Normative Data Accounting for a Non-binary Qualitative Independent Variable

This chapter focuses on the derivation of regression-based normative data that account for a non-binary qualitative independent variable (such as Level of Education, with outcome values low, average, and high). It is explained that the inferential procedure that is based on \(T^{*}\) test statistics (see Chap. 3 ) is suboptimal in this setting, because the evaluation of the effect of a non-binary qualitative independent variable requires the testing of multiple hypotheses simultaneously. The General Linear Test procedure is introduced as the preferred inferential framework in this setting. In the latter approach, statistical significance is evaluated by comparing the fit of a so-called unrestricted model (that contains the independent variable of interest in the mean structure) with the fit of a restricted model (that does not contain this variable). The methodology to derive regression-based norms that account for a non-binary qualitative independent variable is illustrated using two case studies. In particular, norms that account for the impact of Level of Education are derived for the Openness subscale score of the International Personality Item Pool and the Fruits Verbal Fluency Test score. The normative analyses are conducted using the NormData package in the R software.