A politician you don’t know approaches you on a market. Due to lack of time, you do not take the leaflet handed to you and return to your shopping. If you feel like you have guessed which political side he or she belongs to, you have used a stereotype, beliefs about the personal attributes of a social group. Don’t blame yourself, it’s human, it’s not necessarily disconnected from reality, and any social group is likely to be the subject of it, whether defined by its age, its gender, its profession , its geographical origin, or even its university sector.
Politicians are no exception. We were even able to show that people were able to classify them as left or right simply by looking at their photograph. The results were observed in the United States, Switzerland and Germany. We would therefore be able to identify political affiliations based on appearance.
But how do we go about it? The question is twofold. On the one hand, it is a question of determining the clues that we identify and which give us an indication of political affiliation. On the other hand, it is a question of understanding how we treat them to arrive at a diagnosis of the political affiliation of the person we have seen. The study that we published in the Journal of Economic Behavior & Organizationc the economist Carmelo Licata tries to offer some answers.
Are stereotypes statistical?
Two main families of theories describe this process. The first assumes that people are rational and follow statistical principles. According to her, people form beliefs that reflect all the observable characteristics of the groups considered. They should then, on average, correctly infer group membership from all available cues.
The second family of theories, proposed by Nobel Prize winner Daniel Kahneman and his co-author Amos Tversky, assumes that people use simplified reasoning, heuristics, in this case the representativeness heuristic. According to this heuristic, people focus on those characteristics whose frequencies differ most across groups. This heuristic simplifies the judgment problem but can lead to overreactions. For example, if the proportion of left-wing MPs with a mustache is higher than the proportion of right-wing MPs with a mustache, people will infer that any MP with a mustache is left-wing, even though there are few deputies with a mustache even on the left and that certain deputies on the right wear a mustache.
[Plus de 85 000 lecteurs font confiance aux newsletters de The Conversation pour mieux comprendre les grands enjeux du monde. Abonnez-vous aujourd’hui]
To try to decide between these theories, we used the result of an online game in which participants had to classify French deputies on the left or on the right based on a photograph.
Online riddles
In January 2012 journalists and developers competed to design applications that could be used by journalists during the “Hack the Press 2” event. An anonymous developer known by the nickname “wax-o” created a website on which Internet users had to sort anonymous photographs of real French MPs left or right. The site has gone viral. The database, which could be freely downloaded, ended up containing almost 70,000 observations.
The photographs came from the French Parliament website and represented the 554 deputies in office from 2007 to 2012, 333 from the right and 221 from the left, 447 men and 107 women. No personal information was provided. Internet users therefore had to rely on their stereotypes.
We supplemented the wax-o site with an online survey. In addition to guessing the political affiliations of anonymous photographs, respondents were asked to rate MPs on a scale of one to five in terms of their attractiveness, competence and trustworthiness. Before taking the test, they had to specify their gender and their political orientation – left, right, or undeclared.
Respondents correctly classified more than 50 percent of the photographs and did better than deciding randomly, regardless of the sample considered. They were therefore able to extract useful information from anonymous photographs, like their American, Swiss and German counterparts before them.
Left or right?
But how did they do it? To answer this question, we must start by determining whether there were indeed differences between left-wing and right-wing deputies.
We focused on male MPs for statistical and practical reasons – men made up 81% of the sample, and because men’s dress is more standardized than women’s, it is easier to code. The appendix to the study shows, however, that we find the same results when considering the MPs together or only the MPs.
We coded the characteristics of the deputies: being Caucasian or not, the color or absence of a tie, wearing a mustache or a beard, smiling in their photo while baring their teeth. They thus established a form of composite portrait of the deputy from each side, what we call the prototype of the left and right deputies.
In the sample, right-wing MPs were less likely to be non-Caucasian and less likely to have a beard or mustache than their left-wing counterparts. On the other hand, right-wing MPs were less likely to wear glasses and more likely to smile than their left-wing counterparts. Right-wing MPs also wore red ties, ties of another color, or no tie at all less often than their left-wing counterparts, but these differences were statistically smaller.
In any case, there were visible and statistically significant differences between right-wing and left-wing deputies.
A part of truth
Secondly, we studied the characteristics that led Internet users to classify an MP as right or left, in order to detect the stereotypes used and compare them to the prototypes of MPs.
The results provide a remarkably stable description of stereotypes across surveys. Being Caucasian, wearing a blue tie, being hairless are characteristics that Internet users associate with the right. Conversely, not being Caucasian, wearing glasses, a red tie, a tie of any color other than red or blue, or no tie at all, having a beard or mustache, are characteristics that Internet users associate it with the left.
These characteristics are indeed those of the prototypes of the two groups of deputies. The stereotypes therefore have some truth. What’s more, Internet users react to the same characteristics whether they declare themselves to be right-wing, left-wing or do not declare their preference. The stereotypes would therefore be consensual in addition to being realistic.
However, the magnitude of the effect of a characteristic on the probability that an Internet user categorizes a photograph on the left or on the right increases strictly with the representativeness of this characteristic. This means that some differences may be exaggerated.
This is particularly the case for subjective characteristics. If MPs on the right score on average marginally higher on competence and those on the left score marginally higher on trust, their scores are so close that they can barely or not at all be discerned from each other. a statistical point of view and these characteristics should therefore not be used to categorize photographs. However, Internet users, from the right and the left, classify the deputies who seem competent to them on the right and those who inspire them with confidence on the left.
And attraction?
The attraction stands out. When an MP is considered attractive on average, he or she is classified as left-wing by left-wing internet users and right-wing by right-wing internet users. However, the attractiveness scores of left-wing and right-wing MPs are on average indistinguishable.
This result suggests the existence of a halo effect, the cognitive bias that causes people to assume when they perceive a positive characteristic in a person that their characteristics in other areas will also be positive.
Concretely, if we find a person attractive, we tend to adorn them with other qualities, being spiritual, benevolent, left-wing if we are left-wing and right-wing if we are right-wing. Attraction is the only characteristic for which we were able to identify this bias. In a sense, the charm is therefore stronger than partisan stereotypes.
