The Washington PostDemocracy Dies in Darkness

Please stop treating Trump’s daily approval ratings like the temperature

President Trump listens as he and NATO Secretary General Jens Stoltenberg participate in a news conference in the East Room of the White House, on April 12. (Jabin Botsford/The Washington Post)
Placeholder while article actions load

I spend a lot of time looking at political Twitter because I am a damaged human being whose undiagnosed masochism is clearly burbling just beneath the surface. In doing so, I will see with some regularity a tweet talking about how the new Gallup daily job approval rating has Trump up X percent or down Y percent and how that means that the thing he did or didn’t do yesterday or the day prior is dooming or buoying his presidency. That’s sort of an exaggeration, but not much. It’s tracked the way a normal person might eyeball the temperature, particularly on Twitter where the community rumbles through about 19 obsessions a day.

It’s also very misleading.

Here’s what Trump’s daily approval rating looks like.

A lot of volatility, certainly! But let’s step away from Gallup for a second.

During the campaign, RealClearPolitics’s Sean Trende raised a point that’s related. He was lamenting that fewer national polls meant that there was necessarily more daily volatility. Why? Because polling averages tend to smooth out fluctuations based on margins of error. In Gallup’s daily poll (actually itself an average of the past three days of responses), the margin of error is plus or minus 3 points. So it looks like this:

Meaning that, in theory, any given set of poll results might fall anywhere into that shaded area. So it’s just as likely that the daily trend looks like this:

To Trende’s point, I made an interactive pitting Trump against Hillary Clinton in a series of polls during the 2016 campaign for president. Crank up the number of polls each day, each of which shares the same margin of error, and you’ll see that the results end up closer to that horizontal line — the actual, Platonic view of voters — than if you only take one poll a day.

With one poll a day, you might get only tiny variability. But that’s not usually the case.

Which is one reason Gallup’s other job-approval metric — its weekly average — is more stable.

Over the course of Trump’s 16 weeks in office, the high and low points he’s hit in his weekly average is 7 points apart. In the daily poll? The spread is 11 points.

It’s certainly possible that Trump does something on Monday that infuriates people and something else on Tuesday that they love, and that his popularity whipsaws as a result. But longer-term trends tell us more about the path that the administration is on.

To that point, notice the change in the last two weekly Gallup polls: From 42 percent approval to 38 percent, thanks mostly to a big drop among independents. That’s the largest weekly change Trump has seen to date. That number, I encourage you to tweet about it.

Perhaps by sharing this article.