Sorry Democrats, your party has racial bias problems too (Part 2)

By Kent R. Kroeger (May 18, 2018)

[This is Part 2 of a two-part article on how racial/ethnic attitudes permeated the 2016 election. Part 1 detailed the issues and complexities of efforts to understand racial attitudes and how they impact the political arena. Part 2 details my own research on this topic with respect to the 2016 presidential election.]

Political pundits are not shy about calling Donald Trump and his supporters racist. From CNN’s Van Jones’ “white lash” scolding of Trump voters on election night to recent news articles on political science research showing ‘racial resentments’ were a strong driver of the 2016 presidential vote, it is likely race will also be a potent issue in the 2018 midterm elections.

A recent dialogue from a public forum hosted by The New Yorker magazine exemplifies this ‘Trump voters are racist” narrative:

“Racial resentment” is largely what Rankine is talking about, and her hypothesis on its power is echoed in quantitative research from the 2016 election showing ‘racial resentments’ were a powerful driver of vote choice (along with sexism).

In another recent public forum on racism, hosted by journalist Robert Fieseler, Harvard University’s Dr. James Herron offered this explanation as to why race continues to underlie many of our national conversations: “Race or racial ideology runs deep in our history and culture. In certain ways, it’s at the core of our political culture. Our identities are shaped by race. So, given its centrality in our history, it’s not surprising that it continues to be relevant.”

According to Herron, racism is the “lens through which people interpret, naturalize, and reproduce inequality.”

While appealing in its succinctness, many academics who study racism consider Herron’s “dictionary” definition far too narrow to capture the importance and centrality of racism in American society.

“Racism refers to a variety of practices, beliefs, social relations, and phenomena that work to reproduce a racial hierarchy and social structure that yield superiority, power, and privilege for some, and discrimination and oppression for others,” writes Dr. Nicki Lisa Cole, a sociology professor in the department of Social Policy and Social Work at the University of York (England). “Contrary to a dictionary definition, racism, as defined based on social science research and theory, is about much more than race-based prejudice — it exists when an imbalance in power and social status is generated by how we understand and act upon race.”

Cole says racism manifests itself in seven forms: representational, ideological, discursive, interactional, institutional, structural, and systemic.

Given racism’s broad scope and impact on American society, the following analysis is admittedly narrow and merely offers evidence that negative racial attitudes are not wholly determined by partisan orientation.

While trends are positive, negative racial attitudes remain ubiquitous.

Part 1 of this article discussed some of the recent studies showing ‘racial resentments’ as central to the vote decisions of many Trump voters.

Though these cross-sectional studies had methodological limitations, I still found their evidence supporting the ‘racial resentment’ hypothesis to be convincing.

Nonetheless, these same studies implicitly reinforced the narrative that the vast majority of racists voters are Republicans, despite the fact considerable research has shown negative racial attitudes cut across party boundaries.

For example, in 2014,’s Nate Silver and Allison McCann analyzed two decades of research on racial attitudes by the General Social Survey (GSS) and concluded:

“There are white racists in both parties. By most questions, they represent a minority of white voters in both parties. They probably represent a slightly larger minority of white Republicans than white Democrats,” they wrote.


The good news is, negative racial attitudes among white Republicans and Democrats have been in a steady decline since 1990, with the drop stronger among white Democrats than white Republicans.

However, the 2016 presidential election reinforced the belief among opinion leaders that Trump and the Republican Party have a near-monopoly on racist voters.

But is that true?

What is the distribution of racial bias in the American vote-eligible public?

To address this question, I conducted a simple analysis of the 2016 American National Election Study (ANES) regarding Americans’ favorability ratings towards White, Black, Hispanic, and Asian race groups (the dataset is available here).

