Understanding How Class, Education and Income Affect Voting Behavior

By Kevin Reuning (@KevinReuning)

The white working class has taken on a mythical status in today’s politics. They are a critical component of the GOP’s future. The Democratic Party does not understand the white working class, but needs to win them back as they’ve been losing them for decades now. The only positive is that the punditocracy might have finally broken from saying “working class” when they mean “white working class”. But what does it mean to be in the working class? And how exactly do they vote?

Those who employ the term “working class” rarely attempt to define it. When they do, the definitions they come up with are often frustratingly limited, such as “those without college degrees.” At the same time, the concept of working class quickly slips into the frame of blue-collar manufacturing work. This is obvious from not just Trump’s emphasis on manufacturing jobs but the media’s. Yet manufacturing jobs make up a fraction of the US job market while service jobs (of all types) continue to grow. Low wage service jobs are different not only because of the type of work, but also because they tend to be significantly more diverse and have recently been at the heart of a new worker movement through the Fight for 15.

In trying to get a better handle on who constitutes the “working class,” I’ve used data provided to us by Greenberg, Quinlan and Rosner and fielded online between June 2017 - October 2018 with a sample of 31,338 respondents. The survey asks not only for education and income (often proxies for class) but also the industry people work in. Industry is not perfect as every industry has a managerial class, but it does provide significantly more information than just income and education.

Before defining our classes we can look at how income and education is stratified across the different industries. In this first chart the colors show the proportion of respondents in each industry with a certain educational attainment. For example, unsurprisingly the majority of individuals working in the legal industry have advanced degrees. In contrast, the traditional blue collar industries such as manufacturing, production, and construction, near the bottom of the chart, have a plurality of respondents with only a high school degree. This is also true for those in food preparation and personal care (part of the new working class service industry).


The pattern for income is less clear. There are a few industries where a majority of participants are making more than $100,000: legal, management and engineering. But the blue-collar manufacturing jobs below do not look very different from many of the other categories. The distribution of income across industries is largely the same.  


Because of this I used work and education to create seven categories of employment. Two categories capture the more traditional working class: “traditional blue collar” and “service blue collar.” Both of these groups include those without a college degree but the industries differ for them. “Traditional blue collar” includes those manufacturing, production, maintenance, construction, and farming industries while the service blue collar includes those in food service, personal care, hospitality, and healthcare support. I then created a large “office” category that are likely to be those working in an office environment but at something closer to entry level. Again they do not have a college education but are working across a large number of industries. Next are the “professional” and “managerial” classes. This group has a college degree (except in the one management industry) and are divided based on if they are likely to be mainly managing others or be working in a specialized field. Finally, I created two categories that span across education level. An “education” group, and then a catch-all “other.”


Using these we can start looking at support for a generic Congressional Democrat. The sample size in this survey is relatively large (~16,000 for which we have occupational data) so we can look at more narrow groups than what we normally look at breaking things down by race, income, and work category.

The below plot shows support for a generic Democrat by race and work category. Race still is an important factor but we begin to see a lot of internal variation as well. Traditional blue collar white workers are the least supportive of Democrats with around 27 percent saying they’d support a Democrat, but those service economy blue collar Democrats (who have the same education level) are significantly more supportive at 38 percent. Forty-seven percent of whites working in professional jobs support Democrats, along with 50 percent of educators.


Among black workers, blue collar service employees are the most supportive at 87 percent while those in traditional blue collar are in the middle of the pack at 80 percent. Black professional, management, and office employees are all all in the same area around 81 percent to 82 percent.

Complicating things one additional step we can also look at how income groups vary across these categories as well. In order to make things a little less complicated I shrunk the income categories we saw above and drop out any group that has less than 25 observations.

What is most interesting here is that once we account for different employment classes income does not lead to a lot of variation. Among white voters the only group with any significant variation are those in management, which scales a large amount of education levels. There is less consistent variation across other racial groups, although some of this is hard to establish because of the smaller sample sizes here.


We’ve already started to accept that the working class is not the same as the white working class. It is clear now though that we also need to think about what we actually mean when we say white working class. Do we mean, as I suspect we often do, traditional blue collar jobs? If our assumptions about the white working class all come from an image of the mythologized blue collar worker of the the 1950s then we are missing how work has changed and how the people who work in this new blue collar service industry are politically different.

Kevin Reuning (@KevinReuning) is an assistant professor of political science at Miami University.

Ethan Winter