Editors’ note: In light of President Trump’s contradictory tweets about Syria over the last two days, we are rerunning this piece from May 3, 2017 on why flip-flopping did not hurt the president politically.

On April 6, President Trump authorized an airstrike against a Syrian air base in response to President Bashar al-Assad’s use of chemical weapons on civilians. This was an abrupt about-face for Trump. In one tweet in September 2013, he chided then-President Barack Obama for being a “very foolish leader” for contemplating an attack on Syria, warning that “many bad things will happen the U.S. gets nothing.”

Did Trump’s “flip-flop” on military intervention in Syria hurt him?

According to conventional wisdom, it should hurt his approval ratings, because reversals allegedly make leaders look wishy-washy, unpredictable and untrustworthy. Several op-eds made this point in the days after the attack. Much of the political science literature on this topic agrees. Political scientists have argued that leaders pay a political price for such a stark reversal, especially about an international powder keg like Syria.

But in a study we conducted right after the president’s Syria intervention, we found that Trump’s flip-flop didn’t hurt him much at all. Instead, people viewed his action through the lens of their existing views of Trump.

Here’s how we did our research

When Trump ordered the attack, we fielded a survey experiment on a sample of 1,800 U.S. citizens, randomly assigning each person to one of three groups.

The first group read a short news article about the recent Wells Fargo scandal, in which the bank opened more than 2 million fraudulent accounts. This story, by design, did not mention Syria or Trump, so we could use it to establish a neutral baseline.

People in the second group read a news story about Trump’s decision to use force in Syria.

People in the third group first saw Trump’s tweet from 2013 advocating staying out of Syria, then read the same story the second group did about Trump’s attack on Syria.

We then asked several questions related to whether people approved of Trump’s handling of the situation and asked for their perceptions of his predictability. If the conventional wisdom is correct, Trump’s approval should be lowest in the third group, where the change in policy is most evident. People in this group should also perceive him as less predictable.

Flip-flopping on Syria didn’t hurt Trump much

Our findings, however, reveal that people’s existing feelings about Trump often overrode the policy reversal’s effect. Indeed, when asking about perceptions of the president’s predictability, the only people who punished him were those who already disliked Trump. The president, therefore, probably needn’t worry about domestic political fallout from this particular shift; it’s unlikely to cost him a significant number of supporters.

Three pieces of evidence support this conclusion. First, we found that people who did not vote for Trump in 2016 were more likely to say he had reversed course, no matter which story they read.

A full 20 percent of Trump voters, on the other hand, were unwilling to say he had changed course, even when shown credible evidence that he had.

If a politician’s reversal even registers in a voter’s mind, then, it is largely a byproduct of previously held opinions about that politician — at least if it’s Trump, in our current polarized era. Citizens do not respond to a policy reversal in a uniformly negative way. Rather, political preferences act as a powerful filter.

Second, we asked whether people approved of how Trump handled the situation in Syria; whether they approved of how he was handling his job as president; and whether they would probably vote for him in 2020. Our results show that Trump’s reversal was seen as a problem only in how he handled the situation in Syria, and even then only for a small group of Trump supporters (see the rightmost set of bars in Figure 1, and compare the pink and green bars to the blue).

This finding may seem contrary to our general argument. But leaders don’t care much about what’s called “situational approval” — especially compared with their job approval or future electoral fortunes. And those two measures didn’t change. For example, when looking at job approval in Figure 2, among Trump and non-Trump voters, all three bars are of roughly equal height, regardless of which news story someone read.

Contrary to the conventional scholarly wisdom, flip-flopping does not seem to have a negative effect on a leader’s popularity.

Finally, we asked whether Trump’s behavior changed how respondents viewed his predictability. People who did not vote for Trump were more likely to see him as unpredictable if they were in the group that read about his flip-flop. Trump supporters, on the other hand, did not change their opinion on this dimension, even if they were explicitly told that he changed his position.

It’s how people view Trump that matters

Pundits and political scientists should temper their expectations for any political backlash after a leader’s change in policy. Even in the case of Trump and Syria, when the reversal was made explicit, only those people who were predisposed to dislike the president responded negatively. This phenomenon is not unique to Trump supporters. Other research has found that people in general tend to overlook flip-flopping so long as they agree with the politician’s current (if only recently acquired) position. Reversals typically gain negative traction only in the context of an election if the reversing politician’s opponent has a consistent track record (see, for example, Hillary Clinton and the Iraq War vote in her Democratic primary battle with Barack Obama in 2008).

In the wake of the attack on Syria, an increasing number of media outlets have underscored how Trump’s myriad flip-flops could be symptomatic of a broader lack of competence or strategic vision. Our research suggests, however, that voters are unlikely to view events in this manner. Given that many of his supporters reject the existence of a reversal they have seen with their own eyes, getting them to see many flips and recognize them as symptomatic of a larger, more serious problem is likely to be a very tall order.

Sarah E. Croco 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). Find her on Twitter @SarahCroco.

Jared A. McDonald 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. Find him on Twitter @JaredAMcDonald.