Americans’ views on race are polarizing based on our partisan sides. But that does not mean our views on race are the factor driving our political decisions. Our political views may be changing our racial attitudes—and even our racial identities. Andrew Engelhardt finds that whites’ prior partisanship often predicts changes in their racial resentment and racial group feelings more than prior racial attitudes predict changes in partisanship. Alexander Agadjanian finds that some Americans changed their racial identities in alignment with their 2016 presidential vote, switching from Hispanic or mixed race to white when shifting to Trump. Race remains central to our politics, but the relationship is more complex than we may assume. 

Guests: Andrew Engelhardt, University of North Carolina, Greensboro; Alexander Agadjanian, University of California, Berkeley

Studies:Racial Attitudes Through a Partisan Lens” and “Changing Votes, Changing Identities?


Matt Grossmann: How politics changes racial views and identities, this week on the Science of Politics. For the Niskanen Center, I’m Matt Grossmann. American’s views on race are polarizing based on our partisan sides. That’s normally interpreted to mean our views on race are driving our political decisions, but new research suggests the reverse causal chain. Our political views may be changing our racial attitudes and even our racial identities. This week, I talked to Andrew Engelhardt of the University of North Carolina, Greensboro about his British Journal of Political Science article, Racial Attitudes through a Partisan Lens. He finds that whites’ prior partisanship, often predicts changes in their racial, resentment and racial group feelings more than prior racial attitudes predict changes in partisanship.

I also talked to Alexander Agadjanian of the University of California, Berkeley about his Public Opinion Quarterly article with Dean Lacey, Changing Votes, Changing Identities? He finds that some Americans change their racial identities in alignment with their 2016 presidential votes switching from Hispanic or mixed race to white when shifting to Trump. They both say race remains central to our politics, but the relationship is more complex than we normally assume. Engelhardt took a broad view of which was causing the other, partisanship or racial attitudes.

Andrew Engelhardt: This article was interested in kind of the link between white Americans’ partisanship, so their identities as Democrats or Republicans and their views about black Americans. Conventionally within political science we think that this relationship really just runs one way from whites’ views about black Americans to their partisanship. So feeling positively about black Americans in a post civil rights era leads someone to identify as a Democrat and feeling negatively about black Americans in this same era, leads somebody to identify as a Republican. But I got curious to see if the relationship might run the other way that partisanship actually changes how people view black Americans and through a series of analyses, I found that this is indeed the case, that whether somebody identifies as a Democrat or Republican does lead them to change how they view black Americans, which is politically important and also important theoretically, within political science.

So some of the political implications suggest that thinking about coalition building with in politics, what political elites are saying, it’s not falling on deaf ears. So when we think about Republican dog whistles, for instance, their partisans are hearing these often either disparaging, implicit or explicit remarks about black Americans, and then Republicans are actually incorporating this into their attitudes. It’s not just that Republican elites are trying to make race salient, they’re actually substantively changing what their underlying attitudes are for their mass partisans. Conversely, for Democrats, a lot of coalition building about addressing racial inequality is landing on sympathetic ears. So discussion of structural racism, concern with black American status in American politics, is not leading white Americans necessarily to defect from the party, it’s leading them to grow sympathetic and adopt these views, opening up the Democratic party to maybe being able to champion policies that black Americans have been pushing for as part of the party, but the party has been otherwise opposed to.

Within political science, it offers some kind of theoretical importance, knowing that views about social groups, which we conventionally view as kind of forming early in life through childhood socialization and then leading to politics, knowing that these views can actually change in response to politics, leads us to kind of be a little bit more precise in terms of how we think about the links between these attitudes and political outcomes, knowing that prejudice is endogenous, that prejudice can actually follow from politics opens up a whole door for avenues of attitude change. I think knowing that political elites aren’t just trying to activate these attitudes that they can change them, I think gives us as political scientists are really big swath of opportunity to think about under what conditions can the views that the advantage have of the marginalized change, or conversely under what conditions can elites perpetuate marginalization through kind of reinforcing negative views of these marginalized groups.

