chapter  8
The Forest Revisited Scales of Measurement and Multivariate Statistics
Pages 38

There are at least four issues that cut across the specific materials of Chapters 2 through 7, but that also require some acquaintance with the material in those chapters for meaningful discussion. The present chapter discusses (a) the applicability of the multivariate techniques discussed so far to data having “weaker” than interval scale properties; (b) the robustness of multivariate significance tests in the presence of violations of the assumptions (especially multivariate normality) used to derive them; (c) how nonlinear relationships among or between variables can be detected and how to handle nonlinearity when it is detected; and (d) the utility of the very general techniques provided for testing the multivariate general linear hypothesis and for modeling structural equations as alternatives to the more specialized procedures described in Chapters 2 through 7. Finally, some suggestions for further study are offered.