APPLIED MULTIVARIATE STATISTICS: WITH SAS® SOFTWARE

6.7. A Random Coefficient Model

Another approach to the modeling of repeated measures data is to use an appropriate random coefficient model. In this approach, one or more regression parameters are assumed to be a random sample from a population of regression coefficients. These models are useful whenever the regression model for fitting the repeated data on a subject can be assumed to be a random variation of a population regression model. This approach can be especially useful due to its capacity to handle unequally spaced and/or unbalanced growth curve type data.

Let y_iu be the p_iu×1 vector of repeated measures on the u^th subject of the i^th treatment group. Then consider a mixed effects model described as

where X_iu and Z_iu are the known ...