Matt Grossmann: Agadjanian finds people change their racial identities and politics might be at the root.

Alexander Agadjanian: In this paper we’re interested in whether political factors can contribute to this dynamic of racial fluidity, and this is the idea that self-expressed racial identification is a flexible trait that can change at the individual level over time. And so the background here is that this concept is commonly seen as a stable trait at the individual level that influences other variables, but itself is not influenced by their factors in turn. And that’s kind of contradicted by growing evidence that a small but notable portion of the population changes how they racially self-identify over time. And so, for example, linked censuses between 2000 and 2010 find about six percent of Americans change their identities. More recently, I found that 8% changed from 2016 to 2020. And so given how important a variable like race is and how much data points like this kind of overturn conventional wisdom, I think scholars have gradually begun to wonder what causes might precipitate this change and a lot of that’s been confined to [inaudible 00:05:18] and economics and pointing to factors like social context and pressures and economic incentives.

But for the most part, it’s been overlooked in the context of political science and political factors that might cause this. And so the purpose of my paper with Dean Lacey is to come in and say that there might actually be good reason to believe politics could be at least one factor contributing to racial fluidity, and that’s because political and social identities like race have over time become increasingly intertwined in terms of party coalitions and people’s individual psychologies and political scientists have also shown that more and more political attachment shape outcomes beyond the purely political realm and that sometimes people update their other social identities to fit with their partisanship, and so building on this work, we think that 2016 election in particular might have created conditions ripe for politics induced racial identity change.

So you have this historical backdrop of growing social political identity alignment, but in 2016, what really happens is that racial identities and attitudes have become so much more salient in this election season than ever before. And some scholars have said that campaign rhetoric framed the election around identity and that rhetoric from Trump embracing white identity and nonwhite hostility, contrasted with racial liberalism and elevated concerns of racial minorities by Hillary Clinton on the other side. And so we think that group pressures related to race and identity led Americans, especially those who are new on each voting side, feel maybe the greatest pressure to conform to change their racial identity to better fit with their new vote allegiances.

And so to test this, we turn to voter, the voter study group panel survey. And so these are reinterviews of the same Americans across 2011, 2012 and 2016. And so here we can measure people who are changing the racial identity from 2011 to 2016, we can measure people who are switching their votes. And so this is kind of providing us with everything we need to examine our theory. In terms of results, we find basically partial support for our expectations. So for 2011, non-whites who didn’t vote Republican in 2012, switching over to the Republican side with Trump on the ballot in 2016 led to an increase in adopting white identity five years later. And so this is robust to various controls and we find that it’s also driven by mixed, originally mixed race and Hispanic identifiers. So there’s other people who change their race too, but these are two groups that really switch their votes and their race in tandem.

But on the other hand, we find that there’s not much support on the other side of things in terms of new democratic voters are not that much more likely to shift from white to non-white identities. This pattern on the Republican side does not really show up to the same extent in an earlier election period from 2008 to 2012. And so that does seem like something unique is happening in 2016. And so the big takeaway from our study then is that we do find evidence that vote choice and racial identities evolve together in expected ways. And so racial minorities who switch to Trump become more likely to identify as white. And so even though there are some limitations, we think this is suggestive evidence that politics might actually have a role in racial fluidity.

Matt Grossmann: They both use panel data. Engelhardt tries to figure out what causes what by looking at change over time in both.

Andrew Engelhardt: Cross lagged models are just a statistical technique to, and we assess kind of average levels of change in an attitude over time. So in this article, I look at panel data. So it has the same individuals reinterviewed at two points in time. In both waves I have measures of their views of about black Americans and measures of their partisanship, so their identifications as Democrats or Republicans. What the statistical models allow me to do is say given where some he was at time one in terms of the racial attitudes, for instance, does knowing whether or not they identify as a Democrat or Republican at time one give us any information at all in their racial attitudes at time two? The statistical model allows us to assess then the degree to which racial attitudes measured at time two are associated with prior partisanship above and beyond some of these initial views about black Americans or vice versa for changes in partisanship driven by racial attitudes.

