Alex Todorov

Who they are

Todorov's subject is first impressions treated as data. Faces are read almost instantly, the reading is consistent across people, and the question his work asks is whether anything outside the viewer's head corresponds to it. The election study is the strongest case anyone has made that something does.

The headline number needs flagging. Thinking, Fast and Slow gives about seventy per cent as the rate at which the candidate rated more competent won. That figure is Kahneman's rounding across different chamber types, which had different hit rates, rather than a single statistic reported from a single analysis. Anyone quoting seventy per cent as Todorov's result is quoting a compression. Either use the primary paper's chamber-specific figures or say well above chance and leave the number out.

What the finding is matters as much as how large it is. It is a correlation between a snap judgment made by naive raters and an election outcome. It is not evidence that voters decide on faces, because the same facial cues may correlate with things that genuinely drive outcomes, including incumbency, funding, party strength in the district, and selection by parties who themselves prefer candidates who look the part. Kahneman uses it as a demonstration of substitution, a hard question about competence answered by an easy question about appearance, and that use is legitimate. The causal reading is not.

One further point of fairness. This work sits in the same chapters as the priming literature Kahneman later conceded had not held up, and it is a different kind of study: archival outcomes, large candidate sets, naive raters, a straightforward prediction. It should be judged on its own design rather than by association, which is the standing rule on this site.

Key papers

The vault does not yet record the primary citation for the election-outcomes work, and this site will not supply a paper, year, journal or sample size it has not checked. What is held is the secondary route, below, and the flag on the number that travels with it. The section will be completed when the original paper has been read.

Kahneman, D. (2011), Thinking, Fast and Slow, Farrar, Straus and Giroux. Where the finding reaches a general audience, used as an illustration of substitution and of what you see is all there is, and where the rounded competence figure originates.

Key ideas

Competence read from a face predicts election outcomes at rates well above chance, when the raters do not recognise the candidates.

The seventy per cent figure is a rounding across chamber types, not a reported result. Do not repeat it as his.

Prediction is not causation here. Facial cues may be proxies for incumbency, funding and party selection rather than drivers of votes.

The finding is a clean example of substitution: a hard question about competence answered by an easy one about appearance.

Proximity to failed work is not evidence against a study. Judge the design.

---