Report Nationscape™

Racing Apart

Partisan Shifts on Racial Attitudes Over the Last Decade

October 2021

Political Parties, Race & Ethnicity

Share this Article

Racing Apart

Key Findings

  • Democrats’ and independents’ attitudes on identity-related topics diverged significantly from Republicans’ between 2011 and 2020 — including their attitudes on racial inequality, police, the Black Lives Matter movement, immigration, and Muslims. Most of this divergence derives from shifts among Democrats, who have grown much more liberal over this period.
  • Donald Trump’s actions and rhetoric helped create this divergence. White Americans’ views of Trump as of 2016 were strongly correlated with shifts in their views on these issues between 2016 and 2020. This means that white Democrats placed themselves further in opposition to a candidate and president who made explicit racial appeals.
  • The murder of George Floyd and the protests that followed had a dramatic impact on Americans’ attitudes on policing and the Black Lives Matter movement. In the aggregate, this shift was mostly temporary. Its long term consequence was even greater partisan polarization on racial issues.
  • Despite increases in support for affirmative action and reparations for slavery, support for these policies still lags behind overall concern about racial inequality. Support for reparations also did not change much in response to Floyd’s murder and the ensuing protests.
  • There are early signs that Republicans are reacting against President Biden’s immigration policies just as Democrats reacted against President Trump’s. As a result, Republicans have increasingly less favorable views of immigration.


At the beginning of his 1903 book, The Souls of Black Folk, W.E.B. DuBois wrote “the problem of the Twentieth Century is the problem of the color line.” It would prove a prescient statement not only about that century, but about the 21st century as well. This was painfully clear in the past year, when the murder of George Floyd foregrounded, once again, how the color line is manifest in policing. It was also clear in the politics of the presidential election, in which race and racialized issues — including policing, reparations for slavery, voting rights, the COVID-19 pandemic, immigration, and others — were central.

The 2020 election took place amid a remarkable shift in attitudes about issues related to racial, ethnic, religious, and national identities. After decades of relative stasis in racial attitudes, Democrats and independents became dramatically more liberal on these issues. This leftward shift was already underway as of the 2016 election but it continued into 2020 as well, which we document using unique survey data that has tracked more than 3,300 Americans over almost 10 years. Among Republicans, there has been no clear shift.

The resulting polarization between Democrats and Republicans is not simply a generically partisan phenomenon. It is a phenomenon connected to Donald Trump himself. Partisans often develop their views in response to information from political leaders. In the case of Democrats’ racial attitudes, it meant reacting against the opposing party’s leader. Trump’s brand of identity politics — which included explicit racial appeals and the demonization of various racial, ethnic, and religious groups — helped push Democrats in the opposite direction.

Against this backdrop of partisan polarization on racial issues came the dramatic events of 2020, especially the murder of Floyd and ensuing protests. Immediately after Floyd’s murder, views of the police became less favorable while views of the Black Lives Matter movement became more positive. More Americans came to perceive that Black Americans face serious discrimination and espoused views that were consistent with a belief in systemic racism.

As a whole, this shift in attitudes was largely temporary. Weekly surveys from the Democracy Fund + UCLA Nationscape project show that any aggregate changes had mostly evaporated by January 2021. This shift back to the status quo appeared to reflect the decline in news coverage of Floyd and in the number of protests, as well as the increasingly polarized political debate about the protestors’ tactics and policy demands. By early 2021, Democrats and Republicans had grown further apart on racial issues. Thus, the dramatic events of 2020 did not fundamentally disrupt the attitudes and partisan alignments that had been developing over the past 10 years. If anything, it only strengthened those alignments.

Already, trends in public opinion during the presidency of Joe Biden suggest that further polarization is underway. Just as Democrats shifted away from Trump, Republicans are now shifting away from Biden, especially on the issue of immigration. The upshot is an even greater divide: Now, perhaps more than ever, party politics in the U.S. is organized around fundamental questions about racial justice and American identity.

Emerging Racial Liberalism in the Democratic Party

In the 1990s and 2000s, attitudes about systemic racism and Black Americans did not shift much in the aggregate — a pattern that the political scientists Christopher DeSante and Candis Watts Smith have called “racial stasis.” This was particularly evident in responses to a series of questions that compose “racial resentment,” a measure that captures how much people attribute racial inequality to structural forces, like slavery and discrimination, as opposed to individual-level factors, like whether Black Americans “try hard enough.” Racial resentment is typically measured by asking people whether they agree or disagree with four statements:

  • "Over the past few years, Black people have gotten less than they deserve."
  • "Irish, Italian, Jewish, and many other minorities overcame prejudice and worked their way up. Black people should do the same without any special favors."
  • It's really a matter of some people not trying hard enough; if Black people would only try harder they could be just as well off as white people."
  • "Generations of slavery and discrimination have created conditions that make it difficult for Black people to work their way out of the lower class."1

But since 2011, there have been remarkable changes in these attitudes. This is visible in the Democracy Fund Voter Study Group VOTER Survey (Views of the Electorate Research Survey). Between 2011 and 2020, Americans have become more likely to attribute racial inequality to structural forces as opposed to individual effort (Figure 1). For example, the proportion of Americans who agreed that Black people have gotten less than they deserve rose from 20 percent to 45 percent. The number who agreed that discrimination and slavery prevent Black people from making economic progress increased from 35 percent to 48 percent. In addition, more Americans disagreed that Black people should overcome prejudice without special favors (from 20 percent to 37 percent) and that Black people could be just as well off as white people if they tried harder (from 31 percent to 45 percent).