So the advantage then is that I can kind of assess change over time on average. And then also in this framework assess which of these constructs in a relationship, partisanship, or racial attitudes is doing a little bit more to explain the increase in correlation between these two concepts that we see just in cross sectional data. One of the challenges though, and especially for this comparison is partisanship driving dynamics, or is racial attitudes, driving dynamics? One of the challenges as with all analyses is measurement error, where if we’ve poorly measured our variables of interest, then we’re going to underestimate, for instance, the temporal stability or how kind of crystallized and persistent the attitude is, which might lead us to then falsely conclude some other construct, like partisanship is leading to changes in that attitude. So one of the things I was sensitive to in this paper and did a lot of supplementary analysis to look at is just to make sure that what I’m finding isn’t just due to differences in how well we measure different constructs.

Another challenge that these models face is simply with any sort of other observational study, the challenge of omitted variables. There could be other factors that like first point in time that are beyond partisanship or beyond racial attitudes that are leading to changes in the other, but it seems that through some additional analyses I’ve done, the results are robust to considering these other potential alternative explanations, like values or views about immigration or other salient policy debates.

Matt Grossmann: Agadjanian uses panels to look get changes in both race and voting.

Alexander Agadjanian: So the key benefit of the panel data we’re using here is basically we’re able to measure the dependent variable we’re most interested in. So this whole conversation is about overtime variation in racial identity. And so let’s say a single survey at one point in time, doesn’t really allow us to capture what we care about. Panels that are asking race, let’s say in 2011 and also five years later in 2016, that can allow us to pinpoint these people who are changing their identities over time, and in a similar way, on the independent variable side, we know who is switching their votes from one election to the next. And so this is based on where are the action we think is happening? Panels allow us to capture this best.

On the limitation side, I think there are some important ones to consider. And I think the main one is that we’re still limited in the strength of causal claims that we can make. So maybe there’s, I think maybe the most important one there is the potential for reverse causation. So maybe instead of people switching their votes and then changing their race and response, maybe the reverse is happening, some racial assimilation happens first. People become white in their identities and then change their votes in response. I think we can’t really disentangle that well. Putting that aside, I think there’s still value in establishing just the correlational nature of the relationship and just documenting that race and vote change together, still ought to revamp earlier understandings of how these two variables interrelate and how we understand race.

Matt Grossmann: Let’s dig into each study, starting with Engelhardt’s. He did find evidence that racial resentment drove partisanship in the 1990s.

Andrew Engelhardt: So I think that the results in 1992 to 1994, where I find that whites who view black Americans negatively according to a measure of racialism. So just given this structural discrimination versus individual effort, explanation for black American status, that people who were higher in racial resentment became more likely to identify as Republicans in 1994, relative to where they were in 1992. But throughout the paper and including these analyses, I also look to see if just simple viewing black Americans more negatively relative to white Americans, a measure of just affect, so relative favorability, whether that also offers a similar picture into dynamics. And I don’t find any relevance for affect, which is interesting because we would think that both of these two measures were capturing something similar. But the things that I take away from this is that it’s actually an advantage of the more political nature of racial resentment.

A lot of what goes into this measure can be highlighted through one of the questions that is often used to measure it. If black Americans would only try harder, they would be just as well off as whites. It’s, in part, an attribution of black American status. One of the key policy debates between 1992 and 1994 is featured in the Gingrich revolution in Newt Gingrich’s contract with America, which focused on welfare reform and kind of the merits of providing assistance to folks in need, where a lot of the critics of welfare policy and a lot of the proponents of welfare reform focused on the lack of deservingness and a need to get basically free loaders and moochers off of public role, public dole. A lot of that rhetoric then is directly targeting this racial resentment construct. So it makes sense then that racial resentment would show changes or sorting with individuals moving to the Republican party based on these attitudes and affect wouldn’t, just in part because one measures actually more directly capturing the types of rhetoric that politicians are offering while the other one is a little bit more removed from the political climate.

