Examining Employee Attrition Through Job Attitudes and Work Demands: An Empirical Analysis Using Logistic Regression

Abstract

Purpose: Employee attrition imposes direct replacement costs and indirect losses of knowledge, productivity and morale. This study examines how job attitudes (job satisfaction, environment satisfaction, relationship satisfaction, job involvement and work-life balance) and work demands (overtime, frequent business travel, commuting distance and career stagnation) relate to the likelihood that an employee leaves the organisation.

Design/methodology/approach: Drawing on the job demands–resources model, conservation of resources theory and social exchange theory, nine hypotheses were tested on the publicly available IBM HR Analytics Employee Attrition and Performance dataset (N = 1,470; 237 leavers). Welch t-tests, chi-square tests of association and hierarchical binary logistic regression were used, with demographic and compensation variables as controls.

Findings: Overtime was the strongest single predictor of attrition: employees working overtime had 5.79 times the odds of leaving, after controls (95% CI 4.08–8.20). Each one-point increase in job involvement, environment satisfaction and job satisfaction reduced the odds of leaving by 41%, 34% and 33% respectively. Frequent travel, longer commutes and more years since the last promotion raised attrition risk, while longer tenure with the current manager and more training lowered it. Job attitudes added significantly to a controls-only model (likelihood-ratio χ² = 71.0, p < .001) and work demands added further (χ² = 172.2, p < .001); the full model achieved an area under the ROC curve of 0.84. Contrary to expectation, job satisfaction and work-life balance did not buffer the effect of overtime. Attrition rose from 3.8% among employees with no risk factors to 48.4% among those with four.

Practical implications: Managing workload and overtime, improving the work environment, building stable supervisory relationships and maintaining career progression offer the most promising levers for retention.

Originality/value: The study integrates attitudinal and structural predictors of attrition in a single hierarchical model and shows that structural work demands explain more variance than attitudes, a finding with direct implications for HR analytics and retention policy.