Women are somewhat better at divergent thinking, while men show greater variability, meaning there are more likely to fall in the extremes (& previous research say that this is hardly related to creative performance)

Abdulla Alabbasi, A. M., Thompson, T. L., Runco, M. A., Alansari, L. A., & Ayoub, A. E. A. (2022). Gender differences in creative potential: A meta-analysis of mean differences and variability. Psychology of Aesthetics, Creativity, and the Arts. Jul 2022. https://doi.org/10.1037/aca0000506

Abstract: The current study examined gender differences in divergent thinking (DT) using meta-analyses of mean difference and variation. The main objective of the meta-analysis of mean difference was to resolve contradictory findings in the creativity literature regarding the prevalence of creativity among males or females in creative potential. The meta-analysis of variation aimed to test the greater male variability hypothesis (GMVH) in DT. To test gender differences in means (i.e., Hedges’ g), results from 213 studies (k = 1,251; N = 115,289) were analyzed using a three-level approach. Females slightly outperformed males in DT, g = −.065, 95% CI [−.095, −.034], p = ≤ .001. Three-level multiple regression analyses showed that the mean effect size significantly varied by (a) country, (b) DT subscale, (c) type of task, and (d) ability (gifted vs. nongifted). In the second meta-analysis, the GMVH in creative potential was tested by synthesizing the results of 1,152 effect sizes from 187 studies (k = 1,152; N = 101,328). The results confirmed the existence of greater male variability (GMV) in DT, (InVR) = 1.216, 95% CI [1.14, 1.29], p ≤ .001, indicating 21.6% GMV in DT. Multiple regression analyses explained 29.82% of variability in the mean effect (InVR) at Level-2 (within-studies variance), and 5% of the variability in the mean effect at Level-3 (between-studies variance). The mean difference findings support the gender similarity hypothesis, while variation results tend to support the gender differences hypothesis. Limitations and recommendations for future studies are discussed.

Discussion

Gender Differences in Means

Although seminal review research on gender differences in mean DT scores supported the gender similarity hypothesis (e.g., Baer, 2012; Baer & Kaufman, 2008; Kogan, 1974; Runco et al., 2010), an empirical quantitative investigation was warranted. The current results show a small effect size (Cohen, 1988), favoring females. Interestingly, previous findings indicated that females mostly outperformed males in DT abilities (Thompson et al., 2021), yet males have shown a slight advantage in terms of creative performance (Hora et al., 2021). Such a pattern of findings suggest that females may initially show greater creative potential, but that males are able to apply their potential more fruitfully in terms of achievements.

As expected, the overall mean effect size showed high heterogeneity, requiring moderator analysis. Moderator selection was based on possible sources of variability identified by previous meta-analyses in creativity (Abdulla Alabbasi et al., 2021; Abraham, 2016; Acar & Runco, 2012; Acar, Runco, & Park, 2020; Baer & Kaufman, 2008; Paek et al., 2021; Reiter-Palmon et al., 2019; Runco et al., 2010; Said-Metwaly et al., 2020; Kogan, 1974; Thompson et al., 2021). These included year of publication (Baer, 2012; Mar’i & Karayanni, 1983), cross-cultural comparisons (Shao et al., 2019; Storme et al., 2017), age (Cheung & Lau, 2010; Palmiero et al., 2014), DT test (Runco et al., 2016), DT subscale (Karwowski et al., 2016), type of task (Abdulla Alabbasi et al., 2021; Taylor & Barbot, 2021), and ability (Abdulla Alabbasi et al., 2021; Runco, 1993). The three-level multiple regression analysis showed that Level-2 and Level-3 together explained 61.6% of the total variance. The multiple regression analysis indicated that the mean effect size significantly varied by (a) country, (b) DT subscale, (c) type of task, and (d) ability. The smallest gender difference in DT was observed in Asian countries, while slightly larger differences were observed in the Middle East, the United States, and Canada. This finding is consistent with that of some studies on the differences in intellectual abilities across different cultures (Feingold, 1994; Gray et al., 2019; He et al., 2013); however, there was a small difference in DT between participants in different countries (less than g = .10; −.085–.013).

Regarding DT subscales, the results showed a significantly larger mean effect size for fluency, in favor of females, while the smallest gender difference was observed for originality. This is one of the most interesting and significant findings, given that originality is the central feature of creativity (Acar et al., 2019; Runco & Jaeger, 2012). Females also scored higher than males in the composite DT score. Although the composite score is a useful index, it might be misleading since it offers an incomplete assessment of individual differences in creative potential. For educators, the emphasis should be on originality more than any other DT index, first because it is an essential element in any creative work or behavior (Runco, 2014) and, second, because the current findings showed no significant gender difference in originality, unlike fluency, which may be biased against males.

Another important consideration for educators is the type of task used in DT tests when assessing students’ creative potential (e.g., to identify gifted students). The current finding is consistent with some previous studies reporting that females outperform males in verbal tasks (Abraham, 2016; Halpern et al., 2007). This is not to say that educators (or researchers) should avoid using verbal DT tests; in fact, we believe that both task types elicit unique information about an individual’s creative potential. Our recommendation is that both verbal and figural tasks be used to screen students for special programs (e.g., gifted student programs), as using both mediums seem to capture a fuller spectrum of gender strengths. Finally, the comparison between gifted and nongifted samples showed that both gifted females and nongifted females outperformed gifted males and nongifted males, although the magnitude of the effect size was larger in gifted females than gifted males, supporting some previous giftedness and DT findings (e.g., Abdulla Alabbasi et al., 2021; Bahar & Ozturk, 2018).

