Predicting Elections: Child’s Play

In a paper published in Science, Antonakis and Dalgas show that judgments based on political candidates’ faces were reliable forecasts of election outcomes. What is new in the study is that judgments made by little children (from 5-13 years old) were as predictive as those made by adults.

Inspired by the study of Todorov and colleagues (Todorov, Mandisodza, Goren, & Hall, 2005), Antonakis and Dalgas first showed 684 university students 57 pairs of faces from the 2002 French parliamentary election. Political candidates were unknown to the students, which is not surprising given that the election had taken place five years before the experiment in another country.

Each student rated one pair of faces on competence. Controlling for fixed-effects of pairs of faces Antonakis and Dalgas found that the probability of predicting an election result correctly based on the competence ratings was 0.72. These results suggest that voters were inordinately anchored on facial appearances and that they did not ponder additional information on the candidates sufficiently.

Having established that the “face effect” worked well with the French parliamentarians, they used these results as a baseline on which to compare the children. They then used a game played by little children (n = 684, mean age 10.31) and older individuals (n = 160, mean age = 30.49) regarding Odysseus’s trip from Troy to Ithaca.

At the end of the game, players chose who they would want as the captain of the boat. Remarkably the probability of a child correctly picking the election winner based on choice of captain was 0.71. This accuracy was no different when including the older participants; prediction accuracy was unrelated to age.