1 On the development of the measure of racial resentment, see Donald R. Kinder and Lynn M. Sanders, “Divided by Color,” Chicago: University of Chicago Press, 1996. On the prior stability of racial resentment and other racial attitude measures, see Michael Tesler, “Post-Racial or Most-Racial? Race and Politics in the Obama Era,” Chicago: University of Chicago Press, 2016; and Christopher D. DeSante and Candis Watts Smith, “Racial Stasis,” Chicago: University of Chicago Press, 2020. On the meaning of the racial resentment scale, see Cindy D. Kam and Camille D. Burge, “Uncovering Reactions to the Racial Resentment Scale across the Racial Divide,” The Journal of Politics, 2019, 80(1): pp. 314–320.

In large part, these shifts were driven by Democrats and independents. This change was already evident in 2016, but the trends continued between 2016 and 2020.2 Compared to 2011, Democrats in 2020 were far more likely to agree that Black people have gotten less than they deserve (from 32 percent to 73 percent) and to agree that discrimination makes it difficult for Black people to make economic progress (from 51 percent to 77 percent). Democrats were also more likely to disagree that Black people should overcome prejudice without favors (from 35 percent to 64 percent) and to disagree that Black people could be as well off as white people if they tried harder (from 50 percent to 71 percent). Independents showed smaller but directionally similar changes. Between 2011 and 2020, the number of Republicans who agreed or disagreed with these statements did not change substantially. Prior research has shown that partisan polarization on racial resentment began before 2011, but it has clearly accelerated since.(i)

These shifts were evident among white, Black, and Latino Democrats, although most of the change came from white Democrats (Figure 2), which largely closed a historic gap between white and Black Democrats on these questions.

2 On the earlier trends, see John Sides, Michael Tesler, and Lynn Vavreck, “Identity Crisis: The 2016 Presidential Campaign and the Battle for the Meaning of America,” Princeton University Press, Princeton, NJ, 2018. On the trend in the 2011–2016 VOTER Survey, see Sean McElwee, “The Rising Racial Liberalism of Democratic Voters,” The New York Times, May 23, 2018. Available at:

The increase in liberal racial attitudes among Black Democrats was more modest but is still important. Between 1986 and 2016, the trend among Black Americans was toward less liberal responses to these questions.(ii) This conservative shift extended to weaker Black support for government aid to Black Americans, and it also has been attributed to the moderating effect of incorporating Black political leaders into mainstream American politics.(iii) But between 2016 and 2020, Black Democrats shifted in the opposite direction. (There were shifts in the same direction among Black independents and Republicans as well, although these shifts contributed less to the trend overall given the smaller number of Black independents and Republicans.3)

Latino Democrats also shifted in the same liberal direction, although overall their responses to these questions were less liberal than those of Black or white Democrats. Nevertheless, the differences between the racial attitudes of Latino and Black Democrats were generally smaller in 2020 than 2011.

These shifts among Democrats were not simply the result of partisan sorting — with more racially liberal Americans switching into the Democratic Party and more racially conservative Americans switching out. They were also driven by actual changes in attitudes. To see this, we can focus on VOTER Survey respondents who were interviewed in both 2011 and 2020 and identified as Democrats in both of those years. Among these consistent Democrats, there were similar trends in racial attitudes. For example, the percentage who agreed that Black people have gotten less than they deserve increased from 38 percent to 73 percent, and the percentage who disagreed that Black people could be as well off as white people if they tried harder increased from 59 percent to 72 percent.4

Growing partisan differences are also visible for other identity-inflected issues. One is immigration. The VOTER Survey has tracked opinions on three different immigration questions since 2011: support for a path to citizenship for undocumented immigrants, whether undocumented immigrants make a contribution to American society, and whether to make it harder or easier to immigrate legally to the U.S.

In each case, Democratic attitudes have become much more favorable toward immigrants and immigration (Figure 3). Between 2011 and 2020, the percentage of Democrats who supported a path to citizenship increased from 57 percent to 81 percent. The percentage who said undocumented immigrants make a contribution increased from 39 percent to 74 percent. The percentage who wanted to make legal immigration easier increased from 31 percent to 51 percent. Among independents, there were much smaller shifts in the same direction. Among Republicans, there was a modest increase in support for a path to citizenship but no change in views about the contributions of undocumented immigrants and a decrease in support for easier immigration to the U.S.

3 This is based on analysis of the Cooperative Congressional Election Study, whose large sample size (roughly 50,000 cases in election year surveys) allows us to examine subgroups of Black Americans.

4 For more evidence, see Andrew M. Engelhardt, “Racial Attitudes Through a Partisan Lens,” British Journal of Political Science, January 2020.

