What Actually Moves Trump’s Approval Ratings?

Since President Trump was inaugurated, this question has come up over and over again. Trump’s approval rating might be the most important polling statistic in American politics right now—historically such ratings have well-documented influence on midterm elections, and more popular presidents tend to have an easier time getting Congress to pass their policy agenda. But approval ratings aren’t always easy to predict. Sometimes seemingly huge news stories fail to make a dent in Trump’s numbers, while a bad policy move or a good week for Trump can change his polls in simple, straightforward and significant ways.

To get a better handle on exactly what changes Trump’s approval rating and why, I combined a few data sets. I looked at the polling averages from FiveThirtyEight and RealClearPolitics as well as the news archives at Memeorandum (an aggregator that algorithmically figures out which pieces are being cited most in the media). The basic intuition is simple: Looking at which news stories materialize before a rise or fall in Trump’s approval rating might give us some information about what events drive his numbers up or down. Public opinion is complicated and this approach alone doesn’t explain everything about Trump’s approval rating (more on that later). But for now we’ll start with the simple stuff and work our way into the knottier aspects of this question.

Policy, Disaster, Competence: The Real Swings in Trump’s Approval Make Sense

Figuring out which stories shift poll numbers can be tricky, but I took a fairly straightforward approach. I used change point detection (a statistical technique that tries to figure out when some series of data points shifts in a way that’s distinguishable from noise), along with a close subjective look at the data to figure out when Trump’s approval moved by a significant margin. Then I looked at the news stories that preceded these shifts (both from Memeorandum and other sources) and found some patterns.

First, health care matters. On March 6, the day that the House GOP unveiled a bill that would repeal and replace the Affordable Care Act, Trump’s approval sat at 44.5 percent according to RCP. The bill didn’t poll well
—by March 24 it was withdrawn, and by April 4 Trump’s approval rating had fallen to a new low of 39.8 percent. The timeline of Trump’s drop in the polls doesn’t match the events perfectly, but that’s to be expected. Pollsters often have to survey people for a few days before calculating and publishing their results, and aggregators use polls from multiple days. So we should expect the effect of any news story to be delayed.

Obviously there was more than just health care happening at that time. Trump had appointed Cabinet members, fought with judges over a controversial travel ban, announced that he was nominating Neil Gorsuch to the Supreme Court and made a number of other moves that might have figured into voters’ initial assessment.

But as health care receded and other stories (e.g., Syria) came into focus, Trump’s approval ticked up by a few points. And when health care came back into the spotlight in May (the House passed an Obamacare replacement that didn’t poll especially well), Trump’s numbers began to dip again. Presidential approval remained steady for most of the summer, but there was another small drop after the Senate rejected “skinny repeal” in late July.

The relationship between health care and Trump’s polling isn’t simple. It’s hard to tell exactly if/what effect Graham-Cassidy had (it wasn’t popular, but it was proposed around when Trump began to get a bump for his handling of Hurricanes Harvey and Irma). And it seems a bit odd for voters to both punish Trump for pushing unpopular health care bills and for failing to pass them. And, as I noted previously, there’s always more than one simultaneous storyline in the Trump administration.

But after the president made public, sustained pushes for unpopular policy, he tended to suffer in the polls. And while “skinny repeal” didn’t poll well, it’s possible that some voters were at least temporarily turned off by his public inability to push through one of his key legislative priorities. In other words, two things that affect most presidents—policy and the appearance of competence—seem to matter for Trump.

It’s worth noting that non-policy stories may have also hurt Trump’s approval. For example, some polls showed that Trump suffered after his responses to the violent clashes between Neo-Nazis and counterprotesters in Charlottesville, Virginia. The aggregate didn’t move much, but it’s worth mentioning this instance to show that non-policy events can move polls as well.

And Trump’s numbers haven’t just gone down. In late September, his response to Hurricanes Harvey and Irma may have led to a small rebound in his approval ratings. As in these other cases, there were other political and policy stories happening simultaneously, but data suggests that voters viewed his response to those storms as competent and thus his approval rating went up (though he earned worse marks for his handling of Hurricane Maria in Puerto Rico).

This is far from an exhaustive list of events that might have changed the polls, and these swings are overdetermined—that is, there’s more than one cause that could have produced the effect. But the overall thrust of this analysis is that Trump, despite all his oddities and his historically low poll numbers, is conventional. If he deals with a crisis competently (e.g., hurricanes), his approval rating ticks up. If he pushes an unpopular policy (e.g., various health care bills), he suffers in the polls.

But There’s a Lot Happening Outside of Those Big Swings

Obviously there’s more to Trump’s approval rating than simple ups and downs, and there’s more to news than just health care, storms, and Neo-Nazis. And while we’re not going to be able to explain every intricacy of the Trump phenomenon here, it’s worth noting that other important forces can shape public opinion without forcing it directly up or down at some specific date.

For example, the economy may be helping Trump’s approval without showing up in the news much. As my colleague Mike Warren astutely pointed out, Trump had a higher rating on his handling of the economy than other issues in a recent Fox News poll. And his average approval rating on the economy is higher than this overall job approval rating or his numbers on foreign policy. But that aspect of Trump’s approval rating might not always show up in the daily news, since reporters are (understandably) more focused on detailing the day-to-day workings of Congress, the White House, the parties, etc. Moreover, it’s possible that factors that we aren’t actively measuring also pull Trump’s approval up or down.

Additionally, some of the factors that change Trump’s approval rating won’t be captured in a single event. There’s no one moment that’s going to signal to voters that Trump definitely is or isn’t able to get things done. As Trump attempts to fulfill campaign promises (e.g., repeal and replace the ACA, cut taxes, build a border wall), voters will decide whether or not he’s competent and his approval ratings might move up or down as that assessment changes. And there are always long-running storylines that currently break on partisan lines but could wear on Trump’s approval as events develop and change (e.g., Trump’s relationship with Russia).

My point here isn’t to describe all the complex ways that news events or other factors could change public opinion (that might take multiple books) or to list off every news storyline and describe exactly how it’s changed public opinion. It’s to show that some aspects of Trump’s approval are simple and direct, some are complicated and some may not yet be fully understood.