Measuring gender differences in partner violence: implications from research on other forms of violent and socially undesirable behavior

Sex Roles: A Journal of Research,  June, 2005  by Sherry L. Hamby

• E-mail
• Print
• Link

Once again it is helpful to go beyond the partner violence literature to see if partner violence reports are an aberration from other data on gender differences in validity. In some cases, researchers are seeking to explain similar phenomena. For example, in a pattern that closely parallels partner violence data, sex differences are less marked for many self-report measures of juvenile delinquency than they are for official records of arrests for delinquent behavior (Hindelang et al., 1979). Although some authors (e.g., Archer, 2000) minimize the importance of differential validity, it can present a problem in surprising places. Even biological markers can suffer from differential validity. For example, it has been demonstrated that different methods for measuring testosterone are highly correlated for men, but not for women (Shirtcliff, Granger, & Likos, 2002).

Differential Validity for Gender in Self-Report Data

The limited empirical research on whether violence measures are differentially valid for men and women has yielded mixed results. Some studies have found no gender differences in over- or underreporting (Czaja et al., 1994; Huizinga & Elliott, 1986). Gender did not interact with methodological manipulations for reports of psychological aggression, physical assault, and injury in one study (Hamby et al., 2001), but sexual coercion produced greater gender differences for dichotomous categories (47% male and 3% female perpetrators) than standard CTS2 frequency categories (26% male and 22% female). In a direction contrary to most hypotheses, men reported both more offending (compared to arrests records) and less victimization (compared to substantiated reports) than women in one sample of known abuse victims (Maxfield et al., 2000; Widom & Morris, 1997). Widom and her colleagues suggested that identifying as an offender may be more stigmatizing for women, and identifying as a sexual abuse victim may be more stigmatizing for men.

The Effect of Item Content on Gender Differences

Although the empirical database on gender differences in validity is sparse, research on delinquency, bullying, and other aggression show that including a wider array of less criminal or physically aggressive acts will not only increase the total number of people who endorse one or more items but will also make the gender ratios more nearly equivalent. For example, nonviolent delinquent acts, such as running away and underage alcohol use, are reported by fairly similar numbers of males and females, whereas assaults are reported by many more males than females (Hindelang et al., 1979). Similarly, boys perpetrate more physical bullying than girls, but expanding the construct of bullying to include emotional aggression will greatly increase the number of girls identified as offenders (Hamby, Ormrod, & Finkelhor, 2004; Salmivalli & Kaukiainen, 2004). Even in crime data, the percentage of female perpetrators is higher for simple assault than for aggravated assault or murder (FBI, 2003; Rennison & Rand, 2003). It has been known for more than two decades that "sex differences in self-report data are highly contingent on item content" (Hindelang et al., 1979, p. 998).