A reassessment of the time-sharing paradigm with ANCOVA

The purpose of this note is to follow up on the suggestion by Willis and Goodwin (1987) (Neuropsychologia, 25, 719-724) that analysis of covariance is the proper statistical control to eliminate the confound provided by the "initial values problem" inherent in the time-sharing paradigm. The discrepancy between baseline tapping speed of the dominant versus nondominant hand can affect concurrent tapping disruption rates independently of laterality factors. A re-analysis of previous time-sharing data showed that significant laterality main effects and interactions as shown by ANOVA are eliminated with ANCOVA procedures. Thus, when the initial-values problem is statistically controlled, the laterality effect of asymmetrical hand disruption disappears. This result seriously questions the continued use of the time-sharing paradigm as a behavioral index of language lateralization.