A Multilevel Analysis of Associations Between Children's Coloured Progressive Matrices Performances and Self-Rated Personality: Class-Average and Class-Homogeneity Differences in Nonverbal Intelligence Matter.

The relationship between self-rated personality and nonverbal intelligence has been studied in young students, but these studies have generally not considered nested data, despite their allowing us to analyse between-classroom variability. The present cross-sectional study involved third- to sixth-grade students (n = 447) who were nested into their classrooms (n = 32). The participants completed the Raven's Coloured Progressive Matrices (CPM) as a measure of nonverbal intelligence and a personality questionnaire based on the Five Factor Model. At the class level, the study data included class size, class-average CPM scores, and class-homogeneity in CPM performances. Multilevel modelling with class-mean centring of personality predictors was applied to examine class-average differences in CPM scores and interaction effects between personality and class-homogeneity on CPM scores. The results showed significant differences in average CPM performances across classrooms, significant fixed and random slope effects linking nonverbal intelligence and Imagination, and a cross-level effect revealing that Imagination is a stronger predictor of CPM scores when class-homogeneity in intelligence is lower. Beyond confirming the intelligence-Imagination association generally observed in the literature, the present findings emphasise the importance of using nested structures when collecting personality and intelligence data in classrooms. More attention needs to be paid to how the classroom environment affects children's self-reported personality and intelligence test performances.