Interracial cooperation in the United States

Media narratives and policy debates in recent years have often emphasized conflicts across racial and ethnic lines. Whether the topic is immigration policy, hiring, or college admissions, the issues are often framed as one group in competition with another.

This is ironic given that economic and cultural modernization have made interracial contact and cooperation more common than at any time in U.S. history. By “cooperation,” we mean voluntary and presumptively mutually beneficial exchanges, such as those that routinely occur in social and economic interactions. Now more than ever, the United States is a cooperative multiethnic and multiracial nation.

With this in mind, scholars at Gallup and the Brookings Institution’s Center for Community Uplift—a research and policy group dedicated to increasing well-being for people of all races—sought to document the extent of interracial contact as reported by Americans across domains of life, and analyze its effects on well-being and satisfaction. To do so, the research team surveyed a nationally representative sample of approximately 5,000 U.S. adults, asking them extensive questions about their relationships at work, their friendships, their romantic relations, and their family ties, with a focus on how frequently these relationships involve people from other races. The team also used a hypothetical scenario to elicit respondents’ potential racial preferences in a hiring decision. The survey also asked about various attitudes toward discrimination, and the importance given to race versus other characteristics when selecting friends and business partners.

This report analyzes only a subset of the complex, multifaceted data collected through this project as a way to introduce the data and inspire further research. Neither the report nor the survey data provide anything like a comprehensive overview of implicit or explicit discrimination at the individual or institutional scale, which is beyond the scope of this project. Rather, the research provides insights into people’s self-reported frequency and context of interracial cooperation across domains of life, how attitudes are associated with interracial exposure, and how interracial exposure affects well-being. The findings can be summarized as follows:

Explicit racial bias in business exchanges and friendship formation is rare. When evaluating whether to conduct business with someone, 87% of U.S. adults report that the race of a candidate is not important to them at all. When reflecting on friend selection, 83% give this response. Non-demographic, contextually relevant characteristics are rated as far more important to these decisions (e.g., a good reputation in the business context and shared values in friendship selections). There are no meaningful differences across political parties on these self-report measures.

These results align with findings from our hypothetical hiring experiments, which randomly assigned candidate characteristics, including race, and asked the likelihood of preferring one candidate over another. Race was much less important to hiring preferences than other factors. Among U.S. adults, 85% expressed preferences that were race-neutral; 10% expressed a preference favoring Black candidates; and 5% expressed a preference for white candidates. Race explained 4% to 5% of the variation in candidate preferences, whereas candidate qualifications such as interview quality, credentials, intelligence, and recommendations explained 30% to 44% of how people selected a candidate. There are only small differences by race in these preferences.

Non-Black adults (whether they are white or not) are more likely to prefer Black job candidates if they are exposed to more Black people through work, friendships, or dating history; if they believe discrimination has been especially harmful to Black people; or if they identify with the Democratic Party.

Most U.S. adults experience interracial relationships at work. Among all U.S. workers, 83% have a client from a different race than themselves; 80% have one or more co-workers of a different race; and just under half (48%) have a boss, manager, or supervisor of a different race. Despite this, the three largest minority racial groups (Asian American, Black, and nonwhite Latino or Hispanic Americans) have fewer different-race colleagues than would be expected based on national population shares of each group. Racial differences in county population shares can account for much of the differences, but remaining gaps are likely due to occupational segregation within the same labor markets.

Job satisfaction is unrelated to diversity at work overall, but this relationship varies for different racial and ethnic groups. Black workers are significantly more satisfied with their job when they have a higher percentage of non-Black co-workers; the data suggest that their job satisfaction is highest when the share of non-Black co-workers is roughly 70%, but the average Black worker reports that only about 60% of their co-workers are non-Black.

White workers and Latino or Hispanic workers are slightly less satisfied with their job when their bosses or managers are from a different racial or ethnic group than themselves. These two groups are also less satisfied when their co-workers are of a different race or ethnicity than themselves. These associations are modest, and most workers in diverse workplaces still report being satisfied with their jobs; the patterns likely reflect a mix of comfort with similarity, workplace norms, and the types of jobs in which more diverse teams are concentrated.

Interracial friendships are common and successful. Over half of U.S. adults report having a close friend from another race, and 72% have at least one friend from another race, whether close or not.

The quality and quantity of friendships are important components of well-being. To study the link between interracial friendships and well-being, we analyzed the association between friendship contentment and the share of friends who are from racial groups that differ from the respondent’s own race. Overall, there were no significant effects of close interracial friendships on friendship satisfaction among U.S. adults or for any specific racial or ethnic group. For non-close friendships, however, there was a small but significant negative effect of interracial friendships on friendship satisfaction for Black Americans, though not for any other group. This indicates that, in general, the racial composition of friendship networks does not significantly impact friendship contentment either way, except among Black Americans, who are slightly less content with their friendships when their non-close friends are from a different race. To illustrate, the percentage of Black adults who are content with their friendships falls by just 3 percentage points if the Black share of non-close friends falls by one-third. For Black adults, it may be that contentment with non-close friends from other races increases when they become close friends. Other explanations for the specific patterns observed among Black adults are possible, but the survey was not designed to answer questions about possible mechanisms or explanations for these patterns. For all groups, close friendships seem to be valued similarly regardless of whether they are with someone of the same race or not.

Interracial marriages and other familial relationships are growing, and interracial pairing is not related to the quality of those relationships. Among adults who are married or in a long-term romantic relationship, approximately 35% of those under age 30 are in an interracial relationship, compared to 14% of adults age 70 or older. For each of the largest racial groups, there is no association between being in an interracial relationship and the quality of those relationships as rated by the respondent.¹

Overall, these findings provide grounds for an optimistic assessment of America’s future as an increasingly diverse society. Large majorities of Americans reject explicit racial bias when it comes to forging friendships and business relationships. In the case of hiring, a small preference for Black job candidates is consistent with a desire to combat anti-Black discrimination (past and present), but those preferences are vastly outweighed by job candidates’ objective qualities. Together, these results suggest that most Americans prefer racial bias to not drive decisionmaking across important domains of life.

