Using Machine Learning Methods to Examine Turnover Rates in State Health Agencies

Area of Science:

• Public health workforce analytics
• Machine learning applications in human resources
• Organizational psychology

Background:

• High turnover in the public health sector challenges service continuity and knowledge retention.
• Post-pandemic factors like burnout and dissatisfaction exacerbate workforce instability.
• Machine learning (ML) offers potential for improved prediction of turnover intention.

Purpose of the Study:

• To apply ML techniques for analyzing turnover intent predictors in the public health workforce.
• To simulate the impact of enhanced workplace satisfaction on retention.
• To identify key drivers of turnover intention among state health agency employees.

Main Methods:

• Utilized 4 waves (2014-2024) of the Public Health Workforce Interests and Needs Survey data.
• Trained Lasso Regression, Random Forest, and Gradient Boosting ML models to predict intent to leave.
• Assessed variable importance and simulated turnover reduction from satisfaction improvements.

Main Results:

• ML models showed strong predictive performance (AUC 0.78-0.85).
• Job satisfaction was the primary predictor, followed by organizational and pay satisfaction.
• Simulations indicated substantial turnover reduction with modest satisfaction gains, particularly for early/mid-career staff.

Conclusions:

• ML effectively identifies key drivers of public health workforce turnover intention.
• Job satisfaction, organizational climate, and compensation are crucial for retention.
• ML tools can enable proactive, data-driven retention strategies in public health organizations.