Matt Grossmann: But the Obama era instead mostly moved partisans to adopt new racial attitudes.

Andrew Engelhardt: I don’t think that a lot of the stories about the Obama administration, at least the early, the first term aren’t necessarily wrong in terms of Obama racializing the political environment and leading some whites, for instance, to switch from the Democratic to Republican parties, a production of the Tea Party backlash to the Obama administration’s policy agenda. But part of this environment where you have a prominent black man and Tesler’s argument, Michael Tesler’s argument, racializing politics, you also have a prominent black politician who’s the head of the Democratic party that signals to Democrats that we have a positive out group exemplar. So he’s a positive role model of what black political leadership looks like that for most Democrats is going to be hard to ignore. And that just leads to more positive views about black Americans, where people are changing their views based on this prominent individual that they see either on the nightly news or get information about secondhand.

One of the things that’s interesting, and especially in terms of thinking about the 2012 election, is that in thinking about the competition between Obama and Mitt Romney, it’s a lot of focus on social class and economic and equality kind of around the Occupy Wall Street movement, focus on Romney’s past with being capital. But I do find that there’s actually a fair bit of racial attitude change in this period. One of the things that I think about with this is that some of the rhetoric about deservingness through economic inequality could be activating kind of like deservingness around who should or should not get government assistance, which implies black Americans based on the racialization of these policies, including either the connection to then President Obama or just legacies in terms of Martin Gillen’s work in terms of the racialization of welfare and federal spending programs.

One of the things I find interesting too, is that a nice article by Sean Westwood and Eric Peterson, they show that there’s a close connection cognitively between race and political party where changes in affect with parties lead to also changes in racial affect. So even if the 2012 campaign was economic, and on its face it really was, just simple political competition could be sufficient to change partisan views about black Americans just given the close connections between these two things based on decades long political competition around the status of black Americans.

Matt Grossmann: Racial attitudes continued to change with partisanship through the Trump era.

Andrew Engelhardt: Finding continuing patterns through Obama’s second term office and also through the 2016 election, and in fact, in analyses, not in the paper, but looking at the 2016 to 2020 AENS panel, I find again, consistent evidence. So it’s not just an Obama effect, but I also don’t think it’s exclusively a Trump effect. One of the things that is hard to suss out with the analyses that I have is the timing of change between for instance, 2012 and 2016. All I know is that attitudes changed sometime in this four year period. With a little bit more fine grained data, Pew has a question that they’ve asked that basically captures racial resentment, which just asks is black American status due more towards structural discrimination, or is it due to a lack of individual effort and like individual will?

There’s actually a shift among Democrats in terms of the proportion endorsing structural discrimination. In fact, a majority of white Democrats endorse structural discrimination, starting in about 2014, 2015 in the wake of Ferguson. So one of the things I think we miss in terms of emphasizing Trump is that there’s a lot of other stuff in the information environment that similarly speaks to the status of black Americans. We have the whole Black Lives Matter movement. We have the Democratic campaign in Hillary Clinton building on the Black Lives Matter movement. The 2016 campaign was the one that was quite open about advancing and advocating racial inequality, with then Secretary Clinton talking about implicit bias and structural discrimination during a televised presidential debate, which to that point is quite important for signaling the party’s awareness and recognition of these issues. So while Trump’s part of it, right, the Clinton campaign could decide to have done, taken the strategy, given who they were competing against. Any focus on even the 2016 campaign then misses the preceding context, reactions to an Obama, the rise of the Black Lives Matter movement and elevation of the experiences that black Americans are having to a national consciousness.