Other surveys show public support for increasing immigration growing in the same manner, with Democrats shifting more than Republicans. In Gallup surveys, the percentage of Democrats who wanted to increase immigration rose from 30 percent to 50 percent between June 2016 and June 2020. Among Republicans, there was no change. When respondents were asked if immigration was a “good thing” or “bad thing” for the country, the percentage of Democrats who said “good thing” increased from 71 percent in 2014 to 85 percent in 2018. Over that period, the percentage of Republicans who said this rose from 58 percent to 65 percent.(iv)

A final example of this alignment between party identification and identity-inflected issues has to do with views of Muslims. Between 2011 and 2020, the VOTER Survey has asked respondents to rate Muslims on a “feeling thermometer,” a 0–100 scale where 100 indicates very warm or favorable feelings and 0 indicates cold or unfavorable feelings. On this measure, Muslims are one of the least liked social groups in the U.S., compared to a range of racial, ethnic, and religious groups.(v)

But over the past decade there has been an increase in positive views of Muslims, again primarily driven by Democrats (Figure 4). Among Democrats, the average feeling thermometer rating of Muslims increased from 60 to 67 between 2011 and 2020. Among independents and Republicans, there was an increase between 2019 and 2020, but this largely brought both groups to where they were in 2011.5

5 Weekly Nationscape surveys conducted in 2019–2020 also showed increasingly positive views of Muslims among all partisan groups, but especially Republicans. This is consistent with the pattern in the VOTER Survey. See Mayesha Quasem and Robert Griffin, “The Changing Sway of Voter Attitudes Toward Muslims,” Democracy Fund Voter Study Group, April 6, 2021. Available at:

Again, other surveys have corroborated these shifts. For example, in Pew surveys conducted between 2009 and 2017, there was a 16-point increase in the percentage of Democrats who agreed that “The Islamic religion does not encourage violence more than others.” There was no increase in positive views of Islam among Republicans.(vi)

Taken together, these trends in attitudes about Black Americans, immigrants, and Muslims demonstrate that the partisan polarization on racial issues that was so evident in the 2016 election year has only intensified.

The Role of Trump

Why did Democrats’ views on race and ethnicity change so much in such a short time? Clearly there could be several factors at work, especially the evidence of police brutality combined with protest movements focused on racial injustice.

But another factor stands out, particularly in the last four years: Donald Trump himself. His rhetoric helped push Democrats toward more liberal positions on race.

Trump’s hostile statements about racial, ethnic, and religious minorities marked an important departure from dog-whistle politics, of which there is a long history in America. Trump was much more comfortable making explicit racial appeals rather than relying only on the code words or innuendo used by some politicians before him. The shift from implicit to explicit racial appeals under Trump helped simplify these issues for voters, sending increasingly clear signals about where the two parties stood.

Trump’s message mattered because of its effect on public opinion. Americans often develop their opinions via signals from those they perceive as political allies and enemies. On many issues, public attitudes have polarized after Democratic and Republican politicians took different positions. Moreover, political science research suggests that cues from enemies can be more important than cues from allies. Politicians often polarize public opinion by inducing members of the other party to take opposing positions.(vii) For example, even before he became president, Trump’s rhetoric and policies had a backfire effect. During his presidential campaign, his constant push for a U.S.-Mexico border wall appeared to make such measures less popular among Democrats.(viii)

Although any president would likely have faced some pushback, Trump was not just any president. He inspired unprecedented animosity and opposition among Democrats throughout his term.6 His positions on race, immigration, and Islam created an incongruity for Democrats who disliked Trump but were otherwise more moderate or conservative on these issues. The easiest way for these Democrats to resolve this incongruity was to shift their positions away from Trump’s.

This is precisely what happened among white Americans. Even more so than partisan identification, changes in their racial attitudes from 2016 to 2020 were correlated with their feelings about Trump.

Using data from the December 2016 and September 2020 waves of the VOTER Survey as well as the 2016 and 2020 waves of the American National Election Study panel, we calculated how much respondents who participated in both waves changed their attitudes on three dimensions: their racial resentment score, their feelings toward Black Lives Matter, and their feelings toward police. We modeled these changes as a function of several characteristics measured in 2016: respondents’ favorability toward Trump, party identification, ideological identification, age, gender, and education level.

In both surveys, white Americans’ feelings about Trump in 2016 were strongly associated with subsequent changes in their responses to the questions from the racial resentment scale as well as their feelings about the Black Lives Matter movement and police (Figure 5).7 In particular, the less favorably white Americans felt toward Trump in 2016, the more their attitudes shifted in a liberal direction between 2016 and 2020. Feelings toward Trump were also much more strongly associated with changes in attitudes than was party identification — meaning that what appeared to matter more was not being a Democrat or Republican but being a Trump opponent or Trump fan.8

6 See, for example, data from Pew and Gallup surveys.

7 See Appendix for full model results.

8 The same patterns are visible among Black Americans, but less consistently. In the VOTER Survey, Trump favorability is correlated with changes in support for BLM but not changes in racial resentment (which is more strongly associated with party identification) or changes in feelings toward the police. In the ANES panel survey, there are no statistically significant relationships between Trump favorability and racial attitudes once party identification was included in the model. Unfortunately, due to sample size limitations in both surveys, the analysis of other racial and ethnic groups is not possible.