Gender Differences in Variability

The GMVH was initially proposed as a possible explanation for greater male superiority in different cognitive domains throughout human history (see Feingold, 1992 for a historical review). Several meta-analytic reviews of GMVH in intellectual abilities supported greater male variability in most of the tested intellectual abilities (Feingold, 1992; Gray et al., 2019; Hedges & Friedman, 1993). For creativity, studies on gender differences in variability have been conducted in both Eastern (He & Wong, 2011; He et al., 2013; Ju et al., 2015; Lau & Cheung, 2015) and Western cultural contexts (Karwowski, Jankowska, Gajda, et al., 2016; Taylor & Barbot, 2021), and one study was conducted with an African sample (Karwowski Jankowska, Gralewski, et al., 2016). These investigations attributed different findings to cultural differences (e.g., He et al., 2013), type of task (e.g., Taylor & Barbot, 2021), and the obtained or reported variance ratio (VR). For instance, whereas He and Wong (2011) reported a VR of 1.62 for the composite score of the test for Creative Thinking-Drawing Production (TCT-DP), He et al. (2013) reported a VR of 1.30, Ju et al. (2015) reported a VR of 1.06, and Karwowski, Jankowska, Gralewski, et al. (2016) reported a VR of 1.82. However, earlier studies were limited in terms of the assessments used (all used the TCT-DP except Lau & Cheung, 2015; Taylor & Barbot, 2021), type of task (all used figural tasks except Lau & Cheung, 2015; Taylor & Barbot, 2021), the capacity for cultural comparisons, and sample size. By retrieving the raw data from 187 studies, we were able to calculate the (InVR) of a sample of 101,328, providing a clearer picture of GMVH in creative potential. Moreover, we were able to test different moderators (see Tables 4 and 5) to explain the high heterogeneity observed in the mean effect size. The mean effect size obtained in the current study was less than most previous studies (except Ju et al., 2015). Note here that a VR between .90 and 1.10 indicates a small effect size, while a VR greater than 1.10 would indicate GMV (Karwowski, Jankowska, Gralewski, et al., 2016; Lau & Cheung, 2015). The major findings from the moderator analysis were: (a) greater male DT variability was observed in verbal tasks (InVR = 1.249); (b) among DT subscales, GMV was observed in the elaboration subscale (InVR = 1.429); and (c) among DT tests, a GMV was observed in Wallach and Kogan’s tests (InVR = 1.316).

First, regarding type of task, the current findings differ from previous meta-analyses on GMVH in other cognitive abilities. For instance, Feingold (1992) reported that males and females did not differ in verbal tests such as short-term memory (STM), abstract reasoning, and perceptual speed, whereas a large male variability was found in mechanical reasoning, for example. Hedges and Nowell (1995) reported a negligible difference between males and females in the VR for vocabulary (VR = 1.00–1.08) and reading comprehension (VR = 1.03–1.16), compared with spatial ability (VR = 1.27) and mechanical reasoning (VR = 1.45–1.74). The current finding (i.e., GMV in verbal DT) is also inconsistent with Lau and Cheung (2015), who concluded that GMV was supported in figural tasks, while not much variability was observed in verbal tasks. Greater male variability in elaboration is one area that deserves further future investigation, given that none of the previous studies on GMVH in creative potential targeted or assessed elaboration. The same is true for differences in variability between DT tests, which was not tested before.

Limitations and Future Directions

The limitations of meta-analyses often originate in the limitations of the primary studies. First, DT tests are not synonymous with creativity. Studies show mixed evidence on DT tests’ predictive validity, with some suggesting test scores are unrelated to real-world creative achievement (Baer, 1993), and others suggesting that they account for up to half of the variance in creative achievement (Plucker, 1999). Nevertheless, DT tests have consistently been the most popular way to measure creativity (Abdulla & Cramond, 2017; Plucker & Makel, 2010), and thus, synthesis of these results continues to be a useful metric of the state of the creativity literature.

The overall sample of the meta-analysis was also limited because creativity research has tended to be conducted more often with youth rather than adults. The mean age of the overall sample was 13.91, and only 25.4% of the included studies consisted of participants above the age of 18 (see Figure 3).

Fig 3

As such, there was an age ceiling that limits the generalizability of these findings. This is important because there is some evidence indicating that DT increases with age (Fusi et al., 2021; Shah & Gustafsson, 2021), at least up to middle-age (until about 40 years-old; Massimiliano, 2015; Reese et al., 2001). The creativity literature overall would benefit from extending data collection to older samples to gain a better understanding of life span creativity.

Additional steps to diversify primary samples would further improve the generalizability of future meta-analyses of DT data. Though the current study attempted to emphasize cultural variability and included some studies in Arabic, the sample still primarily comprised Western, English-speaking participants. Similarly, few to no studies allowed participants to self-identify outside of the female-male binary. Future investigations from more diverse countries, including those speaking different languages, and with data collection that allows for the representation of nonbinary gender identities would provide richer data.