A new study published by Chiao et al. in the journal PLoS ONE explores the gendered nature of American voting behavior. Subjects were asked to rank politicians — based only on photographs of each politician’s face — along different quality scales, and also to choose among these photographs who should be President. The study concludes that male and female candidates are evaluated on distinctly different terms, and that male and female voters do this evaluation in somewhat (but not dramatically) different ways. The authors conclude that “…contrary to popular notions, people are not necessarily using deliberate and rational strategies in deciding who to vote for, especially when it comes to (voting for) women.” Well, duh. I think we already knew that. But despite the obvious naivete of the researchers in this particular statement, their study is still interesting and well done, and I’d like to explore it a bit further.
Let’s start with a summary of the paper’s conclusions, then explore different ways of looking at this.
Male and female voters judged a series of male and female political candidates on how competent, dominant, attractive and approachable they seemed based on their facial appearance. Then they saw a series of pairs of political candidates and decided which politician they would vote for in a hypothetical election for President of the United States. … All voters are likely to vote for candidates who appear more competent. However, male candidates that appear more approachable and female candidates who appear more attractive are more likely to win votes. In particular, men are more likely to vote for attractive female candidates whereas women are more likely to vote for approachable male candidates.
The authors carry out the appropriate statistical analyses, and can draw these conclusions with reasonable certainty. For instance, the following table highlights which relationships are significant (with asterisks).
You can see that male and female subjects generally regard competence as important in both male and female candidates, everyone likes attractive females and approachable males, but there is some difference between males and females in the assumed strength of these preferences. Here, by “strength” we mean mean statistically at the group level (how many subjects will follow the pattern as opposed to acting in a different way), not how strong the feelings are.
In carrying out this analysis, the researchers used photographs of actual members of congress, and the subjects evaluating them were college students. Preliminary analysis allowed the researchers to remove members of congress who are likely to have been recognized by the test subjects. One great advantage of using actual members of congress is that the researchers were able to obtain voting data to see if the test subjects’ preferences correctly predicted voting outcome. It does. It turns out that perceived competence and dominance (by the subjects in this study) were good predictors of actual election outcomes.
… In particular, perceived competence significantly predicted actual election outcomes for male candidates … but not female candidates. … Taking both gender of voter and gender of candidate into account revealed divergent kinds of facial inferences that predicted actual election outcomes. Perceived competence of male candidates by male voters … predicted actual election outcomes for male Congressional candidates. However, facial inferences by male and female voters did not significantly predict actual election outcomes for female Congressional candidates.
So, to summarize, subjects of both sexes value competence and can identify it (they think) from looking at faces; Both male and female subjects like attractive females and approachable males, but males are more oriented towards attractiveness in females and females are more oriented towards approachability in males. And, even though this was a group of college students looking at pictures, the values attributed by these students to these elected officials corresponded to actual election outcomes.
So what does this all mean? The researchers examine a set of different kinds of explanations.
The authors consider the so called “thin slice theory” which would essentially assert that the voters are accurately assessing the competence of the politicians from facial appearance and nothing else. Everyone is behaving optimally and rationally.
This is extremely unlikely, however, because (as the author’s point out) there is abundant evidence that females are better leaders than males, even though voters more typically choose males. In some studies, males and females are shown to have the same levels of effectiveness and competence in a wide range of leadership roles, even though males and females in such roles tend to gravitate towards somewhat different styles of leadership. Furthermore, where comparisons can be made, female national leaders outperform male leaders in several ways. There have been virtually no syphilitic corrupt female heads of state, for instance. In places like India, which have both female leaders now and then (most countries have none) and varying degrees of corruption, the female leaders are far less corrupt than the male leaders. And so on.
The authors alternatively suggest that the observed pattern could be explained from the perspective of “social role theory.” Essentially, individuals are in roles determined by society, members of society are accustom to this, and this customary attitude is simply applied to new cases in some predictable way. This is a reasonable explanation and can’t really be ruled out from the study at hand.
[The]…present findings … indicate that gender stereotypes predispose us to value divergent qualities in leaders, such as attractiveness in female politicians and approachability in male politicians… Although impressions of competence from facial appearance are ubiquitously predictive of voting behavior, both male and female voters are more likely to vote for female politicians who not only appear competent, but also attractive. Moreover, female voters are more likely to vote for male politicians who not only appear competent, but also approachable.
The authors examine a few other explanations, and eventually propose what is probably their preferred choice:
A third … explanation … based on evolutionary theory is that people automatically evaluate faces using a core constellation of intuitive heuristics critical for other kinds of adaptive decision-making, such as mate selection. … men and women value different qualities in heterosexual mate selection. … men are more likely to prefer women who are physically attractive, whereas women are more like to prefer men who have high social status or demonstrate the ability to garner resources … We suggest that both male and female voters value physical attractiveness in female but not male politicians … [attractiveness] engenders a broader cultural expectation … of high social status roles…. Similarly, female voters value not only competence but also approachability in male politicians due to the importance of qualities such as kindness and warmth in female selection of male long-term partners.