Of course, other factors besides feelings about Trump may also have helped create greater partisan polarization in racial attitudes. One factor was the highly publicized instances and protests of police brutality, which in turn shaped the messages that Trump and other political figures communicated to the public. For example, protests of systemic racism — from mass demonstrations to the actions of Colin Kaepernick and other professional athletes — frequently drew criticism from Trump. His presidential diatribes, in turn, brought even greater attention to the underlying issues and incidents being protested and helped rally the president’s opponents against his positions. This interplay could have helped push those who disliked Trump to adopt more liberal positions on race-related issues during his presidency.

The Impact of George Floyd's Murder and Ensuing Protests

On May 25, 2020, Minneapolis police officer Derek Chauvin knelt on George Floyd’s neck for nine minutes, ultimately killing him. Floyd’s murder led to an extraordinary number of protests across the country. The Crowd Counting Consortium identified about 4,700 separate demonstrations between May 25 and the end of June.(ix) The result was the most sustained attention to policing and racial justice that the country had seen since the beginning of the Black Lives Matter movement.

In the wake of Floyd’s murder, the initial reaction was bipartisan condemnation. Even leading Republicans who had never been sympathetic to Black Lives Matter appeared to side with Floyd and not the police. On June 6, Rush Limbaugh tweeted:

“There is not a single person in this country, regardless of race or political affiliation, who did not think that the nine minutes of video of a policeman’s knee on the neck of George Floyd was absolutely horrifying.”

At the Senate Republicans’ lunch on June 9, Senator Tom Cotton (R-AR) said:

“Young Black men have a very different experience with law enforcement in this nation than white people and that’s their impression and experience and we need to be sensitive to that and do all we can to change it.”9

On June 17, Senator Mitch McConnell (R-KY), the Majority Leader at that point, said:

“When Black Americans tell us they do not feel safe in their own communities, we need to listen. When citizens lack faith in our justice system, we need to respond. And when the equal protection of the law feels to some Americans like a contingency of demographics, we need to act.”10

When Democratic and Republican leaders largely agree on an issue, this creates what the political scientist John Zaller has called a “one-message model,” which tends to create consensus among Democrats and Republicans in the electorate. That is exactly what happened in May and June. For example, in a May 28–29, 2020 YouGov poll, there was bipartisan support for arresting Chauvin: 90 percent of Democrats and 68 percent of Republicans agreed. In a June 7–9, 2020 YouGov poll, 89 percent of Democrats and 72 percent of Republicans approved of “non-violent protests in response to George Floyd’s death.”

Thus, partisan differences were not completely absent, but they were far more muted than is often true on racial issues in general and on police killings of Black Americans in particular. Indeed, Floyd’s murder led some commentators to wonder whether, as one Politico article put it, “a punitive brand of conservatism embraced by Trump and some GOP hardliners is rapidly falling out of step with public opinion.”11,(x)

The initial shifts in public opinion were certainly rapid and dramatic. But not all of those shifts were durable. By the end of 2020, public opinion on a range of questions related to race and policing were fairly close to the status quo prior to Floyd’s murder. While Americans’ opinions in 2020 were still quite different than in 2016 — particularly because of liberal shifts among Democrats — most of this movement does not appear attributable to Floyd’s murder and the ensuing protests.

View on Police and the Black Lives Matter Movement

The public’s views of two key actors — the police and the Black Lives Matter movement — demonstrate both the longer-term pattern of partisan polarization and the shorter-term effects of Floyd’s murder and the ensuing protests. In VOTER Survey interviews with respondents after the 2016 and 2020 elections, views of the police and BLM were measured with the same 0–100 feeling thermometer described in the previous section.

Views of the police were net-positive in both years but clearly less favorable in 2020 than 2016 — an average of 65 vs. 74 (Figure 6). Favorability toward Black Lives Matter shifted in the opposite direction, with an increase in average thermometer ratings among all Americans from 41 to 50.

Both shifts were driven primarily by changes among Democrats and independents. Among Democrats, average thermometer ratings of police dropped from 66 to 53. Among independents, they dropped from 70 to 59. Republican views were stable (an average of 85 in 2016 and 84 in 2020). Similarly, Democrats’ average rating of BLM rose 12 points while Republican attitudes did not change.

Among Democrats, changes in attitudes toward the police were similar across different racial and ethnic groups. Feelings toward police among white (68 vs. 54), Black (59 vs. 46), and Latino (69 vs. 55) respondents dropped similar amounts over this four-year period. Changes in Democratic attitudes toward BLM occurred mainly among white and Latino Democrats. While white (59 vs. 74) and Latino (56 vs. 70) ratings of BLM rose about 15 points, changes among Black Democrats — already more favorable toward BLM than other Democrats — were more muted (81 vs. 84).

To examine the potential impact of Floyd’s murder and the ensuing protests on these attitudes, we rely on Nationscape surveys conducted weekly between July 2019 and February 2021. Existing research using Nationscape surveys has shown notable changes in attitudes after Floyd’s murder, extending into September 2020.(xi) Here we document how much of that change persisted into early 2021.

