The beauty premium in politics, and what the study did not say

The finding is real and the headline version of it is wrong, which makes this a useful case study in how a careful result becomes a silly one.

The research is by Niclas Berggren, Henrik Jordahl and Panu Poutvaara, associated with the Research Institute of Industrial Economics in Stockholm, and it circulated under the title The Right Look: Conservative Politicians Look Better and Voters Reward It. Note the second word of that subtitle. The subject is politicians.

What was measured

The method is standard in this literature and worth understanding, because most objections to the result are actually objections to the method.

Take photographs of real candidates from real elections. Show them to raters who do not know who any of the candidates are, and who are typically not from the country in question, so that party recognition cannot contaminate the rating. Collect ratings of attractiveness and, separately, of inferred competence. Then match those ratings against the actual electoral outcome.

The starting point was already established before this paper: better-looking candidates win more votes, on average. That is the beauty premium, and it is one of the more robust findings in political economy. The question this study asked was whether the premium is distributed evenly across the political spectrum.

What was found

It is not. Candidates on the right were rated as better looking, or as appearing more competent, than candidates on the left, across several countries including Australia, Finland, France and the United States. In the Swedish data the pattern held in both municipal and parliamentary elections, and the beauty premium — the electoral return on looking better — was larger for right-wing candidates.

So there are two separate claims stacked here, and they are often conflated. One is about the distribution of appearance among people who end up on ballots. The other is about how much that appearance is worth at the ballot box. The paper finds an asymmetry in both.

The proposed mechanism

The explanation offered runs through information rather than through aesthetics.

Voters who know a great deal about a candidate have little use for appearance as a signal. Voters who know almost nothing have to infer competence from whatever is available, and a photograph is available. If the electorates of different parties differ in how much they rely on inferred personal competence, as against party label or programme, then the same beauty premium will pay out differently depending on which party a candidate stands for.

That is a hypothesis with supporting evidence, not a demonstrated causal chain, and the authors present it as such.

What the headline did

The version that travelled said that right-wing people are more attractive than left-wing people. That is a different claim about a different population, and the study does not support it. Nothing in the design says anything about voters, party members, activists or supporters. It is about the small, highly self-selected group of individuals who get onto a ballot paper.

The distinction matters because the plausible mechanisms are entirely different. A finding about candidates is a finding about selection: who is recruited to stand, who is placed high on a list, who is put in front of a camera. A finding about the general population would be a claim about correlation between appearance and political orientation, which is a much stranger proposition and one the research never tested.

There is a second layer of distortion in the framing that reached readers, which compared two individual named politicians. Individual comparisons are exactly what an average across thousands of candidates cannot support. A study can find a reliable mean difference between two large groups while telling you nothing whatsoever about any two members of them.

Why the underlying result still matters

Strip the headline off and something uncomfortable remains. If appearance carries electoral weight, and the weight is larger where voter information is thinner, then the beauty premium is partly a measure of how little voters know. That reframes it from a curiosity about looks into an indicator of information quality in an electorate.

It also has a practical implication for parties, and it is not a flattering one: if appearance pays, selection committees have an incentive to respond to that, and any such response operates independently of whether the candidate can do the job.