ABSTRACT
The objective of this chapter is to focus on complexities associated with the multilevel models (MLM) for longitudinal data analysis in psychotherapy research, which may result in proper use or misuse of the modeling structure. The combined model combines the two levels of the model into one equation, which is sometimes referred to as the marginal model. While the MLM structure implies the marginal model, the relationship does not go the other way. When the missing data process depends on the value of the outcome variable, then the missing data are said to be informative. To assess whether the missing data are informative, one common approach is to implement pattern-mixture models. For MLM, the diagnostic approach is similar to the ordinary least squares (OLS) regression model with the exception that the residuals are no longer independent. Violation of the normality assumption affects the parameter estimates and standard errors of the random effects.