In these surveys, views of the police were measured with a four-category scale ranging from “very favorable” to “very unfavorable.” Prior to the Floyd murder, more people had a favorable view of the police than an unfavorable view — especially Republicans but also Democrats to a lesser extent (Figure 7). This was true among white, Latino, and Asian and Pacific Islander Americans as well. Even among Black Americans, views of the police were net positive in the first months of 2020.

That changed immediately after Floyd’s murder. The net favorability rating of police among all Americans dropped a dramatic 24 points, from +44 to +20. This was the largest one-week change in net favorability across any of the social, political, or religious groups measured by the survey between July 2019 and February 2021. This shift was also relatively consistent for Democrats (+33 to +3), independents (+28 to +19), and Republicans (+64 to +43).

However, trends between different partisan groups began to diverge by the beginning of June. While Democratic ratings of police continued to decline, bottoming out at -12 in mid-June, Republican attitudes bounced back almost immediately. By the beginning of August, Republican attitudes about police had largely recovered to their pre-Floyd levels. Democrats’ favorability toward the police increased more slowly over the remainder of the year.

Comparing estimates from the two weeks prior to Floyd’s murder and the two final waves of Nationscape — fielded between January 12 and February 3, 2021 — it is clear that most of the shifts that happened in late May and early June had disappeared within a matter of months. The net approval of police among all Americans was identical in these two time periods (+43 vs. +43) as well as slightly higher among Republicans (+63 vs. +68) and independents (+31 vs. +35). It was only among Democrats that attitudes were lower than those recorded in May (+32 vs. +27).

Roughly the same pattern occurred among different racial and ethnic groups: large drops in favorability toward the police followed by a relatively uniform increase over the next several months. These increases largely brought views of police back to the levels observed before the murder of Floyd in late May.

In Nationscape surveys, a similar question about the Black Lives Matter movement was added in July 2020. Thus, it does not capture changes immediately before and after Floyd’s murder. But we know from public polling that favorable views of BLM increased after Floyd was murdered. In Civiqs polling, for example, the percentage who supported BLM increased from 45 percent to 53 percent. This trend was largely driven by Democrats and independents.(xii)

But as the summer progressed, both Civiqs and Nationscape surveys show a decline in support for BLM (Figure 8). In Nationscape, the net favorability of BLM dropped from +20 to +9 between July 2020 and January 2021. This shift mostly reflects changes among Republicans (-27 to -42) and independents (+14 to -2) but also small shifts among Democrats (+62 to +58).12

There were decreases in favorability among most racial and ethnic groups as well. Between July 2020 and January 2021, the net favorability of BLM declined among white (+9 vs. -4), Black (+70 vs. +48), and API Americans (+31 vs. +19). Attitudes among Latino Americans were not appreciably different in these two periods (+27 vs. +29).

12 These estimates were calculated by comparing data from the first two waves in which this data was collected (July 2 through July 17, 2020) and the final two waves of the survey (January 12 through February 3, 2021).

The upshot is twofold. First, the effects of Floyd’s murder and the ensuing protests on views of the police and BLM were immediately apparent but ultimately not durable. Second, the longer-term trend from 2016 to 2020 is consistent with the trends described in the first section: Attitudes are becoming more progressive — in this case, less favorable toward the police and more favorable toward BLM — but mainly because of trends among Democrats and independents, not Republicans. This means attitudes toward the police and BLM are more polarized by party now than four years ago, much as are attitudes toward Black Americans and immigrants.

Shifts on Discrimination Against Black Americans and Racial Resentment Were Limited

Another survey question that potentially captures the impact of Floyd’s murder and the protests is about how much discrimination respondents believe that Black Americans face (Figure 9). Among all Americans, the proportion saying that Black Americans face “a great deal” or “a lot” of discrimination increased from 55 percent in the two weeks prior to Floyd’s murder to 62 percent in the two weeks after. Similar shifts occurred among all partisan and racial groups.

But again, these changes faded with time. In fact, the number of Americans who said that Black Americans faced high levels of discrimination was slightly higher in the two weeks before Floyd’s murder than it was in the last two waves of the survey, which were fielded between January 12 and February 3, 2021 (55 percent vs. 51 percent). This decline was driven by shifts among independents (46 percent vs. 41 percent) and Republicans (37 percent vs. 30 percent). Democratic views returned to their pre-Floyd levels (74 percent vs. 73 percent).

In addition to this question, Nationscape also asked two of the questions from the racial resentment scale presented in the previous section. Again, there was a noticeable shift in responses to both questions in the weeks after Floyd’s murder (Figure 10). Comparing the two weeks before and the two weeks after, there was an increase in the percentage of Americans agreeing that slavery and discrimination have created economic hurdles for Black people (from 47 percent to 52 percent) and disagreeing that Black people should overcome prejudice without favors (from 23 percent to 28 percent). The shifts on these two questions were driven disproportionately by changes among Democrats (from 62 percent to 70 percent and 35 percent to 43 percent, respectively). Changes among Republicans were negligible over this period.

