Gender Differences in Personality and Social Behavior

The scientific study of gender differences has yielded a wealth of robust generalizations about the way males and females differ across domains, cultures, and developmental stages. This article provides a descriptive overview of gender differences in personality, social cooperation and competition (including aggression and play), and verbal and nonverbal communication. Qualitative statements about the typical behavior of males and females are supplemented with quantitative data from metaanalyses and other large-scale studies. Key methodological issues in the measurement of gender differences are discussed to aid the reader in the interpretation of research findings. The scientific study of gender differences has a long history, spanning more than a century of research (Ellis et al., 2008). Competing theories of gender vary in the role they assign to evolutionary history, cultural practices, endocrine and neurobiological mechanisms, and individual learning; they also vary in the extent to which they regard these levels of explanation as complementary or mutually exclusive. While the field is rife with theoretical debate and controversy, decades of empirical research have yielded a wealth of robust generalizations about the way males and females differ across domains, cultures, and developmental stages. This process has been accelerated by the diffusion of meta-analytic techniques over the last 30 years. Thanks to meta-analysis, the results of multiple studies can be aggregated and corrected for various sources of error; also, the effects of potential moderators can be analyzed in detail. This article provides a descriptive overview of gender differences in personality; social cooperation and competition (including aggression and play); and verbal and nonverbal communication. Whenever possible, qualitative statements about the typical behavior of males and females are supplemented with quantitative data from meta-analyses and other large-scale studies. To aid the reader in the interpretation of these findings, the initial section presents and discusses key methodological issues in the measurement of gender differences. Methodological Issues The Interpretation of Effect Sizes In psychological research, group differences (including differences between males and females) are usually expressed in terms of standardized effect sizes. The most common effect size for group and gender differences is Cohen’s d, the difference between means divided by the pooled standard deviation of the two groups (e.g., d 1⁄4 0.50 represents a distance of half a standard deviation between male and female means). Conventionally, positive values of d indicate that the male mean is higher than the female mean, whereas negative values indicate that the female mean is higher. Values of d can easily be translated into estimates of statistical overlap (Figure 1). For example, d 1⁄4 0.50 corresponds to a 67% overlap between male and female distributions (assuming normality). The precision with which a variable is measured can dramatically affect the resulting effect size, since d becomes artificially smaller as measurement error increases. Many psychological variables are measured with a substantial margin of error; thus, empirical values of d should be interpreted as lower bound estimates of the actual differences. Effect sizes that have been corrected for measurement error are explicitly flagged as such in the remainder of this article. Regrettably, many researchers routinely interpret effect sizes as ‘small,’ ‘moderate,’ or ‘large’ based on the conventional cutoffs (d 1⁄4 0.20, 0.50, and 0.80) originally proposed by Cohen (1988). This practice has little basis in statistical theory and is discouraged by most methodologists (Breaugh, 2003; Hedges, 2008; Vacha-Haase and Thompson, 2004). Ironically, Cohen himself advised against using his tentative cutoffs except as a last resort method for determining sample size in exploratory studies (Cohen, 1988). The substantive Figure 1 Relation between effect size (d or D) and percent overlap between male and female distributions. The illustrative plots on the right show two univariate distributions overlapping by 10%, 30%, 50%, 70%, and 90% of their joint area. Overlap is computed assuming multivariate normality. See Del Giudice, M., 2009. On the real magnitude of psychological sex differences. Evolutionary Psychology 7, 264–279. 750 International Encyclopedia of the Social & Behavioral Sciences, 2nd edition, Volume 9 http://dx.doi.org/10.1016/B978-0-08-097086-8.25100-3 International Encyclopedia of the Social & Behavioral Sciences, Second Edition, 2015, 750–756 Author's personal copy