The key question utilized from that survey was the following:

Using a thermometer scale, how would you rate {Insert racial group: WHITE, BLACK, HISPANIC or ASIAN}? Ratings between 50 degrees and 100 degrees mean that you feel favorable and warm toward the group. Ratings between 0 degrees and 50 degrees mean that you don’t feel favorable toward the group and that you don’t care too much for that group. You would rate the group at the 50 degree mark if you don’t feel particularly warm or cold toward the group.

It is a blunt question — lacking any nuance — that condenses complex social constructs like race and ethnicity into four crude racial categories: Whites, Blacks, Hispanics and Asians.

For example, there isn’t just one Hispanic racial/ethnic group in this country. But, in opinion research, you are forced to work with what the researchers give you. And in the 2016 ANES, one Hispanic racial/ethnic category is what we are given.

Additionally, I concentrated on ‘racial favoritism’ with respect to Blacks and Hispanics only as they are the largest minority groups in the country. Any racial favoritism bias towards Asian or white Americans was excluded from this analysis, though it clearly existed within the data.

Conversely, I included measures of racial favoritism bias for Blacks towards and Hispanics and Hispanics towards Black, as research has shown significant tensions between these two population groups on a number of social issues.

How the ‘racial favoritism’ indicator was measured in the 2016 ANES

Overall, the computation a racial favoritism indicator for each respondent was determined as follows:

For white and Asian respondents I subtracted their racial favoritism score for their race category from their favorability score for Blacks and Hispanics. White and Asian respondents therefore had two racial favorability difference scores (one for Blacks and one for Hispanics). A positive score on either indicated higher favorability for their own race.

For Blacks and Hispanics, the same methodology was employed, however, only one favorability difference score was computed.

I further simplified the analysis by focusing on absolute differentials in the thermometer favorability ratings. An Hispanic respondent giving a ‘20’ favorability rating to Hispanics and ‘10’ to Blacks was analytically equivalent to an Hispanic respondent giving ratings of ‘100’ and ‘90,’ respectively. The racial favorability bias score was ‘+10’ in both cases.

In the final step, the respondent was coded a ‘1’ if either of their favorability difference scores were positive (indicating a favoritism bias for their own race). All other respondents were coded a ‘0’ (indicating no racial favoritism bias).

The Results

The difficulty in measuring racial bias directly is that most people don’t like to think they are ‘racist’ or ‘biased’ in general. When faced with race bias questions, many respondents will seek the ‘socially desirable’ response over their own personal attitudes. This could lead to significant measurement error, particularly a possible under-count of racial bias in the population of interest.

In my analysis of the 2016 ANES, a large percentage of respondents exhibited some level of racial favoritism. I also found this favoritism correlated significantly with other attitudes associated with racial bias, which gives some tentative support regarding the validity of racial favoritism bias measure.

In fact, across the 3,600+ respondents in the 2016 ANES, roughly 41 percent showed some level of racial favoritism towards their own race versus Blacks and/or Hispanics.

The Results

Figure 1 (below) shows the percentage of respondents in the 2016 ANES who favored their own racial group over Blacks and/or Hispanics within each presidential candidate’s voting bloc.

Forty-eight percent of Trump voters favored their racial category over Blacks and/or Hispanics.

But here was the surprise, at least to me. Thirty-two percent of Hillary Clinton voters favored their own racial group over Blacks and/or Hispanics.

Thirty-two percent is nowhere close to zero.

Figure 1: Racial Favoritism by 2016 Presidential Vote Preference

And for those that think non-voters are a greater opportunity for Democrats than Republicans, they may be disappointed to learn that non-voters looked more like Trump voters than Clinton voters when it came to the racial favoritism bias.

When analyzing just ‘white’ respondents, the racial favoritism bias numbers didn’t change significantly, except for Clinton voters, where only 24 percent favored their own race to Blacks and/or Hispanics (see Figure 2 below).

Again, 24 percent is not zero. There are a significant number of Democrats who favor their race over Blacks or Hispanics.