Among all Americans, these changes were largely temporary. By early 2021, the percentage of Americans agreeing and disagreeing with these two statements returned to the levels observed in mid-May. But as with views of the police and BLM, there was evidence of continued partisan polarization. Although the percentage of Democrats agreeing that slavery and discrimination have created economic hurdles for Black people was still higher than in mid-May (67 percent vs. 62 percent), the percentage of Republicans agreeing was actually lower than before the protests (26 percent vs. 34 percent).13

In sum, the consequence of Floyd’s murder and the ensuing protests was not durable changes in public opinion among all Americans but increases in the polarization of Democrats and Republicans.

13 These changes were also calculated by averaging the Nationscape surveys conducted two weeks before Floyd’s murder and the last two weeks of surveys that were fielded between January 12 and February 3, 2021.

Taken together, the public opinion trends show that Americans’ racial attitudes shifted immediately after George Floyd’s murder, but ebbed in the months that followed. The lack of a durable shift among Americans overall, however, concealed a more durable pattern of growing partisan polarization. What can help explain both findings are the trends in news coverage and political rhetoric about Floyd’s murder and the related protests.

The Black Lives Matter movement and the broader issue of systemic racism received an extraordinary amount of coverage after Floyd’s murder. But as with news coverage of many issues, the attention it received was temporary. As the story faded from the news, its effects on opinion faded as well. In the absence of intense media coverage, opinions began to revert to their previous levels.

To show these changes in news coverage, we rely on closed captioning data of cable news broadcasts from the Internet Archive’s TV News Archive to identify the number of clips that mentioned “George Floyd,” “racism,” or “Black Lives Matter” (Figure 11).14 The spikes in coverage were immediately apparent. In the two weeks following Floyd’s death, there were thousands of cable news mentions of these topics.

14 See results from the TV News Archive. This analysis builds on Michael Tesler, “Support for Black Lives Matter Surged During Protests, But Is Waning Among White Americans,” FiveThirtyEight, August 19, 2020. Available at:

But beginning around June 10, the attention cable news paid to these issues began to decline. This likely reflects both the drop in protest activity and the media’s return to other issues, such as the pandemic and presidential campaign. By the end of June, there were only a handful of protests happening daily, whereas in late May and early June, there had been hundreds every day.(xiii)

This trend in media coverage is not surprising. A similar pattern — a steep spike and decline in quick succession — characterizes other phenomena. One example is high-profile mass shootings, which often get fleeting news coverage and thus have ephemeral effects on public opinion about guns and gun control.(xiv) It takes prolonged activism and sustained media attention for the impact of events like the Floyd protests to persist in public opinion.

The partisan polarization in public opinion also followed the re-emergence of polarized rhetoric about race and policing in the summer of 2020. The initial consensus among politicians following Floyd’s murder quickly gave way to the more familiar partisan divide. Conservative media began to focus on less popular elements of the movement, such as calls to “defund the police” and the very rare examples of violence or property damage at protests. Fox News, for example, was much more likely to cover both “defund” and “Portland” (the site of ongoing protests and some violent demonstrations) in the summer months of 2020 than CNN and MSNBC.15

These partisan differences were further magnified by divergent reactions to a Kenosha, Wisconsin, police officer’s shooting of a Black man, Jacob Blake, on August 23, 2020. On the political left, the Blake shooting sparked protests in Kenosha that extended to boycotts by NBA and WNBA athletes. But, unlike after Chauvin’s murder of Floyd, conservative politicians and media figures closed ranks around the Kenosha police. Attorney General William Barr, for example, contrasted the two incidents by stating: “Floyd was already subdued, incapacitated in handcuffs, and was not armed… In the Jacob [Blake] case, he was in the midst of committing a felony, and he was armed. So, that’s a big difference.”(xv) Barr’s description of the Blake shooting was hotly contested — Blake was shot in the back seven times — but this narrative quickly took hold on the right and helped polarize public opinion: Only 13 percent of Trump voters thought that the Blake shooting was unjustified, compared to 81 percent of Biden voters.16

15 See the volume of “defund” content by cable news channel and the volume of “Portland” content by cable news channel.

16 These results are from an August 27-28, 2020 YouGov poll.

Public Support for Affirmative Action and Reparations

The combination of long-term and short-term shifts in racial attitudes has resulted in an American public, and especially a Democratic Party, that is more consistently liberal on racial issues. People are, for example, more likely to see racial inequality as the result of structural forces, like discrimination, as opposed to a lack of individual effort on the part of Black people.

But previous research has long shown a gap between people’s willingness to endorse racial equality in principle and their support for policies that would help bring about equality in practice. This “principle-policy paradox” has been manifest, for example, in debates over school busing — which resurfaced in the Democratic presidential primary when Kamala Harris criticized Joe Biden’s opposition to a school busing plan when he was a Delaware Senator in the 1970s. Debates over reparations for slavery are another example. After some Democratic candidates endorsed reparations in the primary, now-President Biden established a commission to study the topic.

