Back to the basics: Abstract painting as an index of creativity

ABSTRACTResearchers have invested a great deal in creating reliable, “gold-standard” creativity assessments that can be administered in controlled laboratory settings, though these efforts have come at the cost of not using ecologically and face-valid tasks. To help fill this critical gap, we developed and implemented a novel, face-valid paradigm that required participants to paint abstract pieces of art, which were later rated for creative quality. We first sought to evaluate whether there was good convergence among creativity ratings provided by independent raters. Next, we examined whether its measure of creativity correlated with (a) existing creativity measures and (b) individual traits (e.g. openness, fluid intelligence) that are typically correlated with indices of creativity. Our findings indicate that our abstract-painting paradigm is feasible to implement (independent ratings of the creativity of the paintings converged well), and that its measure of creativity significantly correlated with some of the gold-standard indices of creativity (thereby providing convergent validity). These findings suggest that having participants engage in abstract painting provides a valid index of creativity, thereby opening new opportunities for future research to index a more-face-valid measure of creativity. Disclosure statementNo potential conflict of interest was reported by the author(s).Declaration statementThe authors report there are no competing interests to declare.Notes1. We also explored how ratings of an expert (i.e., a professional painter) on Painting Creativity scores would affect results. With expert ratings, κ increases from .37 to .38, and EAP reliability increases from 0.78 to 0.81, indicating consistency between the two groups of raters. Before and after addition of expert ratings, rater severity remained similar ([−0.17, 0.67] to [−0.36, 0.61]) and range of scores remained similar ([−1.90, 1.68] to [−2.09, 1.60]). Importantly, correlation with the AUT remains non-significant, from r = .077 to r = .093.