No differentiation of personality by intelligence: consistently unidimensional factors’ variances and internal consistencies do not differ by intelligence

We tested the Differentiation of Personality by Intelligence (DPI) Hypothesis and overcame two limitations of prior studies. Prior studies typically examine personality across groups created by splitting the sample using the median/tertile intelligence values. In this study, we examine a range from 2-10 intelligence subgroups. Furthermore, prior studies have assumed that factors are consistently unidimensional across intelligence subgroups. Assuming factorial structures can be problematic, as metrics in question (e.g., Cronbach’s α) work on the premiss of unidimensionality. We therefore heeded factorial structures. We used personality (High Potential Traits Inventory) and verbal intelligence (the Word Meaning subtest of the General Intelligence Assessment) data from 6038 respondents. Results showed no consistent DPI pattern for variances, while internal consistencies increased with higher intelligence. Furthermore, factorial structures varied across intelligence subgroups. Heeding the factorial structure by analyzing only consistently unidimensional subfactors, the before observed effects on internal consistencies disappeared. We conclude that a) there is no support for DPI for scale variances and b) that prior findings testing the DPI for internal consistencies may be an artifact driven by not considering factorial structures. On a practical note, we caution that personality scales may not be consistently unidimensional across subgroups varying in their verbal intelligence.