Transforming Labor-Management Conflict Through a Data-Driven Approach: Addressing Dissatisfaction and Motivation Under Uncertainty

In increasingly complex public-sector environments, employee dissatisfaction and labor–management conflict pose significant challenges to organizational effectiveness. This study analyzes 823 administrative separation records and 822 survey responses from Queensland, Australia to examine predictors of voluntary turnover and interpersonal conflict. Because several outcomes are rare and the data contain potential separation, Firth penalized logistic regression is employed as the primary estimator to reduce small-sample bias, with standard logistic models reported as sensitivity checks. The Firth models yielded no statistically significant predictors, a finding interpreted as informative evidence of measurement constraints, low variance in outcome variables, and limited explanatory power of dichotomous indicators in administrative datasets. Sensitivity analyses using standard logistic regression revealed associations between lack of recognition, work–life balance, and voluntary exit; however, these results are presented cautiously due to potential bias under rare-event conditions. Interpersonal conflict exhibited only directional (non-significant) relationships with job insecurity and weak supervisory support. Rather than presenting a definitive predictive model, the study contributes methodologically by demonstrating how penalized regression and sensitivity checks can clarify interpretation when null results arise. The paper recommends refined measurement (using multi-item scales), richer longitudinal or mixed-methods designs, and the integration of unstructured data to better capture the drivers of employee dissatisfaction and conflict in future research.