Posted on May 6, 2021

The Social Construction of Racism in the United States

Eric Kaufmann, Manhattan Institute, April 7, 2021


Executive Summary

This paper begins with a version of Tocqueville’s paradox:[1] at a time when measures of racist attitudes and behavior have never been more positive, pessimism about racism and race relations has increased in America.

Why? An analysis of a wide variety of data sources, including several new surveys that I conducted, suggests that the paradox is best explained by changes in perceptions of racism rather than an increase in the frequency of racist incidents. That is, ideology, partisanship, social media, and education have inclined Americans to “see” more bigotry and more racial prejudice than they previously did. This is true not only regarding the level of racism in society but even of their personal experiences. My survey findings suggest that an important part of the reported experience of racism is ideologically malleable. Reports of increased levels of racism during the Trump era, for example, likely reflect perception rather than reality—just as people have almost always reported rising violent crime when it has been declining during most of the past 25 years. In addition, people who say that they are sad or anxious at least half the time, whether white or black, are about twice as likely as others to say that they have experienced racism and discrimination.

None of this means that racism is an imaginary problem. However, efforts to reduce it should be based on strong empirical evidence and bias-free measures. The risks of overlooking racism are clear: injustice is permitted to persist and grow. Yet there are also clear dangers in overstating its presence. These go well beyond majority resentment and polarization. A media-generated narrative about systemic racism distorts people’s perceptions of reality and may even damage African-Americans’ sense of control over their lives.

Key Findings Include:

  • Eight in 10 African-American survey respondents believe that young black men are more likely to be shot to death by the police than to die in a traffic accident; one in 10 disagrees. Among a highly educated sample of liberal whites, more than six in 10 agreed. In reality, considerably more young African-American men die in car accidents than are shot to death by police.

—Ideology, not education, influences the extent to which people are incorrect on police shootings and traffic accidents.

—Black Trump voters are almost 30 points more likely to get the question right than black Biden voters.

—Conservative whites are almost 50 points more likely to get it right than liberal whites.

—African-Americans who strongly agree that white Republicans are racist are 40 points more likely to get the question wrong than those who strongly disagree that white Republicans are racist.

    • Black Biden voters are twice as likely as black Trump voters to say that they personally experienced more racism under Trump than under Obama. Black Trump voters reported a consistent level of racism under both administrations. Black respondents who strongly agree that white Republicans are racist are 20–30 points more likely to say that they experience various personal forms of racism than African-Americans who strongly disagree that white Republicans are racist.
    • Reading a passage from critical race theory author Ta-Nehisi Coates results in a significant 15-point drop in black respondents’ belief that they have control over their lives.
    • A slight majority of African-Americans and whites overall felt that political correctness on race is demeaning to black people rather than necessary to protect them. Among blacks, the difference between liberals and conservatives was 3 points (51% of the liberals thought it was demeaning vs. 54% of the conservatives). Among whites, however, there was a nearly 20-point divide between liberals and conservatives (43% of the liberals thought it was demeaning vs. 62% of the conservatives).
    • Liberal African-Americans with a college degree are nearly 30 points more likely to find a statement by a white person such as “I don’t notice people’s race” or “America is a colorblind society” offensive than African-Americans without degrees who identify as conservative. Among whites, the gap between liberals and conservatives is 50 points.
    • When asked to choose between a future in which racially offensive remarks were so heavily punished as to be nonexistent and one where minorities were so confident that they no longer felt concerned about racial insults, black respondents overall preferred, by a 53%–47% margin, the resilience option. White liberals preferred the punitive option, by a 71%–29% margin; black liberals chose the second option by just 6 points, 53%–47%.
    • In general, African-Americans’ opinion on race issues appears to be less affected by ideology and partisanship than white opinion. In a 2015 Pew survey, 20 points separated “very conservative” and “very liberal” African- Americans on whether racism is a very big problem. The gap between “very conservative” and “very liberal” whites was 65 points; the gap between “very conservative” and “very liberal” Hispanics and Asians was 40 points.
  • Exposure to social media and other media appears to be related to survey respondents’ views of both the national prevalence of racism and their personal experience of it.


The Media and Public Perception of Racism

It is well known that the media, with their ability to frame events and social trends, have an impact on public opinion. This is especially the case when it comes to the visibility and political prominence of certain issues, what political scientists call issue “salience.” For example, there is a close relationship in Europe between media coverage of immigration and salience—the number of people saying that immigration is the most important issue facing their country.

