1 Introduction

By definition, the primary focus of the study of human individual differences is on variation. As we have seen, the covariation between family members can be especially informative about the causes of variation, so we now turn to the statistical techniques used to measure both variation within and covariation between family members. We start by reviewing the calculation of variances and covariances by hand, and then illustrate how one may use pograms such as SAS, SPSS and PRELIS [SAS, 1988,SPSS, 1988,] to compute these summary statistics in a convenient form for use with Mx. Our initial treatment assumes that we have well-behaved, normally-distributed variables for analysis (see Section 2.2). However, almost all studies involve some measures that are certainly not normal because they consist of a few ordered categories, which we call ordinal scales. In Section 2.3, we deal with the summary of these cruder forms of measurement, and discuss the concepts of degrees of freedom and goodness-of-fit that arise in this context. During this decade advances in computer software and hardware have made the direct analysis of raw data quite practical. As we shall see, this method has some advantages over the analysis of summary statistics, especially when there are missing data. Section 2.4 describes the preparation of raw data for analysis with Mx.