The original formulation of the beer question invites the question of why voters would care so much about something that is exceedingly unlikely to happen. If you invert it, however, voters start to look a lot less irrational. After all, they can’t foresee all the decisions politicians will need to make once in office, and have few ways of holding them accountable if they don’t follow through on their promises. So they need to estimate which politicians are most likely to understand and advance their interests.

A candidate’s attitudes toward “people like me” thus become a powerful heuristic. If a candidate generally likes people like me, then it seems plausible that he will look out for my interests in a wide range of scenarios. If he dislikes people like me—if he would hate sharing a beer with me, and secretly thinks I’m trash—then he is far more likely to sell me out.

The inverted beer test provides an explanation for why Coakley’s gaffe was so harmful: A politician who finds the idea of spending time with Red Sox fans unpleasant is not going to balk at selling them out. It also explains why Gordon Brown calling an elderly lady “bigoted” in 2010, or Hillary Clinton describing some Donald Trump voters as a “basket of deplorables” in 2016, resonated: It seemingly confirmed many voters’ suspicions that they are elitists who look down on “ordinary” people.