ABSTRACT
Statistical data analysis is model driven. Certain assumptions must be made regarding the distribution of the data and the dependency between the pretest and posttest measurements in order to obtain the significance of a treatment effect. We saw the beginnings of this in the last chapter where we specified the linear model for pretest and posttest measurements. This chapter and the next (Relative Change Functions) describe the most common methods used to analyze pretest-posttest data. Recall from the previous chapter that in order to assess “change,” some measure of baseline activity, which we have called the pretest, must be made prior to application of the treatment. Also recall that the expected value of the posttest score, the measurement collected after application of the treatment, is dependent on the pretest score. As was seen in Chapter 1, analysis of posttest scores without considering the pretest scores may result in analysis bias and erroneous conclusions.
