Academics and organizers often occupy different worlds. But just as academics benefit from knowing and interacting with practitioners of organizing, so too do organizers benefit from understanding research on organizing practices. The purpose of this series is to help translate the work of academic researchers into a language all organizers can speak and suggest paths of collaboration between organizers and academics.


Paul Frymer and Jake Grumbach’s new paper, “Labor Unions and White Racial Politics,” explores the role that union membership plays in reducing white racial resentment. Using data from election surveys, Frymer (a professor of politics at Princeton) and Grumbach (an assistant professor of political science at the University of Washington) demonstrate that white union members, in general, have lower levels of racial resentment and are more likely to support policies intended to address racial inequities. Even more significant, the mere act of joining a union is strongly correlated with a reduction in racial resentment.

But is this just a coincidence? Or a case of self-selection? Perhaps people with lower racial resentment are just more likely to join unions. If it is more than coincidence, what explains it? What does this research tell us about the popular conception of racist white working-class union members? And what can unions and other progressive organizations do with this information? 

To get answers to these questions, labor organizer and former academic Dave Kamper interviewed one of the co-authors, Professor Jake Grumbach. The interview has been edited for clarity and length. 


JG: I'm a professor of political science at UW in Seattle. I'm a faculty affiliate at the Labor Center there, the Bridges Center for Labor Studies. They bring together organizers and rank-and-file workers who are interested in these topics, and [interactions with those workers] informed the paper.

 I've always been interested in labor issues and generally been a part of social movements my entire life. I interned for UNITE HERE while I was in college. But the background is mostly family background. My grandfather, my mom's dad, was a Black journalist in Detroit and Chicago, an editor of the Michigan Chronicle and then the Chicago Defender, these Black newspapers. He was on the labor beat and one of his main causes was integrating labor unions and creating a multiracial labor-based movement. You can read some of his pieces as far back as the 1940s, talking about how Dixiecrats hated civil rights and were anti-labor. He had the pictures of Walter Reuther and Martin Luther King walking and these sorts of things, and it was a nice story to come back to this. And, actually, I wish he was still around to give some comments.

DK: How did you come to write this paper?

JG: So Paul Frymer and I met while I was doing a fellowship [at the Center for the Study of Democratic Politics at Princeton]. I loved his work; he's long been interested in historical qualitative research on labor and race issues. He has a book called Black and Blue about the labor movement and Black politics. It’s been a huge influence in my thinking over the years.  And I said, "Somebody had to have already studied whether labor unions reduce racism and racial conflict and make white workers less racist." I was shocked to find there wasn't this study yet.

DK: That's an interesting question to speculate on: why hasn't that work been done?

JG: In my discipline of political science, there's been siloed research in different areas. Intersectionality is such a hot buzzword now, but the intersection of race and class and thinking about an interracial labor movement in this way has just not been a focus of political science.

DK: The paper talks about racial resentment, and you're talking to me about racism. How are those two things connected to each other?

JG: Sociologists, psychologists, political scientists, to some extent economists, a lot of us are interested in uncovering racism as an attitude. But there are obstacles to finding explicit or implicit racism among people's attitudes in surveys. In surveys, people are unlikely to say, "Yes, I believe Black people are innately inferior as a racial group." There's what's called social desirability bias. Racial resentment is more specific. It's a list of survey questions, essentially, that are called the Racial Resentment Index, and what they get at is not overt racism, but rather questions such as, “Do you deny the existence of racism as an explanation for racial inequality?” “Do you agree with the statement that Black people would be just as well off as white people in the U.S. if they just worked harder?”

 A lot of the conservative push since the '60s has been about: government is an unfair giveaway from hardworking white taxpayers. So given that connection between race and economic policy in the US, I think these questions actually do get at a key part of racism.  

DK: And, as a researcher, is there also an advantage in having a set of questions that's been used in lots of other contexts, so that you have some kind of baseline?

JG: Such a great point. So, for example, if you want to see whether racial resentment is increasing or decreasing across time, then it's important to use the same questions so you can make apples to apples comparisons.

DK: The racial resentment questions in your paper are specific to Black Americans. Is that just because that's how the field's developed?  

JG: The racial demographics of the U.S. have changed through the immigrant waves post-1965 as more Latino and Asian Americans were integrated into the U.S. That wasn't the case initially when the racial resentment stuff really hit. And then also, the Black/white color line really does structure race in America.

DG: Let's talk about the data. You used data that already exists. Can you explain a bit about what that data is, how it gets collected, how you get to use it?

JG: In academia, there's a ton of inequality on who has the resources to field these huge expensive surveys, where you get thousands of people to answer questions. It's very expensive and people don't answer the phone much and [it] costs a ton of money. So there's been this push for more publicly available data sets. We used three key data sets. The first is the Cooperative Congressional Election Study, CCES, which is a very famous one that started in the mid-2000s. It was originally focused on whether politicians were being responsive to the attitudes of their constituents.

