- The five assumptions for a repeated-measures design are:
2. Homogeneity of variance.
3. Sphericity (that is: equality of the variance of the differences between each pair of values).
4. The individual variance covariance matrixes are the same for all levels of G.
5. The variance matrix meets compound symmetry (that is: constant variances on the diagonal and constant variances off the diagonal).
Note that, if we have compound symmetry, we will meet the sphericity assumption. Vice versa, it is possible, though not likely, to have sphericity without meeting the assumption of compound symmetry.
- An important advantage of repeated-measures designs is that it provides insight into a longitudinal process (that is, a process over time). In behavioral research, one is often interested in the development of certain behavior over time. This can not be obtained by a cross-sectional analysis.
- Two important drawbacks of repeated-measures designs are (1) It is sensitive to sphericity of the data; (2) the data set must be complete, or at least not contain too many missing values.