Matt Grossmann: Attitude change is partly due to changing norms, but there is some real change.

Andrew Engelhardt: Political leaders are in part providing just information. It’s not saying that I endorse X policy or oppose X policy, it’s just in their speeches and their rhetoric, they’re just talking about or elevating or providing attention to the status of black Americans. So that’s one way to think about this as being real, rather than kind of just parroting back. It’s information rather than specific endorsement or opposition to a policy. But more generally we could think about this, and I like to think about this in some ways as political elites, creating group specific social norms. So if Democrats are advocating or discussing racial inequality, it signals to Democrats in the mass public, okay, we should be more kind of sensitive to the status of black Americans, the status of people of color in the United States that could lead to parroting backed views that are reflective of this norm, but people haven’t actually internalized it.

So in other work I look at, and say that this is kind of insincere, but party driven change, that there’s a group specific norm that people should that understand that they should meet and they don’t want to appear to not be toeing that norm. The complication then comes from, if this group specific norm exists, people could actually internalize it. Say, we know that as Democrats, we should pay attention to racial inequality. I care a lot about being a Democrat. So I’m actually going to internalize this. I’m going to try go out and actually care about these attitudes. So that’s where I think the difference between follow the leader, which establishes a group specific norm that people may or may not care about, and actually a follow a leader with internalization that people actually care about and are at least voicing attitudes that are consistent. What I’m still trying to work on, in upcoming work is the degree to which that these attitudes actually change behavior, or actually when the rubber hits the road that they’re going to affect people’s political decision making.

Matt Grossmann: Partisanship also drives racial attitude change among racial minorities.

Andrew Engelhardt: If I’m claiming that a lot of what’s going on is an information environment account, we should find similar dynamics for other groups. It’s not just exclusively why it’s changing their attitudes about black Americans. So I focus on data in the second term for the Obama administration, so 2012 to 2016 and also in 2016. And I find similar sort of reciprocal effects for both black Americans, Latino Americans and Asian Americans with more evidence for partisanship driving racial attitude change, than really racial attitudes driving changes in partisanship, which is interesting to think about. At first we might question whether these analyses are worthwhile because I’m again using this racial resentment measure, which was constructed to measure whites’ attitudes about black Americans, but work by Cindy Cam and Camille Burge suggests that the measure is interpreted in similar ways between black and white Americans.

And in some other analyses, I’ve done that I’ve not published, I’ve found that it works in a similar way comparing Latinos or comparing Asian Americans. So it seems to be capturing something meaningful. And then my interpretation of the result, that party structures or changes racial resentment for even non-white Americans, is that a lot of what this attitude is, is constructed by the social and political context that people are in. It is in part formed early, but it’s reinforced through what people glean from the political environment. If politicians are either elevating and celebrating a group or denigrating that group, this type of rhetoric is going to be incorporated into people’s attitudes, not just white Americans.

Matt Grossmann: Agadjanian and Lacey looked beyond racial attitudes to racial categorization, finding that few people switched their race, and few people switched their partisan votes, but the switches that did happen were strongly related.

Alexander Agadjanian: Of non-white respondents in 2011, about 4.6 changed their identity to white. It’s about one percentage point lower on the other side, in terms of original whites changing to non-white. On the independent variable side vote switching, we have a little bit more movement where around 14% of people who didn’t vote Republican, either voted Democrat or didn’t turn out in 2012, 14% switched to the Republican side in 2016, whereas let’s say around 11% of original Non-democratic voters switched to Democratic column in 2016. On both the independent and dependent variable sides, there aren’t a lot of people switching and especially in terms of their race, people are still pretty stable over time. So that’s something to keep in mind in terms of how we’re checking the relationship between these two variables.