On at least one issue — affirmative action — support for policy has shifted in ways that parallel views about the origins of racial inequality (Figure 12). Between 2011 and 2020, the proportion of Americans who said they favored affirmative action programs for women and racial minorities increased from 29 percent to 44 percent. Once again, these shifts were driven primarily by Democrats, whose support increased 27 percentage points over these nine years. Independents and Republicans shifted in the same direction, but by much less. Among Democrats, the shift was also driven primarily by white Americans. Their support for affirmative action increased 34 percentage points between 2011 and 2020 — almost completely closing the gap between white Democrats and Black and Latino Democrats.17

17 To be sure, public opinion on affirmative action is sensitive to how the program is described. (See, for example, Laura Stoker, “Understanding Whites’ Resistance to Affirmative Action: The Role of Principled Commitments and Racial Prejudice,” 2008; See also Jon Hurwitz and Mark Peffley (eds.), “Perception and Prejudice: Race and Politics in the United States,” Yale University Press, 1998, Print.) Thus, we do not claim that these numbers are canonical estimates of support for affirmative action in the absolute. Our claim is merely that opinions have changed over time in ways that mirror party polarization on racial issues more generally.

In a similar vein, Gallup surveys show that support for reparations roughly doubled between 2002 and 2019 — from 14 percent to 29 percent. This shift was driven by Democrats (from 25 percent to 49 percent) and independents (from 15 percent to 32 percent). There was no significant change in support among Republicans (from 4 percent to 5 percent).18

But the principle-policy gap is especially large when it comes to reparations, which remains unpopular with the majority of Americans. For example, in post-election Nationscape data, only 39 percent of Democrats supported reparations, while 34 percent opposed it and 27 percent were unsure. There were also stark racial differences among Democrats — with two-thirds (66 percent) of Black Democrats supporting reparations compared to one-third of Latino (34 percent) and white (33 percent) Democrats. Support was even lower among independents (19 percent) and Republicans (12 percent).19

Even among those who agreed with the idea that slavery and discrimination have created obstacles for Black people, only 42 percent supported reparations. There were stark differences across racial lines too: 69 percent of Black Americans who agreed that slavery and discrimination have been hurdles also supported reparations, but just 43 percent of Latinos and 35 percent of white Americans did. Even in a group that seemed to accept the existence of systemic racism — white Democrats who agreed that slavery and discrimination helped create racial inequality — only 41 percent supported reparations.

The murder of George Floyd did little to change this (Figure 13). Support for this policy among all Americans was basically the same in the survey waves conducted immediately before and after Floyd’s murder (26 percent and 24 percent, respectively). At most, there was a small increase in support among Democrats. By early 2021, Democratic support for reparations was still somewhat higher than it had been in the weeks prior to Floyd’s murder (40 percent vs. 35 percent).20

18 See the Gallup trends.

19 Estimates in this paragraph and the following paragraph are from waves collected between December 3, 2020 and January 18, 2021.

20 Estimates in this sentence are from the two final Nationscape waves that asked about reparations. Data was collected between December 24, 2020 and January 18, 2021.

As these results demonstrate, the liberalization of racial attitudes, including within the Democratic Party, comes with important limitations. Even among groups that would seem predisposed to support policies that would mitigate racial inequality, such as white Democrats who acknowledge the importance of systemic racism, a substantial fraction does not. This fits a familiar historical pattern. Rising public support for racial equality during the civil rights movement often failed to produce corresponding increases in white Americans’ support for policy interventions that could help achieve that ideal.21 The lack of support for reparations among white Democrats is yet another manifestation of this longstanding principle-policy gap.

21 For a detailed review of this principle-implementation gap, see John Dixon, Kevin Durrheim, and Manuela Thomae, “The principle-implementation gap in attitudes towards racial equality (and how to close it),” Political Psychology, 2017, 38: pp. 91–126. For a more detailed investigation of white Democrats’ support for policies that would help Black Americans, see Meredith Conroy and Perry Bacon, Jr., “White Democrats Are Wary of Big Ideas to Address Racial Inequality,” FiveThirtyEight, July 14, 2020, Accessed September 2021. Available at:

Continuing Polarization in the Biden Era

Even with Donald Trump out of the White House, polarization on issues related to race, policing, and immigration has continued. Indeed, in some ways it has worsened.

The Derek Chauvin case is one example. The once bipartisan support for arresting Derek Chauvin in May 2020 eventually gave way to an enormous partisan divide over actually convicting him. In a poll right before the jury’s decision in May 2021, Biden and Trump voters were over 60 percentage points apart in their support for conviction (87 percent and 24 percent, respectively).22

This divide over the Chauvin verdict speaks volumes about the polarization of American public opinion on racial issues — especially comparing it to public opinion about another very prominent instance of police brutality against an unarmed Black man that was caught on camera. In 1992, when four Los Angeles police officers were acquitted of using excessive force in the videotaped beating of Rodney King, Democrats and Republicans were both overwhelmingly opposed to the jury’s decision. A May 1992 CBS News poll found that 81 percent of Democrats and 71 percent of Republicans disagreed with the verdict.(xvi) Thirty years later, though, even video-recorded incidents are not sufficient to break through the separate realities that partisans occupy on issues of race.

That polarization is growing even larger, too, as Democrats articulate a more progressive racial agenda under President Biden. Unlike the Trump-era polarization, further polarization could be driven by Republicans pushing back on progressive policies. Multiple surveys, in fact, already show that Republicans have become increasingly hostile to immigration as President Biden reverses many of the Trump administration’s restrictive immigration policies.