The second is the American National Election Study, ANES; that's been going since the 1960s. It's really extensive so it's really expensive, fewer individuals per wave of the survey, but it had those racial policy questions that we can talk about whether you support increased aid to Black Americans or affirmative action and such. And then the last one was this new survey, the Voter Study Group survey. That’s done by a set of researchers that were more specifically interested in studying racism in the U.S. and that started in, I think, 2012.

DK: You talk in the paper about panels. That's describing a situation where literally the same people are being interviewed over a period of time?

JG: These days, social scientists, especially quantitative social scientists, are really interested in whether you can make a causal argument rather than a correlation argument. If you remember back to high school, correlation is not causation, right? Does eating popsicles cause you to wear shorts rather than pants that day? No, they're correlated because it's hot out. 

Similarly here, it could be the case in our study that the white workers that join labor unions are more racially progressive and less racist on their own due to other reasons. They may be in industries that may be more unionized. They may come from places geographically that tend to be more progressive and where white people tend to be less racially resentful already.  

So those are what are called confounders that break down the causal argument. There's a high burden of proof in quantitative social science these days.

If it's non-panel it's called cross-sectional, where you're comparing people at the same point in time: a union member versus a non-union member who's similar in every other demographic way. But you can never get every demographic difference. You can never get every little thing about their childhood or their school. There are subtle differences that could still, again, mean this is a correlation not causation story. But the panel data means no, you're comparing that person four years ago to that same person now.  

So that's the important thing about our panel data; we were able to see, okay, what are people's racial resentment levels before they join a union and then after they join a union? 

DK: Right. And just to be clear, are we talking about people who identified as signing a union card or who simply said they are represented by a union? Because those aren't the same thing.

JG: I think it asks if they're a member of a union, which we interpret as signing a union card. In statistical terminology, it's a measurement error to the extent people just make a mistake when they respond that way. And people do. A lot of people think they're members of a union because the workplace is represented. But that measurement error actually cuts against us. That makes us less likely to find this effect of reducing racial resentment. So the fact that we still found people reducing their racial resentment after joining unions was additional strong evidence.

DK: Would it be helpful from your perspective to be able to fact check that? You kind of hope it's noise that cancels itself out, but as researchers, is that the sort of thing you would like to have someday?

JG: So this is true. Any surveying researcher would like the verified data. In studies of voter turnout, when you ask somebody, “Did you turn out to vote?,” there's all that social desirability bias to say, "Yes, I definitely did my civic duty," when you may not have. So people try to validate this with true voter turnout from the state voter files. But in other areas, for example, union membership or income, their self-reported versus their actual income, these sorts of things are not typically verified. There's not a ton of us labor-oriented researchers yet, but yeah, we'd absolutely like that.

DK: Because potentially that's something that a union could offer to a researcher. That we potentially could give you access to data or certain pieces of it could be verified as true or not because we're giving you our database with our members in it and that sort of thing.  

JG: This is actually a really tough sort of research problem that myself and others are thinking through. The most fruitful stuff I have seen recently is in the wave of teacher strikes, they were able to do surveys of teachers involved and teachers not involved and parents of kids in the schools with strikes happening and in schools without strikes, those things were really great. And those were done in conjunction with the unions. That is really exciting and could be the potential for more real partnerships with unions and researchers.

DK: I want to turn to your results. I would imagine that some of the people who read this interview are going to go look at the paper, and they're going to look at a chart like this. Even I, who theoretically have some training in this, my eyes glaze over pretty fast.




JG: Absolutely.

Dave: I want to ask a couple of questions to help someone who reads this to understand what they're looking at. At the top, you have these four columns, labeled 1, 2, 3, and 4. Are those different equations that you're using to measure?

JG: Exactly. Some of the models control for things like gender, income, geography.  Others don't. But the point here is you don't want a researcher secretly looking through every possible statistical model. In theory, we could run hundreds of different statistical models and pick the one that showed what we wanted to find, right? For that reason, you show a lot of different specific model specifications. Some do include controls for gender or different ways of measuring different things. It's doing the due diligence. It's not us picking some weird spurious thing out of hundreds of possible statistical models.

Then in terms of the results, the thing to look at is the scale of all these things, whether it's the racial resentment scale or the policies that benefit African Americans scale, they're all scaled from zero to one. When you see a -0.05 in the number that's not in parentheses, that means becoming a union member reduces your racial resentment by 0.05 out of one, or about 5% of the overall.

These are four-year panels. You get somebody; they join a union. Then you interview them again [four years later], and during that time, they reduce their racial resentment by 5% of the whole scale, which is actually pretty substantial.

DK: Is there a way we can compare that to, say, the impact of education? Somebody who graduates high school versus someone who has a college degree, what's their change? Is there one, and how does that compare?

JG: In our study, we can say the effect we find for joining a union is consistently about the same as something like education, going up a level in education, so going from non-high school graduate to high school graduate, high school graduate to two-year degree, two-year degree to four-year degree.