To put a number on it, the models we’re running in the background here, one numeric takeaway from our results is that among originally non-white non Republican voters probability of changing one’s racial identity between 2011 and 2016 goes from 0.03 to 0.49, for those who switched into a Republican vote in 2016. So it’s an increase of 0.46 from the original likelihood of changing their identities. So it’s pretty sizable, but again, these are still small numbers to start, and so that should be an important thing to keep in mind in the background.

Matt Grossmann: The changes were concentrated among more racially flexible identities.

Alexander Agadjanian: It’s happening among both mixed race and Hispanic respondents. I think it makes sense for different reasons for both cases. So for example, mixed race Americans, obviously there’s naturally a lot more flexibility in how they identify, even though they have mixed parentage, they may still identify with a single monoracial identity as Oren Davenport’s work explores in detail. And for Hispanics, on the other hand, the government in the census often seems to tell them that they’re white and assumes that they’re white and that, and that might give them some license to cross this group identity boundary. And there’s also some interesting data that Hispanic identity decays across generation, and so maybe as they assimilate more into American society, they have more reason to move into a white label.

But still I think unclear to us, and we don’t really have a good reason as to politics is really affecting these groups. It’s clear why they undergo racial identity change, but it’s not totally clear why politics is driving change here specifically. One note is that among other large non-white subgroups like blacks and Asians, there’s not even enough movement into the white category to look at this well. And so that might be a factor, maybe if there’s more movement in the future, then maybe political factors and vote switching or, or partisanship will tend, will ultimately affect them as well.

In terms of the nature of asking about racial identity, in this survey respondents were basically offered to choose between white, black, Hispanic, Asian, Native American, mixed, other, and Middle Eastern, and only allowing them to choose one of those many options. And so it kind of speaks to the wide variability in how survey approach asking about race. Sometimes they ask about Hispanic or ethnicity separately, and don’t include Hispanic in the race question, which often confuses Hispanic respondents anyways. There’s still a lot of flex in how we ask these questions. I think it still works in our case, even this single choice forced response because the mixed label is still there for people with more flexible identities and with more than one racial identity attachment, in addition to the other label that allows for follow up written in identities. And so I think this is still, and it’s still also important that it’s signaling a primary racial identity attachment.

Matt Grossmann: People also change other identities along with partisanship, but this result was more specific to Trump.

Alexander Agadjanian: I think this, our work is pretty related to that by Michelle Margoles and Dave Campbell and coauthors, and Pat Egan as well. I think all of us kind of have a similar type of mechanism in mind that when political passions become closely linked and almost in set from social identities, whether it be race or religion, especially evangelical Christianity, group pressures, and desires to just resolve cognitive dissonance for people within congruent combinations of that, of identities, might impel people to really sort into more politics consistent social identities.

In our case, there’s also an interesting wrinkle in a few ways in that for one, we have kind of more of a triggering event in mind because of the 2016 election that might lead to this politics driven identity alignment, and separately, we are also account for partisanship and changing partisanship over time to see if that kind of fits with the story of these other authors. And it turns out that the Trump switching whitening relationships still holds when controlling for partisanship change and changes in partisan identity are not really what’s correlating with racial identity change, but instead it’s partisan voting. And so maybe there’s a story here that behavioral change matters more than attitudinal change. And so it’s a bit different from the other cases in that sense, although I wouldn’t rule out partisanship as a cause in other context. And so maybe it just failed to be a trigger in this case, but it’s not totally inconsistent with others’ accounts.

Matt Grossmann: Engelhardt says the findings fit together well, but also point to the current political context.

Andrew Engelhardt: One of the things I like about Alexander Agadjanian’s piece and his co-author, is that it highlights how the present political context is providing again, a wealth of information about the status of different and racial and ethnic groups in the United States. My implication from that is people are changing their views about these groups where the implication that they have is that people are actually distancing themselves, for instance, from groups that are marginalized or thought less of, that they don’t, they can’t derive the self-esteem that being a part of that group provides in the present political context.