Moreover, past research consistently shows that exposure reduces racial and ethnic bias.²³⁴⁵Likewise, results from this survey show that exposure to Black people predicts pro-Black hiring preferences, and more generally, exposure to other races through dating predicts greater comfort when interacting with people from other races.

Given the connection between contact and prejudice, findings from this survey imply that explicit bias and prejudice in everyday interactions may fall in the future as most adults in contemporary U.S. society are exposed to people of different races on a routine basis, including as clients, co-workers, supervisors, friends, and family members. These relationships are rated as just as good as those with people from the same race, with very few exceptions (noted above). This suggests that people may be able to find common ground and build strong relationships with one another regardless of their racial background—a pattern that will contribute to the social and economic health of a multiracial and multicultural society like the United States.

Given the significance of interracial cooperation in our current and future lives, surveying the quality and frequency of Americans’ current interracial relations and their trajectory can provide insights into how to reduce racial and ethnic divisions as well as bridge gaps where they exist. This report investigates the scope and quality of interracial contact in Americans’ lives—at work, in friendships, and within families—and explores how these interactions relate to well-being, satisfaction, and explicit discrimination.

Despite major civil rights milestones, race remains a focal point of political and social conflict in the United States. In 2022, 41% of U.S. adults worried “a great deal” about race relations in the country, and today, only about half of American adults say race relations between white and Black people are good.⁶⁷Prominent policy debates pertaining to race relations have come up in recent legal cases involving affirmative action and gerrymandering.⁸

While racial conflicts have garnered media attention, longer-term trends suggest that the conditions for more cross-race exposure and collaboration are improving. Black-white residential segregation in 2020 stands at roughly half of what it was during its peak in the 1960s.⁹ Discriminatory attitudes, such as hostility to Black-white marriages or racial hiring preferences, have fallen dramatically since the middle of the 20th century.¹⁰¹¹ The racial gap in political party identification has reached a historic low in recent years.¹² And perhaps as a result of better race relations over the years, the Black-white gap in life expectancy fell from seven years to four years between 1963 and 2023.¹³

Yet data collected over the last 10 to 15 years show that Americans still exhibit a preference for people of their own race when it comes to dating, school selection, friendship, where they live, and even at what restaurants they eat.^14151617 These preferences are sometimes the result of physical segregation across U.S. cities, towns, and social networks, which can further contribute to racial divisions. Economically, homes and commercial properties are undervalued in Black-majority neighborhoods and opportunities for wealth accumulation and cultural transmissions are limited.¹⁸ Segregation not only restricts opportunity, but also perpetuates economic inequality by keeping resources, relationships, and capital separate across groups.¹⁹²⁰²¹

Same-race preferences are likely sustained and strengthened by lack of exposure to people of other races. Intergroup contact theory, proposed by social psychologists, demonstrates that sustained, cooperative interaction across racial lines, especially under conditions of equality, reduces bias and builds empathy.²²²³ Through repeated contact, perceptions shift, stereotypes weaken, and trust grows. Childhood exposure to Black students, for example, reduces white students’ bias and discriminatory attitudes both in the short and long term.^24252627 

Decades of behavioral and social science research explain both why racial divisions persist and how they might be bridged. Social identity theory and intergroup cognition research suggest that humans instinctively categorize others into “us” and “them” across a range of characteristics and arbitrary features.²⁸²⁹³⁰ This natural tendency can create favoritism toward one’s in-group and subtle exclusion of out-groups, even without overt prejudice.³¹³² For example, classical studies show that students who are assigned to groups based on eye color start to show biased and prejudiced behavior toward other groups, especially if differences between the two groups are reinforced.³³ Anti-Black racism took in-group bias to extremes when paired with cultural efforts to justify slavery, as historians discuss.³⁴

In the modern era, in-group favoritism by race (or discrimination against the out-group) is almost certainly economically inefficient, beyond moral concerns about prejudice and discriminatory decisionmaking. As economist Gary Becker laid out decades ago, discrimination in decisions that pertain to markets (e.g., decisions that a company makes to maximize profit) is irrational.³⁵ In a basic human-capital-based model, a company loses profits when it hires or promotes workers based on characteristics that are irrelevant to expected productivity, such as race, ethnicity, gender, and sexual orientation.³⁶

The same logic applies to friendships and intimate relationships, with the proposal that discrimination predicts a loss in enjoyment or the well-being benefits of relationships. Imagine, for example, the potential loss of opportunity for a rich, meaningful, and satisfying friendship between a Black student and a white student who decide not to be friends based on race, despite having much in common.

In the workplace, one theory of diversity in management science proposes that diversity expands the pool of useful knowledge, leading to learning and better group performance.³⁷ Yet other theories in management science suggest that people who share things in common (either demographic characteristics or cultural traits) may be more attracted to and more cooperative with one another, leading to better team performance. This may be particularly true if social categorization (negative assumptions based on race or other demographics) negatively affects the behavior of teams.³⁸ In other words, racial diversity in the workplace is likely to be most harmful to team performance when racial stereotypes are held and reinforced at the workplace. Yet contact theory suggests that exposure to out-group members—especially in instances of collaboration toward a common goal—will reduce prejudice even if individuals initially approach an interaction with stereotypes about the out-group. Consistent exposure and collaboration can, therefore, eliminate the potential negative effects of diversity.³⁹

Empirical evidence shows mixed findings on the effects of diversity on workplace outcomes. A meta-analysis of diversity and team performance finds that diversity by occupation, skill, or education has small positive associations with team performance, but demographic diversity (including by race, age, and sex) has small negative associations with the same outcome.⁴⁰ A meta-analysis of diversity based on differences in religion or country of origin finds no overall effects on performance.⁴¹ In some studies, when positive effects on creativity and job performance were found, simultaneous negative effects were also found on social cohesion at the workplace or job satisfaction.⁴²⁴³ Other studies, however, show positive effects of demographic diversity on performance, including quality of care and financial outcomes in the health care industry and financial performance at the business unit level across retail and hospitality industries.⁴⁴⁴⁵

Beyond workplace interactions, scholars have studied how diversity affects satisfaction with social relationships.⁴⁶ The most intimate relationships do not seem to be adversely affected by racial differences, according to a recent large-scale meta-analysis.⁴⁷ This contrasts sharply with summary research published several decades ago, which often found that marital satisfaction was lower when there were cultural or racial differences between partners. This was thought to arise, in part, because of disapproval from family or community members. Thus, rising approval for interracial marriage may make the marriage more enjoyable for couples.

