Manfred te Grotenhuis (RU) and Rob Eisinga (RU), together with Subu Subramanian, Rense Nieuwenhuis, and Ben Pelzer, recently published an article on The LSE US Centre’s daily blog on American Politics and Policy. The article is titled Better poll sampling would have cast more doubt on the potential for Hillary Clinton to win the 2016 election.
The full article can be read here. Below a summary.
Most polls indicated Hillary Clinton to win the 2016 US presidential election by a comfortable margin. Even in notorious swing states Florida, Michigan, Pennsylvania, and Wisconsin Clinton was predicted to have a win over Donald Trump. One explanation for the unexpected gap between the predicted odds of winning and the actual outcome is the extremely narrow electoral margin in swing states. However, in new research Manfred te Grotenhuis, Subu Subramanian, Rense Nieuwenhuis, Ben Pelzer, and Rob Eisinga show that if the polls would have used truly random samples of 1,500 voters drawn from the population of voters, they would have predicted a 7 to 3 win for Trump. After a small bias was introduced in the population the odds reversed in favor of Clinton, just as most polls predicted.
The take home message is that a relatively small polling bias which saw Republicans underrepresented in a number of key states tipped the polling clearly in favor of Hillary Clinton.
Manfred te Grotenhuis