Are Men Reluctant to Assault Women Even When Intoxicated?

Abstract

Alcohol intoxication leads to anti-normative behavior. Because violence against women is more anti-normative than violence against men, we suggest that the effects of alcohol on violence against women should be stronger. We found support for this hypothesis in an analysis of more than 1,100 interpersonal disputes reported by male prison inmates and male community members. We find that the tendency for men to be more willing to threaten and attack male adversaries than female adversaries is weaker when men are intoxicated. When respondents were moderately or extremely intoxicated, they are just as likely to target women. Apparently, only sober and slightly intoxicated men are inhibited about attacking women during disputes. We suggest that the effects of the chivalry norm decrease as men become intoxicated.

Keywords:

• Violence
• alcohol
• interpersonal disputes
• gender
• chivalry

Disclosure Statement

No potential conflict of interest was reported by the authors.

Notes

1 We prefer the term “chivalry norm” because it focuses specifically on a norm requiring men to protect women, whereas other descriptors (“paternalism” or “benevolent sexism”) are more general. In addition, we avoid any assumption about the origins of the chivalry norm. While some have argued that chivalry reflects sexism and patriarchy (e.g. Glick & Fiske, 1997, 2001), others have suggested that the norm may reflect an attempt to protect children by protecting mothers or an attempt to protect individuals who are physically weaker and vulnerable (e.g. Campbell, 1999; Felson et al., 2023).

2 In comparison, approximately 27 percent of property offenders, 24 percent of drug offenders, and 36 percent of public order offenders had consumed alcohol at the time of their offense.

3 Intoxication does not necessarily lead to antinormative behavior. For example, intoxication should increase the likelihood of violence when the most salient cues are provocative (e.g. threats and insults, yet decrease the likelihood of violence when cues are inhibiting (e.g. police presence).

4 It may be that some of portion of the relationship between alcohol and violence is spurious (e.g. Fagan, 1990; White, 1997). For example, it may be that people with low self-control are more likely to engage in both drinking and violence.

5 The strongest effect was observed for intimate partner violence (0.45), followed by general aggression (0.46), then sexual aggression (0.32).

6 A different question is the extent to which our estimates are affected by selection bias. This is distinct from concerns over generalizability. Selection does not necessarily result in biased estimates, and important epidemiological research has been conducted to understand when selection induces bias (e.g., Hernan et al., 2004). The fact that our sample includes community members helps mitigate concerns about the idiosyncratic characteristics of offenders.  However, while community members have never been arrested, they are associates of the offenders, so they may also engage in more anti-normative behavior than the general population.

7 A more complete description of the dataset and its sampling procedure is provided by Berg and Felson (2020) and Rogers et al. (2019).

8 Since respondents were not asked about how many drinks they consumed nor the type of alcohol, we were unable to estimate a blood alcohol level (BAC).

9 We are unable to study disputes between male intimate partners since there were only ten cases.

10 It is possible that alcohol has different effects for intimate partner violence that has a control motivation, is more frequent, or involves serious injury (e.g., Johnson, 1995; Johnson, 2008). Future research should explore this possibility.

11 The sample sizes were smaller when we conducted these analyses. There were 835 observations for the incarcerated sample and 304 observations for the community sample.

12 One reviewer suggested controlling for other substances that the respondent may have used at the time of the incident. Respondents were asked two items about their drug use: one about their marijuana use and one about any other drug use. We created a single variable from this information and controlled for it. We did not find that it changed our results (Appendix 1 and 2).

Additional information

Funding

National Institute of Justice, Grant/Award Number: 2012‐91288‐PA‐I.