With respect to friendships, previous work has found that subjective well-being increases with the share of same-race friends up to a point before declining.⁴⁸ These findings are interpreted in terms of a balance between cohesion in an individual’s larger community and their smaller network of people with whom they regularly interact. For example, an individual’s preference for exposure to only same-race interactions can undermine their sense of connection to their community if they live in a more diverse place.

Most communities in the United States are not monoliths, and population projections suggest that by 2050, current racial minorities will represent more than 50% of the U.S. population.⁴⁹ Strong preferences for same-race interactions at work, in families, and in friendships, therefore, do not only limit opportunities for rich and meaningful interactions with diverse others who collectively represent the currently multicultural U.S. society, but may also be maladaptive for the near future at an individual and societal level by restricting exchange and cooperation as well as limiting ties to homogenous groups.

Recruitment

Gallup mailed an invitation to 100,000 randomly selected U.S. addresses that were stratified within groups using four racial classifications (Black, Latino or Hispanic, white, and all other), two age groups (using age 30 as the cut point), and two education groups (with or without a bachelor’s degree). Black, Latino or Hispanic, and other households that had a higher likelihood of not being white (based on a set of demographic characteristics, including name and neighborhood racial and ethnic composition) were oversampled. Younger populations and less educated populations were also oversampled.

The overall response rate was 5% for the entire sample. Response rates could be estimated by race using the sample flags, which are not always accurate. These rates were as follows: 9% for white households, 2% for Black households, 3% for Latino or Hispanic households, and 3% for other households that did not identify as white.

Via a mail card, respondents were asked to take the survey online if they were at least 18 years old. A $1 bill was included in the mailer as a token of appreciation. The survey was administered through Qualtrics.

Sample

Overall, 4,883 adults completed the survey. Of those, 179 identified as American Indian or Alaskan Native; 472 identified as Asian American; 499 as Black or African; 606 as Latino or Hispanic; 88 as Middle Eastern or North African; 52 as Native Hawaiian or Pacific Islander; and 3,578 as white or European. Racial and ethnic categories were not mutually exclusive, and respondents could select more than one category.

Among those who completed the survey, 2,513 participants identified as female; 2,364 as male; and six did not respond to this question. The average age of the sample was 53 years old, with a standard deviation of 18 years, a median of 55 years, and a range of 18 to 100 years.

Results were weighted to represent the U.S. population. The estimates used for creating survey weights came from the 2023 American Community Survey, using the joint population distribution of race x age, race x gender, race x education, and race x region. The estimated margin of error is 2.1 percentage points for the sample.

The study was reviewed and approved by Gallup’s Institutional Review Board.

Survey questions

Interracial interactions by domain

Survey items asked respondents to think about specific relationships in different domains (e.g., colleagues, bosses, best friends, romantic partners, dating, familial relationships) and quantify how many individuals from each racial or ethnic group they have these relationships with.

For example, respondents were asked, “Think of bosses, supervisors, or managers you have worked with over the past 12 months. Enter the number of bosses, supervisors, or managers you worked for from each racial/ethnic group.” For close relationships, they were asked similar questions (e.g., “Think about your very close friends. How many of your very close friends identify as each racial or ethnic group?” or “What is your spouse or partner’s race and/or ethnicity?”). They were also asked to reflect on the race and ethnicity of individuals they have gone on dates with in the past 12 months.

In each domain, an index of hetero-racial exposure was calculated based on the share of relationships with people whose racial identity differs from that of the respondent. For example, hetero-racial exposure to close friends would be calculated as 75% if a Black person had four close friends, only one of whom is Black.

Explicit discrimination in deciding whom to interact with

Participants were also asked questions that measure the degree to which they would discriminate in decisions about friendship and business transactions based on an interaction partner’s race or ethnicity. Specifically, they were asked to rate how important the race or ethnicity of someone is to them when evaluating whether or not they want to be friends with them (on a scale of 1 to 5, 1 representing “not at all important” and 5 representing “extremely important”). Similarly, they were asked this question about deciding whether or not to conduct business with someone.

Participants rated the importance of other characteristics for each of these relationships, including sex, sexual orientation, political affiliation, shared interests, shared values, and education. Additionally, for friendship decisions, they were asked to rate the importance of other relevant features such as whether the person has friends or not, religious background, marital status, and age. For business decisions, respondents were asked to rate the reputation of the imagined partner and the favorability of terms governing the business transaction.

The relevance of race in hypothetical hiring scenarios

To mitigate concerns about social desirability in explicit answers about the importance of race in business transactions (i.e., the likelihood that respondents provide the answer that is socially desirable, regardless of their own private beliefs), we implemented an additional method that involves judgments about other characteristics in addition to race.

Participants were asked to imagine a hiring scenario in which they have to pick one of two candidates (Candidate A or Candidate B) for a job, based on information about each candidate on five of the following 10 characteristics. Each characteristic had two levels (the variation in these levels is shown in bold below) that were randomly assigned to one of the two candidates in question. 

• Race (Black; white)
• Years of relevant experience (one year; five years)
• Education (bachelor’s degree from Georgia Tech; bachelor’s degree from the University of Georgia)
• Recommendation status (highly recommended by former manager/neighbor for skill and work ethic; willing to provide references from past managers)
• Age (34; 28)
• Sex (male; female)
• Place of birth (born in U.S.; born in Mexico)
• Interview performance (gives excellent answers during interview; does not stand out in the interview)
• Intelligence (displays high general intelligence; seems to have average intelligence)
• Network connections (a friend of a friend; not in your social network)

One group of participants was asked to imagine hiring for a non-STEM field (home renovation), in which they saw the following prompt:

“Imagine you are hiring someone to renovate your home, using a design you have approved. If you had to pick between two candidates, knowing only the following five characteristics about them, which would you be most likely to hire?”

The other group was asked to imagine hiring for a STEM-related job (software application development), in which they saw the following prompt:⁵⁰

“Imagine you are hiring someone to contribute to the development of a new health-care-related software application. If you had to pick between two candidates, knowing only the following five characteristics about them, which would you be most likely to hire?”

