As I posted a couple of months ago, a higher level of violence in a society may lead women to prefer more masculine appearing men. In such an environment, picking the healthiest appearing male is more important than the level of parental care the woman expects the man to give.

The latest issue of Evolution and Human Behavior has an article examining the link between female preference and violence, with Jeffrey Snyder and colleagues examining whether a woman’s fear of crime might be a predictor of her preference for “aggressive and formidable” mates. Unlike earlier research, which focused on actual violence levels, Snyder and colleagues' targeted their hypothesis at the woman’s perception of her vulnerability to crime. This makes sense, as the need for an aggressive man is likely to be a function of both the level of crime and the woman’s ability to defend herself. The woman’s perception of her vulnerability should capture both of these elements.

Snyder and colleagues also framed the trade-off around outcomes to the women instead of reproductive outcomes. Instead of asking whether the aggressive man will deliver a healthy child or invest in parental care, the trade-off discussed concerned violence to the woman by the aggressive man versus the protection he can offer the woman.

Using three internet based studies of United States women, Snyder and colleagues supported their hypothesis through the discovery of a positive relationship between a woman’s perception of her vulnerability to crime and her preference for aggressive men. However, there was no or a very weak link between female mate preference and actual crime rates (which were determined by zip code). This is somewhat confusing, as it suggests that fear of crime may not be rationally based. Snyder and colleagues' hypothesis would predict a weaker link between actual crime and preferences than between perceived vulnerability and preferences, but there should still be a link.

A further issue with the results is the low-level of effect that the perception of crime has. While it comfortably passes the significance tests, fear of crime can only explain (at most) 5.5 per cent, 7 per cent and 6 per cent of the variation in preferences for aggressive men across the three studies (and that is including other variables in some of the regressions including race, age, education and inequality). This suggests that while perceived vulnerability is significant in a statistical sense, it has very little predictive power.

Having mulled on this study for a couple of days, I am not sure what to make of it. I find the hypothesis attractive, but the absence of a link to actual crime leaves me with a large number of questions about the survey methods used and suggestions for follow-up research - a number of which were also noted by Snyder et al.

First, it would be useful to get some more variation in the sample. The study participants were highly educated, with less than ten per cent of the sample in each study having a level of education at high school level or below. As a result, there are likely to be very few in the sample who live in a violent area. This variation may be particularly important if the survey subjects are subject to levels of crime too low for protection from aggressive men to matter.

More variation could be introduced by including other countries or particularly high crime areas. In those countries or areas, a more masculine male may deliver much stronger benefits and be more strongly preferred.

A related observation is that for this study’s well-educated participants, the easier way to avoid crime would be to marry a rich man. As a result, women in this sample might want to avoid aggressive men. However, that may not be a feasible choice in Sudan or for some inner-city residents. Are these preferences stronger where the best response to violence is a violent response?

Second, and as suggested in the paper, finding out what these women actually do as opposed to their survey responses would be useful. Apart from the obvious benefits to seeing revealed preferences, this might also help to calibrate the responses. In each survey, the women were asked their responses to characterisations such as “bad-boy”, “broad shouldered” and “strong”. If a well-educated woman thinks an accountant with a motorcycle is a bad boy, that is probably a different level of masculinity compared to someone seeking physical protection from a real risk of crime.

I would also like to know more about the women in the survey. Are they in a partnership or married? How tall are they? Have they been a victim of crime? Does their fear of violence come from in the home or from people they know? How much property do they have which could be appropriated in a crime? This might help find some explanatory variables with some real predictive power. However, to test the basic hypothesis, we need a sample with more variation in the levels of violence and ideally, a sample in which we can observe real choices.

I don’t believe that this story sheds much light on my earlier ruminations on violence (most recent here) and the importance of a shift away from violence to allow characteristics such as hard work, intelligence and patience to be rewarded and spread through the population. It could be argued that as the study was conducted in a developed country, and among educated women in that country, we would expect violence to be a trait associated with low fitness. You would expect women to generally favour other traits, with the aggressive characteristics to be secondary and only accepted if they do not come at the cost of economically important traits. Again, to test this idea, we require more variation in the sample. We would need some of the sample to come from populations in which crime brings benefits to its purveyors and results in reproductive success.