Cross-Validation of Regression Equations to Predict Performance in a Pursuit Tracking Task

Abstract

A double, cross-validation procedure was used to validate regression equations which predict training time to learn a two-dimensional pursuit tracking task. Motor skill and information processing tasks were used as predictors. The results yielded a reliable regression equation for each training condition, and these equations were quite similar in cross-validation. Subsequently, a regression equation based on pooled data from the original and cross-validation sample was calculated for each training condition. To establish the usefulness of a regression approach for selecting training strategies, these equations will be used in a future study where students will be matched, mismatched, and randomly assigned to various training alternatives.