Controlling for autocorrelation in social interaction analysis.

Presents a simple method to control for autocorrelation (serial dependence) in sequential, categorical data of 2 interactants. Relying on either frequency of probability measures, a chi-square test or z test for lagged dependence is first computed and a correction factor is then applied to the result. The limitations that apply to this method are briefly discussed. (14 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)