Factor and simplex models for repeated measures: application to two psychomotor measures of alcohol sensitivity in twins

Abstract

As part of a larger study, data on arithmetic computation and motor coordination were obtained from 206 twin pairs. The twins were measured once before and three times after ingesting a standard dose of alcohol. Previous analyses ignored the time-series structure of these data. Here we illustrate the application of simplex models for the genetic analysis of covariance structures in a repeated-measures design and compare the results with factor models for the two psychomotor measures. We then present a bivariate analysis incorporating simplex processes common and specific to the two measures. Our analyses confirm the notion that there is genetic variation affecting psychomotor performance which is "switched on" in the presence of alcohol. We compare the merits of analysis of mean products versus covariance matrices and confront some practical problems that may arise in situations where the number of subjects is relatively small and where the causal structure among the latent variables places a heavy demand on the data.