Assessing the prediction of employee productivity: a comparison of OLS vs. CART

Abstract

This research compares the results of utilising an ordinary least squares (OLS) approach vs. a classification and regression tree (CART) approach for identifying employees with a high likelihood of being productive. Relevant performance data were collected from 378 employees of a large garment manufacturer. Past research (Markham et al., 2006) has shown that a combined genetic algorithm with an artificial neural network substantially outperformed (R² = 0.30) an equivalent OLS solution (R² = 0.14) when predicting individual level productivity. The current research compares the use of CART to OLS using the same data set. With an R² of 0.43, the CART results were even more powerful in identifying and classifying high performance employees. The implications of this finding for the field of productivity research and employee selection are discussed.