The same appears to be true for race. Gallup data show that the civil rights era of the 1950s and early 1960s, as well as the race riots in the late 1960s, saw the public salience of race spike (Figure 1). The public salience of race then remained muted until 1992, when the Los Angeles riots, in the wake of Rodney King’s beating, sent questions of race to the top of 15% of the public’s priority lists. Since 2014, a series of events (including the protests in Ferguson, Missouri, after the shooting death of Michael Brown by a police officer; and the election of Donald Trump) pushed the race issue above 10% salience. In 2020, despite the Covid-19 pandemic, the George Floyd/Black Lives Matter protests elevated race back to the top spot: it was named as the leading concern by nearly 20% of the public in mid-June 2020. This was the highest salience level recorded for race since the late 1960s, eclipsing the Rodney King spike.

Media events affect the prominence of issues of race and racism in the public consciousness, but they also shape how people evaluate the quality of race relations. Other Gallup data show that during 2001–14, nearly 70% of Americans said that relations between whites and blacks were good. After the Ferguson protests, this fell to 47%, hovered in the low 50s between 2015 and 2019, and has since tumbled to 44% following the BLM protests (Figure 2).

The Decline of Racist Attitudes

The increasing pessimism over race relations stands in contrast to the steady, long-term liberalization among white Americans across a range of racial attitudes measured in the leading General Social Survey (GSS) since 1972. In the 1970s, for example, nearly 60% of white Americans agreed with the statement that blacks shouldn’t “push themselves where they’re not wanted.” This response had declined to 20% by 2002, when the question was discontinued. The share of white Americans who agree that it is permissible to racially discriminate when selling a home declined from 60% as late as 1980 to 28% by 2012.[9]

Approval of black–white intermarriage rose among whites from around 4% in 1958 to 45% in 1995 and 84% in 2013, according to Gallup.[10] In 2017, fewer than 10% of whites in a major Pew survey said that interracial marriage was a “bad thing,”[11] and, as Figure 3 shows, few now oppose a relative marrying someone of a different race. The actual share of intermarried newlyweds rose from 3% in 1967 to 17% in 2015.[12]

For decades, American National Election Studies (ANES) posed a question of whether minorities/ blacks should help themselves or whether the government should help them more. There was a gradual rise in support for government assistance to blacks during 1970–2016 of about a half-point on a seven-point scale.[13] Meanwhile, police killings of African-Americans declined by 60%–80% from the late 1960s to the early 2000s and have remained at this level ever since.[14] Racist attitudes and behaviors have sharply declined, though the problem has not been eradicated.

The Racism Paradox

The increasingly sour national mood on race relations in the U.S. may likely be related to the higher salience of race since the 2014 Ferguson protests. While it is too early to be definitive, the emergence and rapid spread of social media may account for this. Combined with smartphone citizen journalism, social media mean that knowledge of white-police-on-black-suspect violence is more likely to circulate widely, where it can ignite riots and boost the salience of the race question. Thus, even as the number of such incidents is declining, each event is more likely to be captured alive and to possess a higher media multiplier effect.

Evidence that social media may be shaping perceptions of racism is provided in Figure 4, which shows that black respondents on social media in 2016 were considerably more likely to report experiencing discrimination than those not on social media. This is a statistically significant effect that holds when controlling for age, education, income, partisanship, ideology, gender, and contact with whites. On questions about whether black people have experienced people acting suspicious of them or thinking that they are not smart, the gap between those on social media and those not on it reaches as high as 20 points.

People who care passionately about an issue (or see it flagged in the media) tend to overestimate its prevalence. For example, Americans and Europeans routinely overestimate the population shares of minorities, immigrants, and Muslims while underestimating the white share. In France, the average person in 2016 thought that the country was 31% Muslim; the correct answer was 7.5%. In the U.S. and Canada, the same survey shows that people estimated their countries to be 17% Muslim, compared with the actual 1% and 3%, respectively. Anti-immigration whites overestimate more than liberal whites.[15] Meanwhile, minorities tend to overestimate their share of the population more than whites do because they extrapolate from their locale to the nation. Black respondents in a 2005 survey said that the U.S. was 38% black rather than the actual 12%, and Hispanics said that the country was 39% Hispanic rather than 13%.[16]

In terms of racial discrimination, a 2019 study asked people how many résumés a black person would have to send out to get a callback from an employer if a white applicant gets one callback for every 10 applications. It found that Democrats thought that a black person would have to send 26 résumés to get one callback, while Republicans said 17. The correct answer was 15. Overall, blacks were not significantly more likely than whites to overestimate discrimination: partisanship, rather than race, is what apparently led to misperceptions.[17]

Perceptions about trends over time are somewhat more accurate. In Western Europe, for example, concern about immigration is connected to actual immigration levels over time and tends to rise when inflows are high.[18] But this is not always the case. In the U.S., crime rates were flat between 1989 and the mid-1990s, and then fell every year until 2019. Unmoved, a majority of Americans in every year but two since 1989 said that crime had risen over the past year. In 2019, 64% of Americans said that crime had risen in the previous year, even though it had actually continued its gradual post-1990s decline.[19] Emotive issues that feature in the news affect people’s perceptions of the size of a problem.

