Why has the United States become so politically polarized?

Many have argued that social media, where users can find their viewpoints reinforced with slanted news stories and the partisan commentary of friends, has played a role in reinforcing tribal political identities.

That explanation has been percolating long enough and loud enough that it has even reached the Oval Office. In an interview he gave before leaving office, President Barack Obama gestured to the rise of social media as a key factor in the continuing political polarization of the United States, arguing that Americans were trapping themselves within filter bubbles, limiting their own perspectives.

“The capacity to disseminate misinformation, wild conspiracy theories, to paint the opposition in wildly negative light without any rebuttal — that has accelerated in ways that much more sharply polarize the electorate,” he said to The New Yorker.

But a new working paper suggests that the demographic groups that have experienced the most political polarization in recent years are the ones least likely to be consuming media online.

The paper, issued last month by the National Bureau of Economic Research and written by economics professors from Stanford and Brown Universities, found that the growth in political polarization was most significant among older Americans, who were least likely to use the internet between 1996 and 2012, the years for which data was available when the paper was written.

Jesse M. Shapiro, an economics professor at Brown and one of the authors of the paper, detailed the paper’s findings in an interview on Wednesday.

“We found that basically, polarization is rising at least as fast for older Americans as it is for younger,” he said. “So a simple story that says polarization is rising because people are consuming media online has some trouble explaining that fact.”

Mr. Shapiro and his co-authors declined to weigh in on the still open question among political scientists of the best way to define and measure the impact of polarization, many of the attempts devised by professors of political science and including measures of ideological and partisan affect, partisan sorting and issue divergence.

Employing those measures, and data from the American National Election Studies, they divided respondents by their estimated internet usage and found that growth in polarization was consistently higher among older people. For instance, within the index of polarization trends they created, the authors found that among respondents age 75 and older, the increase in polarization between 1996 and 2012 was 0.38 points, compared to just 0.05 points for adults under the age of 40.

Mr. Shapiro added that the authors were working on updating their results with more recent data, but that he thought the results were unlikely to change.

“My guess is that digital media are not the culprit in the big scale changes that we’re seeing, though they may be playing some role,” he said.

Evidence of the trend toward polarization continues to emerge. Last week, The Cook Political Report released the 20th edition of its partisan voter index, which assigns each of the 435 house districts in the country a score based on its political orientation in comparison to the national average. The report found that there were only 72 districts in which politicians for both parties were competitive.

The statistic represented “a 20 percent decline from just four years ago, when there were 90 swing seats,” the report found.

Mr. Shapiro cited the rise of income inequality as one potential explanation for the rise of such hyper-partisanship, but emphasized that he did not have any data to support the hypothesis.

But he said that any explanatory factor would have to make sense equally across demographics — something that the rise of social media failed to account for.