Each participant had seven chances (experimental trials, or “tasks”) to pick one of two candidates, who were matched on at least one characteristic among the five that described their profile. In each trial, they were presented with the profile of each candidate (A and B) and asked how likely they would be to give the job to Candidate A or B. Responses were measured on a 5-point scale: “very likely to prefer A,” “likely to prefer A,” “I am neutral between the two,” “likely to prefer B,” or “very likely to prefer B”.

Our analysis took two forms. First, we used the pooled data from each candidate type (software or home renovation) to estimate the causal effect of each candidate characteristic. The main focus is comparing the effect of “Black race” on hiring preference to every other candidate characteristic. For the software job, this analysis was based on 17,941 observations from 2,563 respondents. For the home renovation job, it was based on 18,137 observations from 2,591 respondents. Errors were clustered at the respondent level and respondent fixed effects were included to average out idiosyncratic response patterns (e.g., always using the ends of the scale).

Second, we conducted another analysis to recover the causal effect of Black race on candidate preference at the individual level. To do so, we regressed candidate preference on Black race and an index of candidate quality. The latter is needed because trials had their own randomized combination of candidate qualities and not all characteristics were shown in each trial. The second variable is constructed as a collapsed index of all candidate characteristics; it is the predicted preference for each task based on a regression of candidate preference on every characteristic except candidate race. Thus, if a respondent task only included information on education, skill, experience, age, and race, the predicted candidate quality would be based on the average marginal effect of the first four qualities using data from all respondents.

Finally, for each respondent, we predicted the hiring preference based on race and predicted candidate quality (using everything but race). The coefficient for race and its error were used to calculate summary statistics, determine approximate significance (using a standard t-statistic threshold of 1.96), and calculate the r-value (effect size).⁵¹ Our preferred summary measure is the respondent’s r-value for Black race. A positive value suggests a pro-Black hiring preference, whereas a negative value suggests a pro-white hiring preference. If the point estimate included zero in 95% confidence intervals (corresponding to a t-statistic of 1.96), we concluded the effect was not significant.

Well-being and domain satisfaction

Satisfaction at work was measured by an 11-point scale, with 0 representing “not at all satisfied” and 10 representing “completely satisfied.” The question invited participants to rate, “Overall, how satisfied you are with your job, the work you do for pay” if they indicated that they received income from work in the past 30 days.

Satisfaction with friendship was measured on a 7-point agreement scale. Participants were presented with the statement, “I am content with my friendships,” and asked to rate their level of agreement from 1 (completely disagree) to 7 (completely agree).

Satisfaction with marriage/romantic partnership was measured on an 11-point scale, with 0 representing the “weakest and least loving relationship you can imagine” and 10 representing the “strongest and most loving relationship you can imagine.” The question asked participants to rate their relationship with their spouse or romantic partner if they were married, engaged, or in an exclusive romantic relationship.  

Other questions

The survey included other questions that measure a variety of behaviors, attitudes, and beliefs. Some of these include in-group favoritism, general well-being and life satisfaction, beliefs about the effects of historical racism on life today, and demographic characteristics including but not limited to race, age, gender, sex, education, and marital status. Some of these additional survey items are used for analysis and presented with the findings throughout the report. Others are used as control variables.

Classification and analysis of race

Following the latest U.S. Office of Management and Budget guidance, we asked each respondent the following item to identify race, allowing “Latino or Hispanic” to be a separate racial category:⁵²

“What is your race and/or ethnicity? Select all that apply.”

1. American Indian or Alaskan Native
2. Asian
3. Black or African
4. Latino or Hispanic
5. Middle Eastern or North African
6. Native Hawaiian or Pacific Islander
7. White or European

Although we over-sampled minority populations, there were not enough respondents to report American Indian or Alaskan Native, Middle Eastern or North African, and Native Hawaiian or Pacific Islander alone. We examined multi-racial combinations, and only Latino or Hispanic-white and American Indian-white groups had enough participants to report, and the latter only when no other restrictions were made (for employment status, for example).

We decided against reporting summary statistics for groups that contained unclear combinations of racial groups, and therefore limit reporting of most group-specific outcomes to the four largest single-race groups: Asian American only, Black only, Latino or Hispanic only, and white only. We sometimes include the two distinct multiracial groups for which the sample size met reporting standards: Latino or Hispanic-white and American Indian-white, along with a third miscellaneous group that includes all other multiracial combinations, as well as the single-race groups that had too few respondents to distinctly report on (1, 5, and 6 in the above list).

In this report, grammatical references to racial and ethnic groups adhere to the Brookings Institution style guide, which closely follows that of the Associated Press.⁵³

We asked respondents to report the race of contacts using the same categories. For friends and dates, we allowed them to identify the number of multiracial friends and then indicate which races those people belonged to. For example, if the person had one close multiracial friend, they could put a “1” next to any of the races, and we would count this person as one person from each group. This “double counts” a multiracial friend, but our subsequent analysis puts this value in both the numerator and denominator. If someone said they had one white friend and one multiracial Black-white friend, our calculation would suggest two-thirds of their friends are white and one-third are Black. This avoids having to ascribe an ancestral percentage to each multiracial person and/or counting people as half a person from an ancestral group.

For work contacts, we allowed people to indicate the number of people who are multiracial, but did not follow up by asking them to identify the specific races due to space and the unlikelihood that respondents have accurate information about their work contacts’ racial backgrounds. Therefore, we removed multiracial work contacts from the numerator and denominator of this category, because it would be unclear which specific groups they would be categorized as, and if treated as its own “multiracial” category, the members would come from very different multiracial backgrounds (e.g., Black-white, Asian American-white), which would yield ambiguous findings.

Across all domains, for each respondent, we calculated the number of contacts in each racial group and the total number of contacts, as the respondent reported. We then calculated the percentage of contacts in each group as the number of contacts in that group divided by the sum of contacts.

