalently for male and female faculty has been largely untested, despite its direct relevance for equitable university management.

To test an integrated structural model relating transformational and servant leadership to workplace harassment, teacher self-efficacy, and life satisfaction among Peruvian university teachers; to formally test the mediating role of self-efficacy against directly estimated residual effects; and to examine whether the structural relations differ by gender.

A quantitative, non-experimental, cross-sectional study with explanatory scope.

506 university teachers from Peru (49.6% men, 50.4% women; 61.66% from public institutions).

Structural Equation Modelling with robust WLSMV estimation on ordinal indicators, including directly estimated residual paths from both leadership styles to life satisfaction. Indirect effects were evaluated with the product-of-coefficients method and bootstrap bias-corrected accelerated confidence intervals ( $B\mathrm{=5000}$ ; ML sensitivity analysis). Measurement invariance across gender was tested with a configural-metric-scalar (threshold) sequence under WLSMV, and structural gender differences were assessed with a formal likelihood-ratio comparison of free versus constrained multigroup models, followed by Wald tests of path differences. Robustness to age, university type, and region was verified through a covariate-adjusted sensitivity model.

Both transformational ( $\beta =-0.123$ , $p0.01$ ) and servant ( $\beta =-0.186$ , $p0.001$ ) leadership were negatively associated with workplace harassment. Servant leadership showed the strongest effect on self-efficacy ( $\beta \mathrm{=0.505}$ , $p0.001$ ), which in turn predicted life satisfaction ( $\beta \mathrm{=0.429}$ , $p0.001$ ; $R^2\mathrm{=0.393}$ ). Servant leadership exhibited complementary partial mediation: both its indirect effect via self-efficacy ( ${\beta }_{\mathrm{std}}\mathrm{=0.217}$ , BCa CI $\left[0.146,0.354\right]$ ) and its direct residual effect ( $\beta \mathrm{=0.251}$ , BCa CI $\mathrm{[0.198,}\mathrm{0.597]}$ ) were significant. Transformational leadership showed a small indirect-only effect ( ${\beta }_{\mathrm{std}}\mathrm{=0.033}$ , BCa CI $\left[0.003,0.101\right]$ ) with a non-significant direct path. Serial paths through harassment were not significant. Measurement invariance across gender was supported ( $\Delta$ CFI $\le 0.003$ at every step), licensing structural comparison; full structural equivalence was rejected ( $\Delta {\chi }^2\mathrm{(8)=17.86}$ , $p\mathrm{=0.022}$ ), with Wald tests localising the differences exclusively to the transformational-leadership paths: its protective effect against harassment operated among women ( $\beta =-0.238$ , $p0.001$ ) but not men ( $\beta =-0.022$ , n.s.), whereas its effect on self-efficacy operated among men ( $\beta \mathrm{=0.157}$ , $p0.05$ ) but not women ( $\beta =-0.024$ , n.s.). All conclusions were robust to age, university type, and region (all $\left|\Delta \beta \right|\le 0.012$ ).

Servant leadership functions as a gender-equivalent organisational resource for faculty well-being, whereas transformational leadership operates through gender-specific pathways-protecting women against workplace harassment and strengthening men's efficacy beliefs. These findings provide differentiated, data-driven guidance for university management aligned with Sustainable Development Goals 5 (gender equality) and 16 (just institutions), indicating that leadership development must be gender-responsive rather than uniform.