Washington State University analysis that used anonymous cellphone data shows families may not have been able to stomach each other’s political views after Donald Trump was elected president. Holiday visits appear to be shorter than in 2015.

Share story

Like many Americans, Ryne Rohla was burned out on politics after the bitter 2016 election campaign. So when guests at his grandparents’ home kept interjecting comments about Donald Trump and Hillary Clinton during Thanksgiving dinner, Rohla slipped out a little early.

The Washington State University graduate student wasn’t alone.

In an analysis published Thursday in the journal Science, Rohla and co-author M. Keith Chen report that political polarization reduced the amount of time families spent together on the most quintessential of American holidays.

Among families with divided political loyalties — picture your MAGA-cap-wearing brother sharing turkey with your firebrand feminist cousin — Thanksgiving celebrations in 2016 were 30 to 50 minutes shorter than among families with shared political sympathies.

Heavy political advertising aggravated the effect, like dumping gasoline on smoldering hostilities. In the parts of the country most barraged with television ads, family gatherings were two to three times shorter.

Collectively, Americans lost out on 34 million hours of “cross-partisan Thanksgiving dinner discourse in 2016, owing to partisan effects,” the paper concludes. And that’s not good for families or society, said Chen, a political economist at the University of California, Los Angeles.

“This kind of no-holds barred, incredibly polarized political campaign and political advertising does more than just make democracy hard,” he said. “It also directly hurts people.”

The study backs up anecdotal and media reports of people filled with dread at the prospect of having to argue with, or ignore, relatives at the opposite end of the political spectrum, said Rohla, a doctoral candidate in economics. “It’s really consistent with my own personal experience of how people behave.”

Perhaps more surprising is that the researchers were able to glean so much detail about Americans’ private family celebrations and political views.

Two sources of data were key: Location readings from 10 million smartphones, and an interactive map created by Rohla that was the first to burrow into 2016 election results at the precinct level.

Stripped of any identifying information, the phone data compiled by a company called Safegraph uses navigation, weather and other apps that track the location of each phone, with the owner’s approval. To determine the owners’ home precincts, the researchers determined where each phone was usually located between 1 a.m. and 4 a.m. in the weeks before Thanksgiving.

The analysis included people who were home Thanksgiving morning, traveled elsewhere during the afternoon or early evening — presumably to spend the holiday with family or friends — then returned home that night.

The researchers used the precinct voting results to infer probable party affiliation, and identify Democrats who traveled to Republican-leaning precincts for the holiday and vice-versa, as well as people who traveled to areas that shared their political affinities. Data from 2015, when the average Thanksgiving stay was about 4.5 hours, were used for comparison.

Among the study’s findings was that fewer Democratic-precinct dwellers traveled for Thanksgiving in 2016 than in 2015. Those living in Republican precincts who traveled to Democratic areas bailed out sooner, cutting their visits by 50 to 70 minutes.

In Washington, the biggest effect wasn’t in deep-blue King County, but in surrounding areas, Rohla said. Pierce County residents who traveled to politically divergent areas lost about 35 minutes of family time, while those from Skamania County lost 39 minutes.

“It’s really the people in the middle (of the political spectrum) who are the most adversely impacted,” said Rohla, who includes himself in that category.

The researchers posted an earlier version of their results in 2017 but have since refined their analysis and discovered large effects.

The analysis is far from perfect, omitting people who made extended Thanksgiving visits and making assumptions about party affiliation based on where people live.

The “big data” approach doesn’t actually prove that people were leaving early due to political spats, cautioned John Wilkerson, director of the Center for American Politics and Public Policy at the University of Washington. “It is also not the final word,” he wrote in an email. “But it is a creative use of innovative data to investigate an interesting question that would be hard to investigate otherwise.”

To try to nail down the role of politics, Chen and Rohla dug into the impact of political advertising in 2016, compared with 2015 when no ads were being aired. In swing states like Florida, Orlando and other cities were bombarded with as many as 26,000 television ads during the long campaign. Thanksgiving visits by people who live in those areas were two to three times shorter.

“If you were in an area in the top 1 percent of political ads, your Thanksgiving dinner declined by an hour and a half,” Chen said. Since there was no similar effect in 2015, the researchers believe they can reasonably point the finger at politics as the driving force.

Family tensions at Thanksgiving are nothing new, but the toxic effects of political polarization can erode family ties and social bonds, Rohla said.

“Personally, I think that’s concerning,” he said. “To me it’s a symptom of a broader decline in the social fabric of the United States.”

The researchers are now examining data from last Thanksgiving, to see if those toxic effects have diminished.

“American families are remarkably resilient,” said Chen. “I want what we found in this paper to be short-lived.”