Same-race exposure is calculated as the percentage of contacts in a given domain that are from the respondent’s own racial group or groups. For multiracial respondents, we calculated the combined number of contacts for the groups identified (up to two). We did not calculate exposure to multiracial respondents with three or more racial identities. These cases accounted for fewer than 100 respondents. For example, for a Latino or Hispanic-white respondent, same-race exposure to a co-worker would include the percentage of white co-workers and the percentage of Latino or Hispanic co-workers. Exposure to other races is 1 minus the percentage of same-race(s) contacts.

Validation of survey data

The patterns of responses in our data indicate that respondents classified the race or races of their contacts with at least some accuracy. Exposure to other racial groups at the county level, as reported by census data, is significantly correlated with exposure to other groups as reported by respondents. County population exposure predicts different-race exposures to managers (r = 0.57), friends (r = 0.27), dates (r = 0.25), clients (r = 0.27), and co-workers (r = 0.38). County-level exposure to Black respondents, specifically, is significantly correlated with greater reported exposure to Black people. For example, this includes friends (r = 0.40), co-workers (r = 0.38), and clients (r = 0.29).⁵⁴ Across respondents’ racial groups, each group shows a higher preponderance of friends in their own racial group when they live in a county with a higher population share of that group.

The survey modules that explicitly measure the importance of considering race when making decisions about friends and business partners raise concerns about accuracy in reporting. Social desirability bias is always a concern when survey respondents are asked to share views on a controversial topic, and racial discrimination is certainly one of those.⁵⁵ For example, respondents may want to project high moral standing to themselves or researchers, even in cases when this does not match their sincere beliefs.

This is important to keep in mind when interpreting the results, but there are several reasons to think that the survey items are capturing valid aspects of discriminatory behavior and intent. First, there are no apparent repercussions to admitting on an anonymous survey that you believe race is an important factor when selecting friends, for example. Second, people who want to appear non-racist (or appear as if race does not matter to them) in the context of a survey are likely to want to appear non-racist in real-world settings, in which they could be accused of discrimination. Thus, concern about social desirability would be a deterrent to undesirable action in both survey-taking and actual behavior, although this does not limit the possibility of undesirable action in circumstances in which accountability is limited or under conditions of anonymity. In general, our estimates of explicit discrimination should be interpreted as lower-bound indicators rather than precise measures of people’s true attitudes or behaviors.

We also have two modules that measure potentially discriminatory attitudes in less explicit ways.

Later in the survey, we asked a series of questions about level of comfort and trust concerning people of different races. The scale was meant to measure in-group bias. This measure is strongly correlated with the importance given to race in hiring and in business (r = 0.43). This is true for single-race Black respondents (r = 0.42) and single-race white respondents (r = 0.43).

Moreover, among white respondents, there was a strong negative correlation between the importance given to race and the degree to which they believed that anti-Black discrimination affected Black people in the 1950s (r = -0.28) and today (r = -0.16). In other words, white respondents who say they give more importance to race in choosing friends and business partners also express greater in-group bias and tend to say that discrimination had small effects on the lives of Black people in the past and has small effects in the present.

Finally, our conjoint study is only moderately explicit as a measure of discrimination in hiring. The module does not ask people to consider the race of candidates or whether they prefer the white candidate or the Black candidate. Rather, respondents decide which candidate they prefer, if any, given a randomized set of candidate qualities; bias is calculated based on the probability of preferring Black (or white) candidates given the average importance that respondents give to the other specific qualities shown. As in real life, respondents can decide to override an initially biased belief and adjust their evaluation, but the fact that race is only one characteristic out of several obscures potentially discriminatory preferences. Bias is revealed in the final analysis.

Among all respondents and among white respondents, we see a negative relationship between people who give more weight to race and pro-Black hiring preferences (r = -0.12 and r = -0.21, respectively). This suggests consistency between people who are explicitly discriminatory and those whose discriminatory preferences are revealed analytically through the conjoint.

Limitations of interpretation

While there is some logical consistency across our survey items that capture different aspects of discrimination, all of these measures pertain to hypothetical or self-reported answers that do not necessarily correspond to the respondent’s real-world behavior in the past or future. We strongly caution against using these data to conclude that a certain percentage of American adults do or do not discriminate in everyday decisionmaking based on race. Discrimination is a complex, multi-domain construct that includes explicit actions and behaviors as well as those that may be more subtle or indirect. No single-item survey measure can possibly encompass it, and any scale will have important limitations.

Workplace

Interracial exposure and diversity in the workplace

How diverse are America’s workplaces today? Nearly half of U.S. workers (48%) report having at least one boss, manager, or supervisor of a different race, and 81% report having at least one co-worker or client of a different race. Thus, interracial contact is very common across the workplace, especially among co-workers and client relationships.

Most workers have at least some exposure to co-workers or clients of a different race. Exposure to different-race bosses or managers is less common overall and across all groups when compared to exposure to different-race clients or co-workers, but for white Americans, exposure to different-race bosses and managers is much less common.

The relationship between each group’s population share and reported interracial contact can be directly considered by comparing potential interracial contact to actual interracial contact. We do this next by showing the percentage of co-workers who are from a different race and comparing this number to the percentage of all U.S. co-workers who are from a different race than the respondent using census data.⁵⁶ These results contrast with those above in two important ways: the overall interpretation and the race-specific results.

First, it is clear that workers are more likely to work with others of the same race than their population share would predict, but this varies widely by race.

For example, 60% of Black workers’ clients are non-Black, but 88% of the U.S. adult population is non-Black. The implication is that Black client networks are heavily tilted toward same-race interactions. About half of this is evidently due to residential segregation: At the county level, our average Black respondent lives in a county that is 74% non-Black, which accounts for about half of the same-race orientation. For Asian American workers, 88% of their clients are non-Asian American, which is close to the non-Asian-American adult population share of 93%. For Latino or Hispanic and Asian American workers, exposure to other groups at the county level explains the entirety of the gap. White workers are more likely to have nonwhite clients than the nonwhite population share would predict using either national population share or local county population shares.

