Polling of the US election has been widely criticised following the outcome of last Tuesday’s ballot. For weeks in the run-up to polling day the polls were widely reported as saying that the result was too close to call. Not only was there little difference between Donald Trump and Kamala Harris in terms of national vote intentions, but this was also the position in seven “swing” states where the outcome would decide who would win the electoral college vote.
As a result, we were warned it might take days for the winner to be known while the final ballots were counted in what could be razor-thin margins in those swing states.
Yet, in the event, people in Britain woke up on Wednesday morning to be told it was clear that Trump had won. Not only was he ahead in all seven swing states, including in the three where Democrat hopes were highest – Michigan, Pennsylvania and Wisconsin – but he also had a decisive lead in the overall national vote. The polls had it seemed once again got it all wrong.
However, now that nearly all the votes have been counted, a closer look at the performance of the polls reveals that – although on average they did somewhat underestimate Trump – the error was less than in 2020. The problem for the polls at this election was that even the smallest error in one direction or the other from the anticipated very tight contest was almost bound to create the impression they had got it “wrong”.
Consider, first of all, the national popular vote. On this the various websites that aggregate the........