Contents lists available at ScienceDirect Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid On the nonlinear association between intelligence and openness: Not much of an effect beyond an average IQ Gilles E. Gignac ⁎ , Bradley Walker, Tiarn Burtenshaw, Nicolas Fay School of Psychological Science, University of Western Australia, Australia ARTICLE INFO Keywords: Intelligence Openness Nonlinear ABSTRACT The linear association between intelligence and openness has been estimated at r ≈ 0.20 to 0.30. However, little research has examined the possibility of a nonlinear effect between the two dimensions. Consequently, task- based intelligence and self-reported openness data were collected from 371 participants (UK community sample). We found that the association was nonlinear, i.e., the positive effect was no longer observed beyond an IQ of ≈ 105. Furthermore, across the 10 openness items, four evidenced positive, linear effects with intelligence, all of which were epistemic openness items. By comparison, several experiential openness items showed inverted U-shaped effects. It is concluded that, beyond relatively low to moderate levels of intelligence, general in- telligence may be unrelated to global openness, especially if need for cognition is considered distinct from openness. 1. Introduction Cognitive intelligence has been characterized by the ability to learn, reason, think abstractly, and solve problems (Deary, 2013). By com- parison, the openness personality trait represents the degree to which a person is willing to entertain and/or enjoy a variety of cognitive and perceptual experiences (e.g., ideas, approaches to expression, un- conventionality, range of emotions, and cultural activities; Goldberg, 1992; McCrae & Costa, 1985). The correlation between general in- telligence and global openness has been reported at between r ≈ 0.20 to 0.30 (von Stumm, Hell, & Chamorro-Premuzic, 2011), suggesting the effect is somewhere between typical and relatively large for differential psychology (Gignac & Szodorai, 2016). Correspondingly, relatively more intelligent people have been suggested to possess relatively more open personalities (e.g., Haslam, 2007). Interpretations of the effect have largely assumed the association to be linear, i.e., ever greater le- vels of openness are assumed to associate with ever greater levels of intelligence. However, such an assumption may be inaccurate. In a recent review, Ackerman (2018) argued that more research should examine the possibility that one or more of the associations may be nonlinear. As a Pearson correlation can only capture linear covar- iance between two variables (Speed, 2011), previous research that has estimated the Pearson correlation between intelligence and openness cannot shed light on the possibility that the association may be non- linear. Arguably, it is important to confirm or disconfirm the linearity of an association between two variables, in the same way that the ob- servation of an interaction can potentially confound the interpretation of a main effect (Davison & Sharma, 1990). For example, if the asso- ciation between openness and intelligence is limited to the low to moderate range of intelligence, it would be unjustified to suggest that ever greater levels of intelligence are associated with ever greater levels of openness. To our knowledge, Rammstedt, Danner, and Martin (2016) is the only investigation to examine the possibility of a nonlinear effect be- tween intelligence and openness. On the basis of a large sample (N > 3000), Rammstedt et al. (2016) reported a significant Pearson correlation of r = 0.05 between intelligence and openness. Ad- ditionally, on the basis of hierarchical multiple regressions, they also reported quadratic openness effects of β = -0.03 and β = -0.04 with verbal and numerical ability, respectively. Although only the numerical ability quadratic effect was significant, the negatively directed stan- dardized beta-weights suggested an inverted U-shape trend between intelligence and openness (no scatter plot was reported). Thus, beyond a certain level of intelligence, the association between intelligence and openness may not exist. Although the results of Rammstedt et al. (2016) offer some insights, we note that they measured intelligence with only two subtests, literacy and numeracy. Additionally, openness was mea- sured with only three items. Thus, in light of results reported by Rammstedt et al. (2016), it was considered useful to test for a nonlinear effect between intelligence and openness using more comprehensive https://doi.org/10.1016/j.paid.2020.110169 Received 4 March 2020; Received in revised form 1 May 2020; Accepted 2 June 2020 ⁎ Corresponding author at: School of Psychology, University of Western Australia, 35 Stirling Highway, Crawley, Western Australia 6009, Australia. E-mail address: gilles.gignac@uwa.edu.au (G.E. Gignac). Personality and Individual Differences 166 (2020) 110169 0191-8869/ © 2020 Elsevier Ltd. All rights reserved. T