For co-workers, all four large racial groups are less likely to have different-race co-workers than their population share would predict, though the gap is very small (1 percentage point) for white workers and entirely explained by county population patterns. Black and Latino or Hispanic workers are the least likely to have different-race co-workers given their population shares. This extends to bosses and managers: Black and Latino or Hispanic workers are much less likely to have a non-Black or non-Latino or Hispanic manager than non-Black and non-Latino or Hispanic population shares would predict. Only 64% of Black workers’ bosses are non-Black, but 91% of all U.S. managers are non-Black. By contrast, 28% of managers are nonwhite, while 19% of white workers’ managers are nonwhite—a smaller gap than for Black workers.

These findings could partly be due to occupational segregation. Black and Latino or Hispanic workers are overrepresented in certain industries, and therefore more likely to be exposed to same-race work relationships. For example, in 2024, 27% of all workers in health care support occupations were Black, while just 13% of U.S. workers overall were Black. The same year, Latino or Hispanic workers were disproportionately represented in natural resources, construction, and maintenance occupations: 36% of workers in these occupations were Latino or Hispanic, whereas just 19% of U.S. workers overall were Latino or Hispanic.⁵⁷

Beyond race, the age of workers is somewhat but weakly related to exposure to different-race contacts. Even though the non-Latino or Hispanic white share of the U.S. population is much lower among 18- to 29-year-olds (56%) than among those age 70 and older (78%), exposure to different-race work relationships is fairly balanced across age levels, reflecting the multigenerational nature of workplaces. Client relationships show the largest sensitivity to age: Workers under age 30 are much more likely to have an interracial client than workers age 70 and older (84% versus 70%).

Political party affiliation has no meaningful connection to interracial exposure. There is no substantial difference between Democrats and Republicans in whether they have interracial relationships at work. Around 80% of Democrats, Republicans, and independents experience diversity at work in interracial relationships with co-workers or clients, and roughly half have a boss or manager from a different race than themselves.

Job satisfaction and interracial exposure at work

Prior findings have shown that racial diversity has unclear implications for job satisfaction.⁵⁸ Research from social psychology suggests that interracial interactions are often accompanied by concerns about appearing prejudiced among white individuals and anxiety about being the target of prejudice among racial minority individuals.^59606162 These concerns lead to interactions that are experienced as effortful, shaping a more negative evaluation of the interaction than same-race interactions.⁶³⁶⁴ Findings from our survey reflect these dynamics in the workplace context.

Overall, there is no association between interracial exposure at work and Americans’ job satisfaction. However, experiences vary for different racial and ethnic groups.

For most groups, there are small negative associations between job satisfaction and exposure to people from different races at work. Yet the opposite is true for Black workers, who are more likely to express job satisfaction when working with non-Black co-workers, managers, and clients.

For both Latino or Hispanic workers and white workers, working with bosses and managers who are from a different race predicts lower job satisfaction. No other group shows a significant relationship, and the association is positive only for Black workers.

For Asian American workers and white workers, having clients from a different racial group is associated with lower job satisfaction—but again, no other group shows a significant effect from exposure, and Black workers show a positive effect that is not statistically significant.

For Asian American, Latino or Hispanic, white, and Latino or Hispanic-white workers, having different-race co-workers is associated with a small and significant negative effect on job satisfaction. For Black workers, there is a significant positive effect. Controlling for household income and employer size reduces the size of the effect for Black workers, but does not entirely explain it. It may be that workplaces with a substantial share of non-Black workers are more likely to have better employee benefits or more engaging management practices, perhaps as a result of differences across industries, occupations, or level of capital intensity.

Together, these findings suggest that interracial workplace interactions have varied effects on job satisfaction. Since interracial exposure is not randomly assigned, we cannot know for sure whether exposure truly affects job satisfaction or is merely correlated with other geographic- or company-specific characteristics. For all groups, whether they are in the minority or majority, repeated exposure reduces stereotypes and may ease initial discomfort; through continued collaboration, interracial interactions become more natural, fluent, and rewarding.⁶⁵⁶⁶ Sustaining diversity, therefore, may enhance its effects on outcomes such as job satisfaction, productivity, and long-term cohesion in the U.S. workforce.

Racial discrimination in hiring decisions: A thought experiment

Anti-Black hiring discrimination has been well documented in the social science literature, usually through audits and field experiments that reveal small but consistent bias against candidates with stereotypically Black names.⁶⁷⁶⁸ However, other evidence from surveys of the general population shows declining explicit anti-Black bias in people’s self-reports.

Before the Civil Rights Movement, nearly half of U.S. adults said they would give a hiring preference to white candidates over Black candidates. From 1946 to 1963, this share fell from 45% to 15%, and nine years later in 1972, it was only 3%.⁶⁹ In 2003, 30% of adults agreed with the statement, “We should make every possible effort to improve the position of blacks and other minorities, even if it means giving them preferential treatment.”⁷⁰

In more recent years, Americans’ views have shifted drastically toward favoring neutrality. In 2019, 74% of U.S. adults agreed that “when it comes to decisions about hiring and promotions, companies and organizations should only take a person’s qualifications into account, even if it results in less diversity,” and only 24% believed a person’s race and ethnicity should be taken into account, in addition to qualifications.⁷¹ In 2024, 2 in 3 Americans reported that the Supreme Court’s June 2023 ruling that ended the use of race and ethnicity in university admissions decisions was “mostly a good thing.”⁷² Two years before that ruling, a majority of Americans (62%) indicated “generally favoring affirmative action programs for racial minorities.”⁷³ Implicit biased attitudes also seem to have shifted over the last decade toward neutrality for specific characteristics, including sexual orientation, race, and skin tone.⁷⁴

Against this backdrop, findings from the current survey’s hypothetical hiring experiments offer a nuanced perspective of explicit bias in hiring preferences. When asked to evaluate a potential business partner, most Americans did not show explicit racial bias. Only 7% rated race as “somewhat,” “very,” or “extremely” important when deciding whether to conduct business with someone; just 2% rated it as “very” or “extremely” important. Instead, characteristics directly related to the task, such as professional reputation and favorable business terms, dominated decisionmaking, with 78% and 66% of respondents citing them as very or extremely important, respectively.

Responses differed across groups. White Americans were least likely to consider race relevant in business decisions: 91% rated it as “not important at all,” while about 4% indicated that race is at least somewhat important to their decisions. Black Americans were the most likely to indicate that race is at least somewhat important: 72% rated race as “not important at all” while about 1 in 5 rated race as at least somewhat important.

