Three years have passed since President Trump was elected to office, but many still believe that it was the white working class that created his victory. This argument has been embraced by the media, which has put forward various explanations as to why the white working class supported Trump, including cultural anxiety, rejection of the establishment and fear of losing their status.

Yet there isn’t much direct evidence that Trump voters were hurting financially, which is generally how “white working class” is interpreted. An earlier TMC post explained how this myth appeared and in what ways it is inaccurate. But why is there no consensus about this?

Whether an average income feels like enough depends on where you live

One reason is that previous accounts have focused on studying how individual incomes affected voting decisions in 2016 by looking at citizens’ total incomes. But what if the relationship between income and who you vote for is related not to how much total money you earn, but how much you earn relative to your neighbors? In a country where income contexts vary widely, using a national scale to measure how people rank in income distributions may not be the best way to understand voting behavior.

This is because absolute income levels vary significantly from one location to the other. The national average household income level of $61,937 per year does not have the same meaning in Clay County, Ga., where households earn an average of $23,315 per year, as it does in Old Greenwich, Conn., where the average income is $236,250 per year. Purchasing power and costs of living also vary considerably between U.S. locations. For instance, a dollar is worth about 20 percent more in Danville, Ill., or Jefferson City, Mo., than in the average U.S. metropolitan area. Housing costs, which can make up a substantial portion of a household’s budget, also range substantially. For example, contrast Manhattan, where the median home value is more than $900,000, to the median home price of $100,000 in about one-third of U.S. counties.

These differences can shape whether we see ourselves as “rich” or “poor.” For most people, your ranking on a national income scale doesn’t conflict with your local income position: If your household makes $200,000, you’re rich pretty much anywhere, while $20,000 a year isn’t rich anywhere. But the story is different for many people at the middle of the income distribution, around $40,000 to $70,000 for a household. Within this income band, many are locally rich, and others are locally poor — even though they may be nationally “average.”

So what does that mean for how they vote?

How we did our research

Our work suggests that when making voting decisions, people assess their economic position relative to members of their own communities rather than to folks in the whole country.

In our research, we analyzed support for Trump among white voters in the 2016 presidential election by using survey data from the Cooperative Congressional Election Study (CCES). Unlike other studies, we incorporated a local measure of income, built by placing individuals in their local income distributions based on census data.

When we include this local income measure to examine support for Republican candidates in previous elections since 2000, our results are clear. We find local income position has a significant association with vote choice. As the data in the table show, support for Trump was strongest among the locally rich — that is, white voters with incomes that are high for their area, though not necessarily for the country as a whole. Of course, a majority of white voters across all of these categories supported Trump; our findings don’t contradict the importance of race in shaping the 2016 election.

These results hold up even when we include controls that also predict vote choice, and when we examine different subsets of white voters: with or without college degrees; men and women; living in different regions of the country; living in areas with high or low percentages of black residents; living in areas with high or low income inequality; living in rich or poor states.

In fact, looking back over time, locally rich whites have tended to support Republican candidates in most elections since 2000.

Politics — and economics — really are local.

These results suggest that we should be paying more attention to how local context affects national voting patterns and partisan identification. This may be particularly important during what economists call the “Great Divergence” — an era in which we’re seeing more inequality not just between individuals, but also across places as economic opportunities and activities concentrate in fewer metropolitan areas.

We may also need to look locally to understand Americans’ other political attitudes and behaviors, especially on issues like housing policy that relate to relative incomes.

Thomas Ogorzalek (@TKOPolitics) is assistant professor of political science and urban studies at Northwestern University and author of “The Cities on the Hill: How Urban Institutions Transformed National Politics” (Oxford University Press, 2018).

Luisa Godinez Puig (@Luisa_Godinez_P) is a PhD candidate in the department of political science at Boston University.

Spencer Piston (@SpencerPiston) is assistant professor of political science at Boston University and author of “Class Attitudes in America” (Cambridge University Press, 2018).