Approval ratings

Harvard Poll CAPSizes Under Its Own Weights

by D.J. McGuire

Last week, Harvard University proved William F. Buckley right – again – about the superior intellect of the first 200 names in the Boston phone book (that might sound weird, but we still get phone books down here, so I presume they’re still around up there). The Ivy League Institution’s latest embarrassment came from its Center for American Political Studies (CAPS), which released a poll claiming President Trump had an approval rating of 47%. Strangely, the Trumpenproletariat missed its cue for a few days, but they’re all over social media and opinion columns with the numbers now.

Curious as to the reasons behind the supposed surge in Trumpularity, I do what I always do when a poll catches my eye – go for the crosstabs. They can provide interesting little nuggets of info that – well, that can make me look a lot smarter than I really am. More to the point, they also provide the “weights” that a poll used to create a “representative sample” – i.e., if fewer white women with college degrees answered then were supposed to answer the poll, they have added “weight” until they’re a share of the sample that matches with America as a whole.

At least that’s the idea. Sometimes, though, a pollster can sets weights that are thoroughly out of balance. A pollster like, for example, the Harvard CAPS.

The first problem I noticed in the Weighted Sample was the 2016 vote, which had Trump ahead of Clinton by 1%. As Clinton was 2% ahead in the actual popular vote, that was a clear tilt in Trump’s direction. There were other examples. Urban voters were 3 points lower (and rural voters 3 points higher) than the 2016 exit poll. Women were two points lower (and men two points higher). Voters over 65 were a whopping 5 points higher in CAPS than in 2016.

But the real kicker came in the education weight. In 2016, 50% of voters were college graduates or higher. CAPS? Only 35%

Given how college graduates easily preferred Clinton – and, if anything, have shifted even further away from Trump since his election – this weight crushes the poll that spawned it.

Democrats should never be complacent. There will be polls that we don’t like and they won’t be this easy send to the recycle bin. This poll, on the other hand, can be safely ignored.


D.J. McGuire – a self-described “progressive conservative” – has been part of the More Perfect Union Podcast since 2015.