Americans don’t hide their racism

By Kent R. Kroeger (March 12, 2019)


This is the second essay in a series dedicated to analyzing the U.S. eligible-voter population using the 2018 American National Election Study (ANES), an online survey administered in December 2018 by researchers from the University of Michigan and Stanford University. __________________________________________________________________

How best to measure racism has long been debated in the political science, psychology and sociology research communities where the measures that have been developed inevitably meet with significant criticism.

The direct, self-reported method, in which researchers ask respondents directly whether they prefer their own race over others, is the most commonly used measure of individual racial bias. Using direct measures, research shows that individual racist beliefs have been in decline since the 1970s, according to longitudinal data from the General Social Survey.

Disputing this conclusion, some social scientists argue that social desirability bias —where respondents give answers based on what they believe to be ‘socially acceptable’ answers as opposed to how they actually feel — invalidate direct racism measures. However, recent research by University of Virginia psychologist Jordan R. Axt compared direct and indirect measures of racism and found that “the best method to measure individuals’ explicit racial attitudes is to ask about them directly.” People are generally honest when they answer survey questions on racial attitudes.

The direct versus indirect racism measures controversy will not be resolved here, however, and I do not out-of-hand dismiss the criticism that direct measures under count actual levels of individual racism.

With that caveat, this article— the second where I analyze 2018 American National Election Study (ANES) data — focuses on a series of direct racial attitude questions asking about respondents’ on a 0-to-100 scale their favorability ratings towards other races and ethnicities versus their own (White, Black, and Hispanic).

In addition, I focus exclusively on white respondents in the 2018 ANES. People of other races and ethnicities can, of course, be racist too. But for the sake of clarity, I target the segment in society most populous and historically most privileged.

How I define a ‘racist’ in this study is also straightforward. If the white respondent rated their race higher than another race or ethnicity (Black, Hispanic) by more than 10 scale points, they are coded as ‘racist.’ Some will argue that this definition is too harsh; while others might argue even a 1-point difference indicates something racist. That is a debate for another day.

Using my definition, there is one striking conclusion in the 2018 ANES data: Ethnocentric and racist attitudes are common among Americans — and no political party or ideology is immune from its presence.

Among vote eligible white Americans, 34 percent favor their race over Blacks or Hispanics, or both (see the third table in Figure 1). Using the political faction segmentation from my previous essay, the highest presence of racial favoritism is among Democrat Centrists (50%), GOP Centrists (49%), and GOP Conservatives (46%). The lowest presence is among Democrat Progressives (10%) and Independents (31%).

Figure 1: Prevalence of racial favoritism among vote eligible white Americans

Source: 2018 American National Election Study (Pilot); Analysis by Kent R. Kroeger (results use weighted data to account for sampling design and response rate differentials)

This racial bias can also be observed in Figures 2 and 3 which plot each respondent based on their favorability rating for white, Blacks and Hispanics. Figure 2 shows white and Black favorability scores and individual cases of racial bias are represented by dots below the diagonal line (i.e., white favorability is higher than black favorability). While most cases plot near the line, we see the obvious correlation between party factions and racial favorability. GOP Conservatives and Centrists are below the line and a high percentage are well below (10 points+) the line. Conversely, Democrat Progressives are mostly well above the line; while a sizable percentage Democrat Centrists appear on both sides.

Figure 2: Racial favorability ratings towards Blacks among vote eligible white Americans

Source: 2018 American National Election Study (Pilot)

A very similar pattern emerges when comparing favorability ratings for whites and Hispanics (see Figure 3). Republicans are generally below the diagonal line, while Democrats are above the line.

Figure 3: Racial favorability ratings towards Hispanics among vote eligible white Americans

Source: 2018 American National Election Study (Pilot)

Before Democrat Progressives start crowing about these findings, keep in mind that racial bias is only one noxious bias present in our society. There are many other types of biases: economic class, gender, sexual orientation, religious, geographic, age, education, political ideology and others. Rest assured, Democrat Progressives possess their fair share of socially corrosive biases. But that, as well, is a debate for another day.

Having established that racism, as defined here, is not uncommon among whites and that there is a political relationship to this bias, how might it relate to our current president, Donald Trump? Probably the least suspenseful question ever asked. But here we go…

First, let us establish that feelings about President Trump are strongly related to political attitudes (and party factions). Figure 4 illuminates that relationship. Plotting political attitudes for all respondents in the 2018 ANES — based on a series of policy-related questions — against how respondents rate Donald Trump, we see a strong relationship.

Figure 4: Political attitudes and favorability towards Donald Trump (All 2018 ANES respondents)

The bluish dots (Democrats) tend to rate Donald Trump low and are to the right along the political attitudes index (i.e., possess liberal political attitudes).

In contrast, the reddish dots tend towards the upper left-hand quadrant of the chart (i.e., rate Trump highly and have conservative political attitudes).

Interestingly, GOP and Democrat Centrists and Independents are sprinkled throughout the chart in Figure 4. There are a lot of them and their views on Trump don’t relate to partisan policy attitudes. I would call that a tactical opportunity for both parties heading into the 2020 election. But that is a topic for another essay.

Based on this essay’s definition of a racist attitude, there is a clear (but not deterministic) relationship between racist attitudes and feelings regarding Trump and political attitudes (see Figure 5 below). Most, but not all, of the black and yellow dots (racists) are in the upper left-hand quadrant of the chart.

However, it is not an air tight relationship. The Kendall tau-b statistic of association for racist attitudes and feelings towards Trump is 0.21 (significant at the 0.01 level, two-tailed test), and with political attitudes it is only -0.19 (also statistically significant).

By comparison, feelings regarding Trump and party faction membership (DEM Progressive, DEM Centrist, Independent, GOP Centrist, GOP Conservative) generate a Kendall tau-b statistic of 0.70, indicating a much stronger relationship. Likewise, political attitudes and feelings towards Trump achieve a Kendall tau-b statistic of -0.65.

Figure 5: Political attitudes and favorability towards Donald Trump by racist attitudes

In a multiple regression model with controls for party identification, age, gender, and political attitudes, the racism index is a significant, independent predictor of feelings towards Trump. However, relative to political attitudes and party identification, the racism index is a minor contributor in explaining feelings towards Trump (see Appendix for the linear regression model summary output).

It is oddly refreshing that Americans appear willing to express their racial biases on a national opinion survey (even if it still may be an under representation of actual racial bias levels in the total population).

If our goal is to reduce all types of racism in this country — assume we can never eliminate it — a good start is creating an atmosphere where everyone feels comfortable talking about their own racial attitudes without fear of shaming and retribution.

German Lopez’ excellent article on is a good place to start the process of generating ideas and strategies for reducing racism. As Lopez points out, there are ways to reduce racial bias, but “calling people racist isn’t one of them.” And that is pretty much the level of dialogue we are treated to on a daily basis on cable news channels and social media. Whatever the motives and intentions of these media lords, they aren’t working to reduce racial bias.

In his 1993 book, Race Matters, Harvard professor Dr. Cornel West argues we must first “understand that racism and race are woven in American history and can never be eradicated without understanding that race matters in everything we consider American.”

Knowing this, it is encouraging that Americans from all perspectives seem willing to share their attitudes and beliefs on race with survey researchers. Its a good start in the long process of creating a healthy environment conducive to racial understanding and acceptance.

  • K.R.K.

Linear regression model for explaining ‘Feelings towards Trump’.

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