This study explored the latent structure of divergent thinking as a cognitive ability across gifted and non-gifted samples of students utilizing multiple-group analysis of mean and covariance structures (MG-MACS). Whereas Spearman's law of diminishing returns postulates lower g saturation of cognitive tests with increasing ability level and consequently, a lower correlation of cognitive abilities in more gifted samples, recent evidence from creativity research has shown that correlations of divergent thinking with intelligence are unaffected by ability level. In order to investigate this conflicting state of affairs with respect to divergent thinking, we utilized increasingly restrictive MG-MACS models that were capable of comparing latent variances, covariances, and means between gifted (IQ > 130) and non-gifted (IQ ≤ 130) groups of students. In a sample of 1070 German school students, we found that a MG-MACS model assuming partial strict measurement invariance with respect to the postulated factor model of verbal, figural, and numerical divergent thinking could not be rejected. Further, latent variances and covariances of latent divergent thinking factors did not significantly differ between groups, whereas the gifted group exhibited significantly higher latent means. Finally, implications of our results for future research on the latent structure of divergent thinking are discussed.
Key words: Giftedness; Confirmatory factor analysis
Many structural theories of intelligence incorporate a factor corresponding to creativity (e.g., Carroll, 1993; Jäger, 1984). Divergent thinking (DT), which has been defined as the capability to generate diverse and numerous ideas (Runco, 1991), can be considered as the core ability for creative achievements. In a classical article, Guilford ( 1950) identified three basic components as factors of DT: Fluency (the total number of ideas generated), flexibility (the number of categories in the ideas) and originality (the number of unique or unusual ideas). However, fluency is usually described as the central component of DT (Hargreaves & Bolton, 1972). In contrast to research on intelligence, DT tests reported in the literature focus on verbal or figural content, thereby neglecting the numerical domain (Cropley, 2000). However, numerical content plays an important role in research on reasoning and problemsolving, where DT is often of central importance (Mumford, Connelly, Baughman, & Marks, 1994). Further, Livne and Milgram (2006) have shown that DT is one important facet of mathematical achievement. Hence, an investigation of numerical DT, and its relationship with verbal and figurai DT, seems necessary to elucidate the factorial structure of DT.
The concept of intellectual giftedness has been defined in different ways across the literature. Some approaches (e.g., Roznowski, Reith, & Hong, 2000) focus exclusively on high intellectual ability (g) as the sole determinant of intellectual giftedness, while others (e.g., Lubinski & Benbow, 2000) perceive giftedness as being multidimensional in nature. The role of creativity (and hence, DT) in models of giftedness varies as well, where some models of intellectual giftedness perceive creativity as a condition sine qua non for outstanding intellectual achievement (Renzulli, 1986), while others perceive creativity as an own form of giftedness (Gagné, 1993). Similar to Roznowski et al. (2000), we take a one-dimensional perspective on intellectual giftedness in this paper, in that subjects with a high level of fluid intelligence are defined as being intellectually gifted. Further, we assume that DT, as a core trait of creative performance, can be conceptualized as a latent cognitive ability that is part of a cognitive taxonomy (Carroll, 1993).
The empirical relationship of DT with intelligence has been intensively researched over the years (cf. Haensly & Reynolds, 1989; Sternberg & O'Hara, 1999). …