The 2016 election scrambled many pundits’ understandings of the fault lines in American political life. Is the country fracturing along rural-urban lines? The coasts versus the heartland? The young versus the old? A new paper offers a different answer: Maybe it’s what’s tucked in the waistband (or not) that counts.

The research, published this month in Social Science Quarterly, finds that the gap between how gun owners and non-gun owners vote for president has widened in nearly every election year since the Nixon era. In 1976, there was a 7.4 percent difference between gun owners and non-gun owners when it came to their likelihood of voting for a Republican presidential candidate. By 2012, the difference had grown to more than 30 percent.

A result of the shift, the paper finds, is that gun ownership has become an increasingly reliable predictor of how a person will vote, independent of age, education, income, or other commonly cited proxies for political allegiance. Only party registration, ideology, and race are better predictors of voting behavior.

For their study, the authors — political scientists Mark Joslyn, Donald Haider-Markel, Michael Baggs, and Andrew Bilbo of the University of Kansas — mined data on gun ownership and voting in presidential elections from the General Social Survey, a nationally recognized gauge of Americans’ personal behavior and political beliefs conducted by the National Opinion Research Center at the University of Chicago. The annual GSS queries tens of thousands of Americans on hundreds of trends, tracking gun ownership alongside categories like household spending, political beliefs, and feelings about quality of life.

As the authors note, guns carry an enormous amount of symbolism for Americans of all stripes. To those who own them, firearms can stand for freedom, security, vigor, and masculinity. To those who don’t, they can signify violence, hierarchy, brute force, and indifference to society.

“We contend that the gun debate reflects a far broader conflict concerning competing conceptions of ideology and culture,” the paper reads.

It’s a conflict that the National Rifle Association, the self-appointed leader of gun culture, goes to great lengths to foment. The group’s political messaging is rife with appeals to cultural solidarity, values, and antipathy to those deemed enemies of freedom — the mainstream media, academics, Women’s Marchers — even if those targets have little to do with firearms. In this framework, Democratic candidates aren’t just gun-grabbers; they’re existential threats.

The Kansas professors’ study can be read as the latest evidence that the NRA’s campaign of cultural warfare is working. Their data shows that the divergence between the voting habits of gun owners and non-gun owners accelerated rapidly in 2008 and 2012. Each time Barack Obama — caricatured by the NRA as a condescending liberal elitist with contempt for traditional values — ran for president, gun owners veered further and further to the right.

The irony is that while Americans may be drifting further apart when it comes to the politics of gun control, there’s a surprising amount of consensus on many aspects of policy. Last month, pollsters at Pew found high levels of support across party lines, even among gun owners, for universal background checks on firearms transfers, as well as preventing gun possession by the mentally ill and those barred from flying under suspicion of terrorism. Americans, gun-owning or otherwise, also broadly oppose allowing the carrying of concealed weapons without a permit.