For some Black respondents, this emphasis may reflect efforts to avoid negative discrimination or stereotypes that may be financially or psychologically costly.⁷⁵ Alternatively, it may be a practical strategy for supporting the economic development of a local community. Black Americans have the lowest share of wealth in the U.S., and even though wealth has increased in the recent years for U.S. households of all racial and ethnic backgrounds, the wealth gap between white and Black families has also increased.

Among Black Americans, attention to race in business decisions may also signal a practical response to perceived barriers resulting from discrimination. Black participants who believed that anti-Black discrimination affects most aspects of Black people’s lives today were more likely to consider race important when deciding whether to conduct business with someone. Of these respondents, 31% said race was at least somewhat important, compared with 12% among those who believed that anti-Black discrimination does not have a meaningful effect of the lives of Black people today.

This pattern suggests that the salience of race in economic relationships may be influenced by how individuals understand structural inequality and the effect of discrimination on their own lives. For many Black Americans, prioritizing business within their community may represent a strategy to build collective wealth and counter the enduring effects of historical exclusion. These findings reflect how experiences with and interpretations of the same circumstances in society may influence perceptions of interracial relations differently for different demographic groups.

Hypothetical candidate selection

To better understand how Americans weigh race in hiring decisions, the survey included an experimental scenario. Participants were presented with two fictional job candidates (“Candidate A” and “Candidate B”), and were asked to choose which one they would be more likely to hire. Each candidate was described by five of 10 possible attributes that varied randomly between trials, including race (Black or white), years of experience, education, recommendation status, age, sex, place of birth, interview performance, intelligence, and network connections. The Methods section of this report provides details about the design and calculations for this task.

Overall, race played a small but significant role in predicting candidate preferences for both jobs. In both cases, Black candidates were more often preferred over white candidates. Still, the average participant gave much greater weight to nonracial factors when expressing candidate preferences. Candidate race explained 4% of the variation in hiring preferences for the home renovation scenario and 5% for the software developer role. By contrast, the quality of the interview explained 37% and 44% of the variation, respectively. For the software developer role, candidate experience, intelligence, having a bachelor’s degree in computer science, and having high recommendations from previous managers were also much more important factors than race in predicting hiring preference. For the home renovation job, having a construction certificate (instead of an English degree) had a notable positive effect on hiring, as did years of experience, a recommendation from neighbors, intelligence, and being a friend of a friend. These factors were more important statistically than race. Sex, age, and nativity were less important than race for both jobs. Female candidates were given a slight edge for the software developer job, but sex had no effect in the home renovation job.

Large majorities made effectively race-neutral decisions: 85% of all adults for the software hiring scenario and 84% for the home renovation scenario. Only 16% of Black respondents expressed a significant preference for Black candidates in the software hiring scenario, and 14% in the home renovation scenario. Those values were 8% and 9%, respectively, for white respondents with a significant Black candidate preference. No group expressed a pro-white preference in higher than 10% of cases, and on balance, the average member of every group was more likely to be pro-Black than pro-white, except Asian American respondents in the software scenario.

The absence of widespread anti-Black bias in this experiment contrasts with effects observed in audits and field experiments scholars have conducted.^76777879 In most of these studies, the research team sends out fake resumes to employers and records whether the employer attempted to contact the candidate. Using random assignment, the candidates differ only by names, which are meant to signal race by using stereotypical African American names or names popular among white candidates. A recent study found a mean effect size of r = -0.03, suggesting a small anti-Black effect in hiring.⁸⁰ By contrast, our effect sizes are not only larger, but in the opposite direction (r = 0.23 and r = 0.19), showing a pro-Black preference.

There are two differences between previous studies and the current survey: 1) Most respondents in our data are unlikely to be involved in hiring decisions, though our home renovation scenario would be relevant to many people who consider hiring independent contractors; and 2) we identify race explicitly, rather than relying on stereotypical names. This removes ambiguity about the candidate’s actual race and overcomes the class-race problem, in which using the most stereotypically Black names confounds race with low levels of parental education.⁸¹ We also gave our respondents more information than what is typically found on a resume, in that we included an overall measure of candidate intelligence, interview quality, and recommendation quality. The addition of this information may have encouraged participants to deliberate on the decision more carefully—and among potentially biased participants, this information may have overridden held racial stereotypes. A limitation of our method is that this is a hypothetical situation.

Consistent with contact theory, we find that the degree of preference for Black candidates among our respondents increased with exposure to Black people. To test this, we set up a model to predict the degree of pro-Black candidate preference as a function of exposure to Black people.⁸² We limited the responses to non-Black adults. We used several exposure measures: the share of co-workers who are Black, the share of recent dates who are Black, and the share of friends who are Black. In each case, greater exposure to Black people predicted a significantly stronger preference for Black candidates. The effects were largest for dates, followed by friends and co-workers.

Additionally, we find that Black respondents who view anti-Black racial discrimination today as especially harmful to their lives are more likely to exhibit a pro-Black hiring preference. Likewise, Democrats were more likely than Republicans to express a pro-Black hiring preference, consistent with political views on affirmative action policy in hiring. Other research shows that Democrats and Black Americans are more likely than other groups to approve of hiring decisions that “take race/ethnicity into account, in addition to qualifications, in order to increase diversity.”⁸³

Friendships

Friendships are among the most personal forms of social connection—and, unlike work or family relationships, they are almost entirely voluntary. Who we choose to befriend reflects our values, preferences, and social opportunities. Yet despite a more diverse nation, Americans’ friendship circles remain segregated, according to surveys on the topic.⁸⁴ Many American adults may not seek interracial friendships because they feel like they have little in common with people from other racial or ethnic groups or due to external barriers such as segregation in neighborhoods, schools, and possibly the workplace.⁸⁵

Those who do form interracial friendships, however, show measurable benefits in various studies on the topic. Students who interact across racial lines exhibit higher intellectual engagement, stronger academic skills, and greater civic involvement.⁸⁶ Research using randomly assigned college roommates finds that interracial pairings lead to more positive attitudes toward out-group members and fewer stereotypical assumptions.⁸⁷⁸⁸⁸⁹

Interracial exposure and diversity in friendship networks

Overall, the majority of Americans (54%) report having a close friendship with someone of a different race than themselves. “Close” friends are defined as those you could turn to for support when things are difficult. Among all respondents, 8% of adults reported no close friends, and the average adult reported three.

