That recent survey from the Public Religion Research Institute and The Atlantic magazine that suggested “cultural anxiety” as the main factor that drove voters toward Donald Trump in the last presidential election? Michael Brendan Dougherty explains at National Review Online why he treats the results with skepticism.

When it comes to social science, I tend to be what you’d call a denialist. People answer surveys in stupidly partisan fashion. And counting their answers seems like a good way to waste time. Republicans and Democrats will give you partisan-sounding answers on whether the U.S. should bomb the fictive city of Agrabah from Disney’s Aladdin or welcome Agrabah’s refugees. No, really. People give social scientists all sorts of crazy conspiratorial answers for a very simple, human reason: They don’t want anyone using their anonymous answers to bolster their partisan enemies. If a pollster calls my house and asks me whether Governor Andrew Cuomo is poisoning the water with a chemical agent, like the villain from a Batman movie, I’m not going to give them the satisfaction of a pro-Cuomo answer.

Less dramatically, Republicans who turn up in surveys over the past year revealed that their partisanship is highly responsive to what their party’s candidate says. See the recent surge of support for government-provided health care among Republicans after Trump seemed to endorse it. If Trump is talking about how cultural change threatens Americans, Republican respondents are going to make efforts to mirror his rhetoric when polled.

But even if you don’t assume that most social science is bunk, there are some problems with a survey like the PRRI-Atlantic one. Once you remind people of their partisan affiliation, all the moral and cultural scripts they associate with their party will come to mind.