Figure 2: Racial Favoritism by 2016 Vote Preference (Whites only)

If you are wondering how racial favoritism bias varies by the respondent’s racial group, see Figure 3 below. Asian-Americans showed higher levels of racial favoritism. The other three groups were statistically similar.

Figure 3: Racial Favoritism by Self-identified Race

Finally, I looked at racial favoritism bias by strength of party preference. No big shocks. As you might guess, ‘strong’ and ‘not very strong’ Republicans showed higher levels of racial favoritism (roughly 50 percent), but “not very strong” Democrats aren’t too far behind.

Figure 4: Racial Favoritism by Party Preference

So, there you go. Racial favoritism (and if you want to call that racism, do so at your own risk) is hyper-prevalent in American society. In terms of percentages, racial favoritism is more likely to be found among Republican voters, but what is striking is how many racial favoritists exist throughout American society.

Further validation of the ‘racial favoritism’ construct

One last analysis I did using the racial favoritism indicator was to see if it correlated with other race-oriented questions in the 2016 ANES. As you can see in Figure 5 (below), the racial favoritism score significantly correlated with every race-oriented question in the survey.


Should we accept that these attitudes, as measured in the 2016 ANES, are real or just merely artifacts of our research methods?

I am convinced these results reflect something real and important in the American body politic. Many in the news media believe negative racial attitudes are largely isolated within the conservative right. This is simply not true. And suggesting so harms the news media’s credibility.

This biased media perspective may be a reflection of the insularity of journalists and political analysts working in the urban centers of the East and West Coasts of the U.S.

In the real world, negative racial attitudes are all too prevalent, regardless of political orientation. But I understand those who will dismiss any suggestion that negative racial attitudes cut across party labels. For many, it violates their own prejudices with regards to white working-class Americans. In addition, I recognize that there are deep methodological flaws in this analysis and some of the analytic conveniences I adopted I would probably reconsider. Survey methods are blunt and too crude to understand complex, socially-constructed concepts like race.

Nonetheless, when the news media suggests ‘racial resentments’ determined the 2016 election, it is important to keep these blanket declarations in perspective. There is a fundamental bias in social science research: the bias of statistically significant marginal effects.

Factors can be statistically significant but not necessarily the dominant forces in determining social phenomena (i.e., effect size). In the case of 2016 presidential vote preferences, past voting behavior, party identification and ideological predispositions were still powerful determinants of how people voted. In the margins, however, which can determine election outcomes, ‘racial resentments’ may well have made the difference in 2016.

There is nothing in the data I’ve analyzed that contradicts that finding.

But those marginal determinants do not tell the whole story. In fact, they can distort the truth if not put into their proper context.

I’m comfortable with criticisms of Trump supporters whose racial resentments animated their 2016 vote decisions. Trump supporters are more likely to exhibit racial resentments. I get it.

But I also expect Democrats to stop acting like virtuous overlords on the issue of racism. While negative racial attitudes are deeply-embedded in the Republican Party, if this analysis is accurate, about a third of Democrats have their own racial biases.

Recent news stories are suggesting the Republicans will continue to emphasize the same issues, such as immigration, that generated Trump’s large advantage among less-than-college educated white voters in 2016. Based on the racial favoritism bias I see in the 2016 data, that strategy may work again as there are a substantial percentage of Democrats potentially open to Republican campaign messages meant to activate racial resentments in voters.

Racism, in any form, is not acceptable as part of any party’s political agenda. At the same time, carelessly associating policy positions or party preferences with racial bias is also detrimental to our national political discourse.

Most Republicans are not racists. And supporting the border wall, or opposing amnesty for undocumented Americans, is not necessarily evidence of racism either. Therefore, it would be a pleasant surprise if the mainstream media stopped pushing that deceptive narrative about Trump voters and, instead, focus on why racial bias continues to be so pervasive within American society.

I’m not holding my breath.

I am a survey and statistical consultant with over 30 -years experience measuring and analyzing public opinion (You can contact me at:

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store