So, what is right and wrong with this picture? The argument given here is an adaptive augment, which assumes that natural selection has shaped humans to be good at certain things, and to respond to certain circumstances in ways that either enhance fitness or avoid diminishing fitness. There are two dramatically different ways in which such models are usually described: The Evolutionary Psychology model (domain specific functions) and the Darwinian Psychology model (general adaptive intelligence). The former would specify that very complex problems can be reliably solved by humans because fairly specific behavioral modules have been shaped by natural selection. These modules are capable of performing some pretty impressive calculations regarding social relations, risk, etc. We are born with these mechanisms, though they must have a proper environment for development. These mechanisms are domain specific in that the same exact logical problem presented in dramatically different contexts are not equally well solved.
In contrast, the latter model suggests that there are few, if any, domain specific ‘modules’ built into the human mind, but rather, a general problem solving capacity that has also been shaped by natural selection but with much less specificity.
There are arguments in favor of both models, but the domain specific module model is to me very questionable and almost certainly wrong. But either way, there is the presumption, not entirely unreasonable, that an evolutionary heritage will affect the way humans react to their environment and make decisions, and it is reasonable to expect that this behavioral repertoire would be manifest in somewhat different ways for males vs. females.
Actually, I was not particularly surprised to find that the male and female subjects in this particular study acted in very similar ways, if in fact choosing a leader was an adaptive behavior. Both males and females would end up with a similar leader, and have requirements (in a leader) that are more group than individually oriented. There should be measurable but not enormous differences between males and females in the preferred qualities of a leader, from an evolutionary perspective, if males and females have lived for a long time in the same mixed sex groups.
However, one could question the validity of the whole choosing a leader thing as an evolved behavior. Humans have probably always lived in groups, but have rarely lived in groups which elected leaders. We have a pretty good idea of the range of leader types and how these leaders get to be in these roles across a wide range of societies, which probably approximates the range of variation extant for a good chuck of our recent evolutionary history.
As the authors of this study point out, traditionally, men are the ‘leaders.’ It is true that across a wide range of societies, men are in fact playing this role. Among hunter gatherers, if there is a leader it is a man, but the role of this male ‘headman’ is usually very vague and weak, and he is responsible for virtually no actual decisions. The headman, where such a thing exits, is generally a liaison with outside groups. In food producing societies (and some human groups have histories as food producers that go back many thousands of years) the leader, again, is always male but may attain this status through a number of processes. Among the Yamomamo, fierceness is key. A man, to be a leader, must be Wawateri (fierce) and this is usually obtained by killing one or two other men (for the appropriate, honorable reasons, of course) somewhere along the line. Yet, Wawateri Yanomamo are not murders. Most leaders have a history, early in their lives, of killing other men in proper warfare or as a socially acceptable vengeance killing, but for their latter years, their resume is full of well negotiated deals, peace keeping, and problem solving with a good mix of saber rattling but no actual deaths.
Among other groups, such as some Highland New Guinea peoples, men rise to leadership roles if they can a) make good deals with other men, and women, to garner resources needed to engage in effective economically based battles with their neighbors, b) can give a really good public talk which organizes the people of his villages, and c) has demonstrated both bravery and wisdom.
It is hard to say what features would make a woman an ideal or at least preferred leader in traditional societies, but women who play the more poweful roles in at least some forager societies are sometimes known for their special abilities as healers or organizers.
In any event, there are no traditional societies that I know of in which the following two things are generally true: 1) People vote for a leader to whom they will delegate a large number of key decisions and b) People would ever have a leader that they did not already know pretty well personally.
This leads me to consider the idea that the subjects in this study were demonstrating an interesting Darwinian glitch. Being asked to pick a leader (even if only hypothetical) from a photograph is a little like, for example, picking a mate for life from hearing the person’s name. That would never happen. The whole idea of a democratically elected leader that influences your life in many ways that you have never met is, from an evolutionary or Darwinian psychological point of view, so out of the ordinary that the vote for the photo experiment is not a valid direct test of an evolved mechanism, or even of a more general adaptive intelligence. These subjects, raised in Western society, are reacting to a test situation (something they are probably rather used to) using the tools of modern pop culture to make a snap decision. These tools involve having casual opinions about often fictitious people, or at least people you’ve never met … like pop stars and, for that matter, non-existent entities such as cartoon characters and superheros. The fact that among these youngish American students some of the guys picked the next president of the United States on the basis of relative hotness is unsurprising. The good news here is that these young, college age kids hardly ever actually vote, so we are not in as much trouble as a society as it might seem…
In the end, there is good news. There is evidence that increased exposure to female politicians reduces the effects of gender stereotypes when people evaluate potential leaders. The findings of this paper demonstrate that gender biases exist, but the link between these biases and our evolutionary history are tenuous at best, and it would seem that society is evolving as we speak to increasingly disregard gender when evaluating leadership effectiveness.
Joan Y. Chiao, Nicholas E. Bowman, Harleen Gill (2008). The Political Gender Gap: Gender Bias in Facial Inferences that Predict Voting Behavior PLoS ONE, 3 (10) DOI: 10.1371/journal.pone.0003666
Ambady N, Rosenthal R (1992) Thin slices of expressive behavior as predictors of interpersonal consequences. Psychological Bulletin 111(2): 256-274.
Eagly AH (1987) Sex differences in social behavior: A social role interpretation. Hillsdale, NJ: Erlbaum.