One of the things that I think is important to note though, and I try to do that through my article is that a lot of these findings are within a certain political context, party driving changes in racial identity party, driving changes in racial attitudes, it’s found in a unique political context. It’s not a law of politics. We’re just in an environment where there’s a wealth of information about racial and ethnic groups, people are attached to their parties, and the implication then is that what we thought might be unmoved movers are actually susceptible to change, but that’s, again, from these specific set of evidence that we have available, which is in the post 2000s era

Matt Grossmann: Agadjanian and also of thinks the work fits together to overturn assumptions.

Alexander Agadjanian: I think my work with Dean, pretty related to Andrew’s work and his findings, it’s maybe not to the extreme extent as it is for race, but up until a few years ago, if you look at how racial attitudes and measures like the racial resentment scale were discussed for decades in political science circles, we see the same type of longstanding assumptions of stability, that racial attitudes are acquired early in life, that stay constant over time, discern an influence on other beliefs and decisions, but are robust to being shaped themselves.

And so Andrew’s work is so valuable in overturning this conventional wisdom in kind of the same way we seek to do. And so I think it fits pretty well and just reinforcing each other in terms of race related variables especially, like identities and attitudes should not be considered as unmoved movers as political science has long conceived of them. And it’s echoing some other points from, again, Michelle Margoles, Pat Egan, and also even [Bird Backer 00:32:01] and others in terms of personality traits that it’s just critical to think more broadly of political attachments and decisions as drivers of other beliefs, identities, and traits, where previously the causal era was thought to be unidirectional.

Matt Grossmann: In fact, Engelhardt says the finding suggest that many analyses of the causes of vote choice, including in 2016, might overestimate the effects of racial attitudes.

Andrew Engelhardt: I think my results highlight some challenges that analyses of vote choice face in the present political context. So conventionally analyses of vote choice are built around cross sectional survey data, like single shots in time, which in the past, if we thought that views of social groups, like racial prejudice or something, is formed prior to politics and pretty sticky, we can just put it in as a independent or predictor variable and look at its relationship with vote choice and be pretty sure that vote choice isn’t also causing racial prejudice, that there isn’t this reverse causality problem. My analyses suggests that we actually have to be more sensitive to the potential for reverse causality, even though I’m not looking at people changing their racial attitudes based on vote choice, in the present political context because vote choice is tied up so closely with partisanship it’s effectively part of someone’s political identity.

So then we’re looking at, in a cross section, a relationship between two things that my analyses show are dynamically related, racial attitudes lead to changes in partisanship leads to changes in racial attitudes. So that suggests that if we want to understand patterns of vote choice, we have to be really concerned with what types of research designs we’re using, what types of data we’re using to actually understand what’s leading people to choose one candidate over another. Simple reliance on a cross-sectional survey and looking at a Hardy correlation between racial prejudice and vote choice. Doesn’t tell us really whether racial prejudice is leading to vote choice, or if somebody’s vote choice leading to their level of prejudice.

And then if we’re comparing, say cross sections over time to say 2020 had, for instance, more or less relevant to racial attitudes relative to today’s 2016 or 2012, if we’re comparing each of these elections cross sectionally, we run into a problem where we might interpret changes in the correlation between some variable and vote choice as sensitive to priming, that there’s an election specific feature that say elevated the relevance of racial attitudes in somebody’s decision making calculus, but it could also be based on my results, a stronger relationship simply because people are better aligned between where their racial attitudes are and where their vote choice is. So that change in the correlation over time, isn’t due to an election specific thing, it’s just this secular trend of people changing their parties to fit with their racial attitudes, and also more importantly, changing their racial attitudes to fit with their partisanship.

Matt Grossmann: But there was a key difference in the results. Engelhardt found that more of racial attitude change is among Democrats than Republicans.

