Last week, President Trump reversed course on whether he’d keep U.S. troops in Syria.

Just a few months ago he’d alarmed his Congressional allies and prompted his Secretary of Defense Jim Mattis to resign when he announced – via tweet – that he would withdraw all U.S. troops from Syria. Now he’s abandoned this position, announcing he’ll keep a small U.S. military force in the country.

Many observers were relieved, having worried that the U.S. would leave a vacuum that ISIS and other Islamist groups would fill. But politically, Trump’s about-face may cause problems. Flip-flops make leaders look unreliable and indecisive – and generally, scholars find, Americans disapprove of leaders who renege on their commitments.

But Trump needn’t worry, our just-published research finds. Americans’ opinions of Trump are so entrenched that they don’t change even when he publicly vacillates on major policy issues. When Trump flips to a position his supporters like and his opponents don’t, it confirms their existing opinions of him. And when Trump flips to a position that his opponents like – however rarely it happens – their approval tends to be fragile.

Here’s how we did our research

Trump’s reversal on Syria is not his first flip-flop. It’s not even his first flip-flop on Syria. In April 2017, Trump authorized an airstrike against Syria – even though, in 2013, he chided Obama for considering the same thing.

To examine how Americans react when Trump backs down from a policy commitment, we examined two separate events. First, we looked at public opinion about Trump’s first reversal on Syria in 2017, as we discussed here at the Monkey Cage last April. Second, to see whether Americans’ responses were similar on different kinds of issues, we examined Trump’s 2018 reversal on imposing tariffs on the E.U.

Shortly after each of these reversals, we fielded a survey experiment on samples of 1,800 U.S. citizens via Amazon’s Mechanical Turk. Both surveys were administered online using volunteer panels, though we obtained large enough samples to have a diversity of political preferences represented.

We randomly assigned respondents to one of three groups in each experiment. First, we asked all respondents whether they thought the U.S. should launch strikes against Syria, for the 2017 experiment, or whether it should impose tariffs on the E.U., in 2018. That way we could see whether they agreed with Trump’s final decision.

Then, those in the first group read a news article that was unrelated to the reversal and which never mentioned Syria or tariffs. We used this group to find a baseline of opinion toward Trump.

Respondents in the second group read a news story about Trump’s final decision on the issue. For the first experiment, this was a story about Trump’s decision to launch airstrikes. For the second, this was a story about Trump’s decision to not levy tariffs.

Finally, those in the third group were shown that Trump had flip-flopped. For the Syria experiment, this group first saw Trump’s tweet from 2013 advocating staying out of Syria – and then read the same story reporting that he’d launched airstrikes. In the tariffs experiment, this group first saw Trump’s tweet favoring trade tariffs – and then read that he hadn’t levied them after all.

After reading the stories, we asked respondents whether they approved of Trump’s handling of the situation and to assess his overall job performance.

Flip-flopping doesn’t hurt Trump

If the conventional wisdom is correct, respondents in the third group – who read about the flip-flops – should give Trump the lowest approval ratings. That’s not what we found. Rather, respondents’ existing opinions toward Trump appeared to override any disapproval of the flip-flopping.

In the Syria experiment, those who opposed the attack and those who supported the attack began with different views on Trump. If they didn’t want the U.S. to strike Syria, they already disliked Trump. When they learn that he launched a strike, it confirmed their disapproval. Meanwhile, those who supported military intervention already supported Trump. Discovering that he flipped to a position they supported barely shifted their deeply-entrenched approval. You can see that in the graph below: All three bars are roughly the same height for each group, suggesting that the different readings had little effect.

We found something different for the tariffs. Trump’s political opponents were more likely to oppose tariffs while his supporters were more likely to favor them – meaning that Trump’s final action should be something his opponents like. When told he did not impose tariffs, support for how he handled the situation shifts from 9 percent to 45 percent.

Should Trump care about this approval bump? We don’t think so. It’s fragile – and disappears when respondents learn that he reversed course, as you can see in the drop from 45 percent to 18 percent in the first set of bars. And when we examine job approval more generally, we find that those who oppose tariffs hold low opinions on Trump’s overall performance in office across the board, even among those who didn’t read about his flip-flop.

In sum, Americans have strong opinions about Trump. Those opinions don’t shift much because of any one decision. Because Trump supporters already approve of him, their approval does not go higher even if he reverses himself and takes a position they support. Likewise, his opponents may evaluate Trump’s actions more positively if he takes a position they agree with, but that support vanishes if they are given any reason to disapprove of him.

All told, Trump’s flip-flopping seems to matter very little given today’s polarized political climate.

Sarah E. Croco (@SarahCroco) is an associate professor of government and politics at the University of Maryland in College Park and the author of “Peace at What Price: Leader Culpability and the Domestic Politics of War Termination” (Cambridge University Press, 2015).

Jared A. McDonald (@JaredAMcDonald) is a doctoral candidate in the department of government and politics at the University of Maryland in College Park whose research focuses on the role of empathy in voting behavior.

Candace Turitto is a D.C.-based lecturer for New York University and a professor of politics and government at California State University, Northridge.