Cold weather isolation is worse in poor and non-white neighborhoods in the United States.

Social isolation can cause a variety of adverse physical and mental health effects and is central to understanding broader social disparities among marginalized groups in the United States. This study aims to assess whether temperature variation is associated with daily social isolation at the neighborhood level. I test a series of two-way fixed effects models to see if mean daily temperature is associated with individuals spending the entire day at home, as measured using smartphone data, across a sample of 45 million devices in 2019 in the United States. Using interaction terms, I specifically examine heterogeneity in temperature effects by neighborhood racial composition and socioeconomic status. The two-way fixed effects models reveal highly statistically significant negative coefficients for the interaction between temperature and neighborhood proportion Black, temperature and neighborhood proportion Hispanic, and temperature and neighborhood residential disadvantage, in predicting the probability of spending the entire day at home. In marginal terms, the findings indicate the gap in the probability of spending the entire day at home between an all-Black neighborhood and an all-White neighborhood grows by nearly 10 percentage points from the warmest day of the year to the coldest day of the year in some parts of the United States. My models highlight how residents of poor and majority Black and Hispanic neighborhoods experience disproportionate social isolation in the form of a greater propensity to spend the entire day at home.