The Corner

This Is Very, Very Odd

The results in the election indicate that political polling before the election did accurately reflect the outcome in most cases — or at least it did if you mashed all the polls together and came up with an average of their results. What didn’t work, for the umpteenth time in a row, was exit polling. This year the exit polling was skewed somewhere between 5 and 8 percent toward Democrats. This is not supposed to happen. Theoretically, exit polls are supposed to be far more accurate than telephone polls because of the sheer number of people interviewed and the depth of the data recorded. You’re not supposed to have to “weight” exit polls — subject them to interpretation — because since they record interviews in the tens of thousands, they should achieve a truly random effect and therefore reflect the vote precisely. So here’s my modest proposal, which will save news organizations millions next time: End exit polling, and do whatever you can to do telephone polls on Election Day. Fox did this and its election-day poll was pretty accurate.

The Latest