More complex analyses

Most of the more complex analyses involve fitting a mathematical model to the data. The two most common approaches are a multivariate analysis of variance (MANOVA) or mixed effect (multilevel) models. Model-based analyses are theoretically more efficient than all the analyses presented above, because they explicitly allow for correlation of repeated measures by including a covariance structure. Adjustment for other possible variables can be made, and time or time-dependent variables can also be included in the model so that time-changing patterns can be considered. However, model-based analyses make a number of assumptions and pose a large number of restrictions. For instance, MANOVA models only use complete cases in the analyses (and thus some form of imputation is required), while mixed models assumed data are missing at random, which is not always an easy assumption to justify. The reader is referred to more specialist texts for a detailed presentation of these models [26].

0 0

Post a comment