Causal Direction Between Workplace Emotions and Employee Turnover

Determine whether negative emotions captured via continuous facial expression analysis in the WELD dataset causally lead to employee turnover, or whether impending turnover causally increases negative emotional states, thereby establishing the direction of causality between observed emotional dynamics and departure outcomes.

Background

The paper demonstrates strong predictive validity for turnover using extended emotional metrics derived from 30.5 months of continuous facial expression data, including perfect aggregate classification in some analyses. Despite these predictive results, the authors emphasize that their analyses are correlational and do not establish causal relationships.

Clarifying whether negative emotions cause turnover or whether turnover-related factors cause negative emotions is crucial for designing ethical and effective interventions, interpreting model predictions, and guiding organizational practices. Establishing causality would validate whether emotion-aware systems should target emotional regulation or address structural workplace factors that precipitate departures.