For example, in Pew Research Center polls, the proportion of Republicans who opposed allowing undocumented immigrants to stay in the country legally increased by 9 percentage points between June 2020 and April 2021.23 Similarly, polling by Civiqs shows a 12-percentage-point increase, since Election Day, in the percentage of Republicans who favor deporting undocumented immigrants.(xvii) This increase occurred immediately after the 2020 election and before the increase in border crossings in early 2021, suggesting that it was a reaction to Biden’s victory.

There is thus little reason to suspect that Biden will preside over any ratcheting down of the country’s identity politics. If Republicans’ negative reaction to his agenda continues, those politics should become even more fraught.

22 See full results and demographic breakdowns from this April 17–20, 2021, YouGov/Economist survey, conducted a few days before the verdict.

23 See the Pew poll results and similar trends in Reuters/Ipsos polling.


At least since Richard Nixon’s 1968 presidential campaign, race has been considered an effective wedge issue for the Republican Party — one that could splinter the Democratic coalition of nonwhite voters, racially liberal white voters, and racially conservative white voters.(xviii) But between 2012 and 2020, the Democratic coalition became far more consistently liberal on racial issues. Democrats are not completely unified, of course, particularly on certain policies that target racial inequality. But they exhibit more unity on many ideas and values related to racial, ethnic, religious, and national identities. In turn, this enables and encourages Democratic politicians to take stronger positions on racial equality. As The New York Times recently reported: “Unlike any other period in recent history, Democrats have infused their agenda with a focus on racial inequality, upending decades of conventional party wisdom that such appeals could prove politically risky.”(xix)

Within the Republican Party, there have not been clear trends on issues like race and immigration. On the one hand, this means that Republicans have not necessarily gravitated toward Trump’s positions. Indeed, previous analyses of VOTER Survey data have shown that Republicans actually have a wider range of views on immigration than Democrats do — suggesting that immigration could be a more effective wedge issue among Republicans, at least if the issue is framed in certain ways. But at the same time, the people who care the most about immigration tend to be Republicans who are more conservative on the issue.(xx) This is the faction of the GOP that Trump best represents — and one that appears increasingly powerful within the party, as the party’s focus on issues like critical race theory illustrates.

In his 1960 book, The Semi-Sovereign People, the political scientist E.E. Schattschneider famously argued that politics is about organizing the “scope of conflict” — that is, which issues are central to debates between contending sides. For a long time, a divided Democratic Party, subtler racial dog-whistling within the Republican Party, and a public characterized by “racial stasis” meant that identity-inflected issues were not always central to U.S. party politics. But in the past several years, a Republican president explicitly targeted racial, ethnic, and religious minorities, the Democratic Party liberalized on racial issues, and therefore American voters increasingly aligned their partisanship and their own racial attitudes.

As a result, Democrats and Republicans disagree not only about the appropriate size and role of government — spending, taxation, etc. — but about basic questions of what America is and who is, or can be, fully American. And because issues like race and immigration tap into strong emotional currents, politics feels more divisive than it might if the scope of conflict mostly concerned income tax rates or cost-of-living adjustments for Social Security benefits.(xxi)

That these conflicts have persisted into the Biden administration, and may even be growing larger, shows that they transcend the words and deeds of any president, including Donald Trump. Today’s identity politics is endemic in the political parties themselves. In this moment, it will prove difficult to organize politics any differently.


  • Appendix

    The results below represent the outputs of an OLS model on white respondents in the ANES and VOTER Survey panels. Dependent variables measure changes in attitudes towards Black Lives Matter, police, and racial resentment from 2016 to 2020. These dependent variables were constructed by subtracting 2016 attitudes from 2020 attitudes among white Americans who were first surveyed in 2016, and then re-interviewed in 2020. The coefficients for Trump favorability show the shift in these three attitudes from 2016 to 2020 that’s associated with moving from rating Trump very unfavorably to rating him very favorably back in 2016, after controlling for all of the other factors in the model. Trump favorability in the VOTER Survey data is a 4-category measure, recoded from 0 (very unfavorable) to 1 (very favorable); Trump favorability in the ANES is a 101-category measure, recoded from 0 (thermometer rating of 0) to 1 (thermometer rating of 100). The other explanatory variables, except for age, are also recoded from 0-1, with 1 representing the largest or most conservative value. All explanatory variables were measured in the 2016 waves of the VOTER and ANES panel surveys.

    Racial ResentmentBlack Lives MatterPolice
    Trump Favorability0.0757***












    Party ID-0.0171








































































    adj. R-sq0.030.0130.0740.0360.0560.03

    Statistical significance: *p<0.1, **p<0.05, ***p<0.01

  1. ↩︎
  2. ↩︎
  3. ↩︎
  4. ↩︎
  5. ↩︎
  6. ↩︎
  7. ↩︎
  8. ↩︎
  9. ↩︎
  10. ↩︎
  11. ↩︎
  12. ↩︎
  13. ↩︎
  14. ↩︎
  15. ↩︎
  16. ↩︎
  17. ↩︎
  18. ↩︎
  19. ↩︎
  20. ↩︎
  21. ↩︎

This report was updated on October 22, 2021 to include notes on API sample limitations in the data.


Don’t Miss Out

Subscribe to our mailing list for updates on new reports, survey data releases, and other upcoming events.

* Required field