ABSTRACT
This chapter considers three actual clinical trial datasets. These datasets reflect response variables or endpoints arising from continuous, binomial, and count data. The associated statistical models for analyzing these types of data include the well-known analysis of covariance (ANCOVA)/multivariate Analysis of Covariance (MANCOVA) for continuous data, logistic regression for binomial data and Poisson regression for count data. The chapter presents these models. The analysis of the diastolic blood pressure (DBP) data takes into account the effects of the covariates: "Age" and "Sex". This analysis is referred to as "ANCOVA". Continuing the multivariate analysis of variance (MANOVA) for treatment difference, the chapter uses multivariate Analysis of Covariance (MACOVA) to incorporate covariates in correlated multivariate outcomes to test treatment difference for the changes from baseline. For logistic and Poisson regression, it emphasizes diagnostics for detecting over/underdispersion for correct modeling and presents several remedies whenever over/underdispersion is pinpointed. The chapter demonstrates how to use R and the R functionalities to analyze the data.
