Research Press Release

Human Behaviour: Pro-Trump areas physically distanced less during the COVID-19 pandemic, predicting health outcomes

Nature Human Behaviour

November 2, 2020

People in counties in the United States that voted for Donald Trump (Republican) in the 2016 presidential election tended to physically distance 14% less than those in counties that voted for Hillary Clinton (Democrat) from March to May 2020. This study, published in Nature Human Behaviour, analysed daily geotracking data from 15 million people and found that partisan differences in distancing were associated with higher COVID-19 infection growth rates in pro-Trump regions.

Numerous polls have found a partisan gap in perceptions and support for preventative health behaviours during the COVID-19 pandemic in the United States. However, it is unclear whether these polls reflect real-world behaviour.

Anton Gollwitzer and colleagues analysed the daily geotracking data from 15 million smartphones in 3,000 US counties to study whether physical distancing in the first few months of the pandemic differed according to partisan leaning. They measured physical distancing in terms of reduction in general movement and reduction in visiting nonessential services (such as restaurants, salons and clothing stores) compared to the previous year using GPS coordinates.

The authors found that people in counties that voted for Trump in 2016 showed a 24% drop in regular activity, while those in counties that voted for Clinton showed a larger 38% reduction. They also observed that partisanship was more strongly associated with physical distancing than other factors, including a county’s median income, the number of COVID-19 cases, commuting to work, population density and racial and age demographics. Furthermore, the partisan gap strengthened over time, even when stay-at-home orders were in place. Finally, using a combination of mathematical models, the authors found that partisan differences in physical distancing related to subsequent infection and fatality growth rates. According to these models, reduced physical distancing in pro-Trump counties was associated with a higher subsequent daily infection growth rate than average.

The authors caution that their data are correlational and do not demonstrate that partisanship in the U.S. caused different physical distancing patterns. However, they conclude that partisanship might be an important risk factor for the current COVID-19 pandemic and potentially other public health crises.


Return to research highlights

PrivacyMark System