Andrew Engelhardt: I found change in racial attitudes among both white Republicans and white Democrats. So on average, Republicans are becoming somewhat more negative in their views about black Americans. Democrats are becoming more positive in their views about black Americans. But what I find is that much more change on average is occurring among of Democrats, and I think that makes sense. The information environment what’s new is a lot of information about the status of black Americans, highlighting, for instance, features of racial inequality that may have been unknown to a lot of white Democrats who are sympathetic to these arguments.

In contrast, a lot of the rhetoric like Trump’s is honestly part of what a lot of American politics has looked like for decades, if not as overt as a lot of Trump’s rhetoric is. So what Republicans are offering is consistent information with a lot of people’s attitudes. So there’s not a lot of room to change. There’s not a lot of new information for Republicans to pick up and incorporate into their attitudes. They’re already negative. For Democrats, there’s a lot more movement, either ambivalence or negativity moving toward positivity.

Conversely too, in terms of thinking about attitude change, we could also think about whether racial attitudes are, doing more to change people, to become Democrats or change people to become Republicans. I find not too big of a difference between these two, but there is more evidence toward folks with negative attitudes, becoming Republicans as being a little bit more of a weight than people with more positive attitudes becoming Democrats. And I think that speaks to some of the tension that the Democratic party right now is experiencing at least with a lot of prognostication of thinking about where the party should go over whether they should advocate for issues championed by [inaudible 00:36:49] the black community, more progressive activists, or try to hold on to some more of the, depending on how you want define it, white working class or white non college educated individuals.

Matt Grossmann: But Agadjanian found that more are switching their racial identity when they become Republican voters.

Alexander Agadjanian: In terms of understanding the asymmetry, the first point to make is that we think it’s not just Trump’s campaign rhetoric that’s at work, but it’s also what’s going on the Democratic side, for example, like John Sides and coauthors find in their book on the 2016 election that voters increasingly perceive Clinton as more racially liberal than previous Democratic candidates, and so this strong message of being white is important to what Republican voters are and racial liberalism and elevating the concerns of racial minorities on the Democratic side, both are contributing to this process.

But in terms of the asymmetry specifically, even though it didn’t fit with our hypotheses that we made beforehand, I think it’s still pretty understandable in light of our theory of group pressures that are at work here. Specifically, the pressure for group conformance is likely stronger for 2011 non-whites entering a Republican voting block that is 88% white, compared to 2011 whites entering a Democratic voting block that is 40% non-white. And so the varying levels of racial homogeneity across parties are probably the key to understanding the asymmetric result, and that’s also supported by other work from Lily Mason and Julie Roski, who argue that identity concurrence pressures are stronger on the Republican side because of this group homogeneity element.

Matt Grossmann: He says the findings may fit with changes in 2020 as well, including Trump’s gains with Latinos.

Alexander Agadjanian: There’s still some interesting trends that might fit well with our story. So for example, I think for some data indicating that in 2010, about half of Hispanics identified as white at a national level, but in some of these swing counties that went to Trump, especially on the Texas border, white identification among Hispanics was a lot higher. For example, Star County almost fully 99% white identifying and Hispanic ethnicity rate was 96%. So maybe while other Hispanics across country are distancing themselves from whiteness, those swinging Trump are reinforcing their white attachment, and that’s just more speculative and definitely more work to be done in terms of how people are reconciling these two elements of their racial identity, especially among Hispanics in the country.

Matt Grossmann: There’s a lot more to learn. The Science of Politics is available biweekly from the Niskanen Center and part of the Democracy Group Network. I’m your host, Matt Grossmann. If you liked this discussion, you should check out our previous episodes with many on related topics. Did Americans’ racial attitudes elect Trump? The roots of the party’s racial switch, racial protest, violence, and backlash, multiracial electoral coalitions for minority candidates, and racial stereotypes in voting for Obama and Trump. Thanks to Andrew Engelhardt and Alexander Agadjanian for joining me, please check out Racial Attitudes Through a Partisan lens and Changing Votes, Changing Identities? and then listen in next time.

Photo credit: iStock