The Great Awokening

Videos of interracial violence circulate on social media; yet this has not led to noticeably cooler feelings between America’s racial groups in ANES surveys.[20] Something distinctive has changed with respect to perceptions of racism since 2014 that cannot be explained solely by technological change.

Ideological shifts are an important independent factor to investigate when trying to explain the racism paradox. At the U.S. state level, Google searches for “racism” are highly correlated with searches for “sexism,” which is, in turn, correlated with the Democratic share of the vote. Liberal states such as Vermont tend to come out highest on both indices, while southern and mountain states score lowest. Searches for “racism,” in short, serve as a useful barometer of left-modernism.

While the salience of racism fluctuates with events, the use of the term “racist” and “racism” has increased in three waves since 1960. Figure 5, based on big data from Google’s Ngram Viewer, tracks the popularity of terms in English-language books. It shows that the use of the term “racism” first rose sharply in the late 1960s, a time of New Left student activism. After reaching a plateau, it surged again and rose to a new level in the late 1980s and early 1990s, when political correctness and speech codes came into vogue. Then, around 2014, there was another upsurge, in tandem with the current period of left-modernist ferment. The use of terms such as racism (or racist/s) took off especially sharply in left-leaning media outlets such as Vox and the New York Times.[21]

A 2018 report, “Hidden Tribes: A Study of America’s Polarized Landscape,” published by More in Common, found that “Progressive Activists,” who make up 8% of the U.S. population, are 3.5 times more active than the “exhausted majority” two-thirds of the population in posting political content on social media.[22] While the rise of social media, citizen journalism, and a surge in online partisan websites has been associated with what Matthew Yglesias calls the “Great Awokening,” it is not associated with right-wing populist voting, which is stronger among older and less educated voters who use social media platforms less.[23] Many left-modernist ideas have older roots in critical theory, but technological change helped left-modernists organize and spread new moral innovations, such as “microaggressions,” or causes, such as gender recognition.[24]

To what extent the most recent “Awokening” would have occurred in the absence of social media is an open question. Whatever the case, the Great Awokening has coincided with a large-scale shift to the left in attitudes toward questions of race, diversity, and immigration, especially among white liberals. Thus, partisanship and ideology increasingly affect perceptions of racism. The importance of ideological differences in perceptions of racism is shown in Figure 6, which reveals that among white conservatives, there has been little to no increase since 1995 in the share who think that racism is a “big problem.” White liberals show the greatest increase, with white moderates in between. The post-2014 trend (see Figure 2) of perceiving worse relations between whites and blacks is, therefore, less a reflection of statistical reality than of rising consciousness of racism, notably among liberals.


Black Public Opinion

Much of the evidence about perceptions of racism so far comes from national samples, which are dominated by white respondents. Though sample sizes for African-Americans in such surveys are typically small and there are fewer black-only surveys, it is apparent that black opinion is characterized by a weaker ideological divide than exists within white opinion. Data from Pew, for example, show that among blacks, 75% of liberals, but also 55% of the smaller group of conservative blacks, say that discrimination makes it harder for blacks to get ahead (Figure 7). By contrast, 17% of “very conservative” whites and 82% of “very liberal” whites agree. A modest 20-point partisan difference among blacks balloons to 65 points among whites. Since 2016, several surveys show that white liberals place to the left of minorities on questions of race, diversity, and immigration.[28]

There is also a substantially larger ideological gap among whites than blacks when it comes to viewing racism as a serious problem. Pew’s 2015 survey, profiled in Figure 8, found that “very liberal” whites evince nearly as much concern over racism (76%) as African-Americans, while moderate (33%) and conservative (11%) whites view racism as a much less important problem. The ideological slope is greatest for whites, with over 60 points separating conservatives from “very liberal” whites. The incline is less steep among Asians and Hispanics and gradual among blacks, with “very liberal” and “very conservative” blacks only differing 20 points in their assessment that racism is a very big problem (64% vs. 84%).