he worse a community’s health the more strongly its voters backed Donald Trump in the 2016 election compared to their support for Mitt Romney, the 2012 Republican presidential nominee, researchers reported on Monday.
The findings suggest that public health “might influence” how people vote, said Dr. Jason Wasfy of Massachusetts General Hospital, who led the study, which looked at factors such as death rates, diabetes prevalence, and teen pregnancy. “The communities that shifted from Romney to Trump in general have worse public health.”
Some political psychologists dismissed that association as probably the result of too much data dredging leading to spurious correlations — such as, famously, that the decline in the divorce rate in Maine is almost perfectly correlated with the decline in margarine consumption.
“We’ve learned over the decades that simply finding a factor correlated with voting doesn’t tell you anything about why people voted as they did,” said political psychologist Jon Krosnick of Stanford University. “It’s tempting to look at a correlation like this [study’s] and say, aha, now we know what caused Trump to win, but we don’t.”
The idea that a community’s health might correlate with its politics nevertheless fits with the view that voters who feel disaffected and left behind, in particular for economic reasons such as losing their job when a coal mine or factory closed, were both more likely to vote for Trump and to have seen their (and their neighbors’) health go downhill from things like abusing drugs and alcohol or developing diabetes and obesity.
“Our results suggest that communities that are more distressed, at least in the sense that distress is measured by public health, moved from Democratic to Republican,” Wasfy said. “Sicker communities overall shifted their votes toward Republican presidential voting in 2016.”
The shift was especially strong in states that have historically voted for Democratic presidential candidates but went for Trump last year, accounting for his Electoral College victory.
Wasfy and his co-authors, including Massachusetts Institute of Technology political scientist Charles Stewart, ran standard statistical tests to see whether poor public health was just a proxy for poverty, unemployment, and education level, all of which are recognized to drive voters’ choices. After factoring out these demographic differences, they found that Trump getting a larger fraction of the vote than Romney is still correlated with public health. Their study was published in PLOS ONE.
But did poor public health in a community cause numerous voters to like Trump more than they did Romney? Perhaps they saw neighbors, friends, and even family members dying young, becoming addicted, getting pregnant as teenagers, and the like, then thought, “America isn’t working for us,” and therefore voted for the outsider, “drain-the-swamp” candidate. “A causal relationship is possible,” Wasfy said, “although we cannot prove [one] from our study design.”
The researchers combined presidential election data for 2012 and 2016 from 3,009 of the 3,142 U.S. counties with data from the Robert Wood Johnson Foundation on public health in these counties. They mashed seven measures — how many days of poor health residents reported; rates of food insecurity (not knowing if you’d eat), obesity, diabetes, and teen births; rate of visiting primary care physicians (a marker of health consciousness); and age-adjusted mortality — into a score of county “unhealthiness” and calculated how it jibed with the Romney-to-Trump shift.
The 88 percent of counties where Trump outdid Romney’s percentage had, in general, the highest unhealthy scores. The link between poor public health and shifting to Trump was especially strong in Iowa, Pennsylvania, Wisconsin, Michigan, Ohio, and Florida, all of which Romney lost and Trump won.
As it happens, The Economist reached a similar conclusion a week after the election, when it took up a “challenge for data nerds” from a pollster to “find the variable that can beat % of non-college whites in the electorate as a predictor of county swing to Trump.” After running lots of statistical tests, The Economist found that life expectancy and rates of obesity, diabetes, heavy drinking, and lack of physical activity together explained 43 percent of Trump’s gains over Romney. A county’s share of whites without a college degree, which many pundits identified as a key factor in Trump’s win, explained 41 percent of the shift to Trump.
The new analysis does not prove that sicker people, as a bloc, voted for Trump more than they had for Romney. Trump’s larger share of votes than Romney’s in unhealthier counties could have happened because sicker people didn’t vote, for example. It also seems to contradict research showing that better health is associated with voting Republican, and raises questions about why very sick people would vote for Trump more than they did Romney. Trump, who promised to repeal Obamacare, “couldn’t be perceived to be the candidate who would take care of people in poor health,” said Stanford’s Krosnick.
On the other hand, people often vote against their self-interest. “There is something to the theory that county-level health outcomes played a role in explaining the shift to Trump, but we should be careful about saying it played the predominant role,” said Patrick Ruffini, a Republican strategist and pollster who issued the “data nerds” challenge. “The share of voters who are white without a college degree remains the best single variable that predicts the shift.”
Yet adding public health data “does explain more than the rate of non-college whites alone, so there is value in highlighting it,” he said. If health is “a proxy for optimism and well-being,” Ruffini added, then poor health “may help explain the appeal of a populist candidate promising to blow up the system.”