We were pretty surprised. I wouldn't be confident in making specific numeric comparisons, but I would say that the effect of union membership, we find, is at least as big as many social cleavages and identity cleavages in American politics, like the gender gap in attitudes.

DK: You mentioned near the conclusion that one of the things that you weren't able to do is break down what kind of union people were joining. That would be interesting to know, wouldn't it?

JG: Yeah, that would be a huge follow-up. We did do a breakdown. We weren't able to do it all that well, but we did look at self-identified professionals versus self-identified manual and service employees. These are vague categories, very broad. You don't see a huge difference, although we don't have a big enough survey to break it down that finely. But to the extent we do, we see that, really, all types of occupational categories see a similar reduction, but maybe a slightly greater reduction in racial resentment among the professionals.  

Same thing with geography, whether in the South or in right to work states. We find no real difference there, either. The decline of manufacturing unionism and non-college-degree, rank-and-file unions is a tremendous problem for American politics and the economy. It would be really important for us to see differences there, if they exist.

DK: You talk in the paper about the possible mechanisms by which this happened because, obviously, it's not some sort of osmosis. Something has to happen, and your survey data doesn't tell you that, but you speculate. Can you talk about what the possible mechanisms were?

JG: I’ve got to shout out all the historians and qualitative researchers and union organizers and labor journalists; they're all really crucial, on the ground, describing in qualitative, real-world terms how people understand themselves and their politics in labor unions and that allowed us to generate these hypotheses about it. 

We rule out some mechanisms. It's not income. It's not just that, "Oh, white workers get a little more comfortable, so they don't feel like they're competing with Black workers." It's not that. Income actually doesn't seem to help you in your racial resentment.

Rather, it's something about unions, and we think, first, it's just the legacy, like the incentives facing unions as organizations and union leadership. They have multiracial workforces, and to expand the union, you better have that workforce pulling in the same direction. That's the first practical term. That's obvious unionism.

The second thing is the way unions are organized. Not every union does this amazingly, but some are more connected. Some are more small-D democratic and really have more frequent political meetings and engage their memberships and train leaders from the rank and file. That's a powerful organizational mechanism. This happens at work, where people spend most of their lives.

Then, finally, it's not that the Democratic Party is some outstanding, shining example of anti-racism and unionism, but rather, making white workers less likely to be Republicans and therefore less likely to be watching Fox News and less likely to be engaging in Republican Party politics, that does probably help somewhat.

The Obama era had some key disappointments for labor. At the same time, you think about Richard Trumka, there was some really powerful anti-racism in his barnstorming for Obama. That was a narrative in 2008 and 2012, this idea that racism is dividing working people and a tool of the powerful. This was during the financial crisis, when it was just like, "Listen, as you white workers bicker and feel threatened by the Black workers at the shop, Wall Street is making away like bandits, and we have no organized power to go to battle."

DK: I also found it interesting that you found that there's a hang-on effect after someone leaves a union. Because you said people who were members of a union also scored lower on racial resentment than others [even after they left the union]. 

JG: That's exactly right. You do see some diminished effect. And we don't know exactly whether it's somebody retires, they tend to be older. It seems pretty plausible that when you're part of the union and you're really in this environment and milieu and politicized on union issues across race, that it persists somewhat, but may diminish a little over time as you're not in a union.

It's really important to think about the limitations of any of these studies. And this is why we need all hands on deck: qualitative, historical research, modern interviews, quantitative research, and surveys. Again, even though there are thousands upon thousands of people in these surveys, there's not all that many people that join a union in any four-year stretch. The percentage is pretty low, so it could just be, even though we do two different panel datasets, there's a small chance we just found a weird set of people that were trending less racially resentful and happened to join unions at that time.

DK: What might a union take away from your research that might change how a union does what a union does?

JG: We need more small-D democracy in unions. In the supplement, the appendices of our paper, we do include some of the literature from unions on racial solidarity campaigns they have waged. And not all of them do it.

I understand the trepidation, how the constant attacks on unions have really dismantled them over the past generation. Now unions have been backed into a corner. Taking that risk of trying to increase the democratic nature of the movement, engaging the rank and file in anti-racism, I think may have really been effective.

There's been a lot of hopeful signs. The teachers' strikes have been really impressive and were wildcats in some cases; this was risky and nobody knew if this would work, and it paid off huge in these deep red anti-labor states. This is the moment we just really have to have some courage.

We're in a really important moment for American democracy, and a vibrant, democratic labor movement is just more important than ever though. You can't just do managerial unionism at this point because you will be cut.

I'd say my bigger feedback is actually for the Democratic Party, that the false choice between race/gender and class has been extremely damaging for the Democratic Party. It's been self-defeating on both fronts. The neglect of the labor movement has led to a white identity politics and cultural politics taking over where labor politics once stood.

So if you want to fight racism, a vibrant labor movement is key. Even though, again, that's risky for Democratic politicians who want to have the tame civil organizations that they're aligned with. But I don't think we have time to be tame anymore.




Created with Sketch.

Related Articles