Consistent with their population size, white adults were the least likely to report having a close friend from another race, but nearly half (48%) still indicated that they did. Multiracial adults generally, as well as multiracial Latino or Hispanic-white (71%) and American Indian-white adults (74%) were among the most likely to have friends from racial groups different from their own. For a Latino or Hispanic-white adult, this means a friend who is neither white nor Latino or Hispanic. Asian, Black, and Latino or Hispanic Americans reported close interracial friendships at similar rates, ranging from 59% to 64%.

Beyond close friends, 64% of U.S. adults report having another friend from a different race, and 72% report either a close or less-close friend. “Other” (less-close) friends were defined as people with whom the respondent regularly communicates with and enjoys being around, but otherwise would not turn to for support. On average, adults reported seven such friends. Every group except Black Americans reported more diverse friendships among more distant friends relative to their closest friends. In other words, the closer the friend, the more likely that they are of the same race as the respondent.

Local population shares are relevant as to whether friends from other races relative to one’s own race are available, both in the formation and sustainability of friendships. Thus, we compared the share of friendships that are interracial to the county population share of each group.

Asian American, Black, and Latino or Hispanic adults consistently have fewer friends outside of their own racial group relative to the county population share of other races. For example, 84% of adults are non-Asian American in the average county where Asian American respondents to our survey live, but only 46% of the friends in Asian American social networks are non-Asian American. Likewise, the average Black adult in our survey lives in a county that is 73% non-Black, but among the friends of Black people, only 29% of friends are non-Black (a 44-percentage-point difference, after rounding). Adults who are white alone or in combination with American Indian or Latino or Hispanic ancestry have friendship networks that are somewhat more racially diverse than county population shares would predict.

Non-racial demographic characteristics are related to interracial exposure, albeit weakly. Younger people have slightly more diverse friendships than older people when measured by whether they have an interracial friendship or as percentage of friends from other racial groups. Democrats and Republicans do not significantly differ in whether they have any friends from another race, nor do independent/third party supporters.

Relationship satisfaction and interracial friendships

The quantity and quality of friendships are highly important to well-being and mental health.⁹⁰ In the present survey, both the number of friends a respondent reported and their contentment with friendships predict higher life evaluation on a 0 to 10 scale (the “Cantril Ladder”).⁹¹

Interracial friendships may require additional emotional effort.⁹² When individuals do not share similar experiences or cultural backgrounds, even well-intentioned interactions can feel less fluent. This “extra work” may explain why friendship satisfaction can be lower among people with more diverse networks. For example, at least in the context of college dormitories, interracial roommates have been shown to experience less positive emotions and less overall satisfaction with each other compared to same-race roommates.⁹³⁹⁴

Findings from the current survey support this pattern, but only among Black respondents. Overall, having a more diverse friendship network is not associated with lower levels of contentment with one’s friendships, but limiting the analysis to Black respondents only, having more friends who are not Black is associated with being less content with one’s friendships. Interestingly, even among Black respondents, this negative association holds only for relationships with friends who are not close friends of the respondent. In relationships with close friends, there is no association between relationship satisfaction and the racial or ethnic diversity of one’s friends among any of the groups. This suggests that close friendships can form and possibly offer well-being benefits regardless of racial or ethnic background.

Racial discrimination in friendship decisions: A thought experiment

While many Americans gravitate toward same-race friendship circles (partly due to structural segregation), explicit racial bias in self-reported friendship preferences is not common. When asked what factors matter most in deciding whom to be friends with, only 2% of Americans rate race as “very” or “extremely” important; 15% indicate that it is “somewhat” or “slightly” important; and 83% indicate that it is not important at all. Shared interests, values, and personality ranked far higher.

The lack of explicit and strong importance given to race when thinking about friend selection is similar across racial groups. A majority of Americans in each group say that race is “not at all important” when selecting friends. Single-race white Americans, single-race Latino or Hispanic Americans, and multiracial Latino or Hispanic-white Americans are the most likely to say race is not at all important. Black and Asian Americans are the least likely to do so. No more than 3% in any group say it is extremely important. Black Americans are the most likely to say that race is very or extremely important, but only 8% indicate this (combining “very” with “extremely”).

The varying levels of importance assigned to race in making friendship decisions can also speak to different preferred models of diversity. Majority-group members in American society have been shown to prefer a colorblind approach to diversity, which is consistent with attributing little value to someone’s race in all social and political decisions, including in hiring and education.⁹⁵ The findings above—especially higher percentages of white Americans indicating that race is “not important at all” to their friendship decisions—suggest that the colorblind approach to diversity may also apply to interpersonal decisions, at least among white Americans. On the other hand, racial and ethnic minority groups are more likely than white Americans to endorse the multiculturalist model of diversity, which emphasizes racial and ethnic distinctions and encourages these distinctions to be celebrated and acknowledged in policy and in day-to-day interactions. One potential consequence of a multiculturalist approach to diversity may be the higher likelihood to “see” and be attracted to interaction with people of the same race or ethnicity as one’s self.

As with business relations, beliefs about racial injustice may further shape attitudes about whom to interact with. Among Black Americans who share the belief that discrimination affects most aspects of life today for Black people, 31% consider race at least slightly important when choosing friends, compared with 12% among those who see no meaningful effect of discrimination on their lives. One possible explanation is that concerns about being a victim of discrimination from out-groups motivate affinity toward one’s own group. Still, race is deemed unimportant in selecting friends even among most Black adults who view discrimination as affecting most aspects of their life today.

Taken together, the data show that Americans rarely endorse explicit racial exclusion, at least when thinking about how they make decisions about friendships. Yet taken together with self-reports about the extent of actual interracial friendships, the findings suggest that structural segregation and unequal opportunities for contact still may limit interracial friendships. Expanding interracial friendships may therefore require both sustained exposure and the intention to be in close relationships with others regardless of race or ethnicity.

Family and romantic relationships

Interracial families represent the deepest form of racial cooperation. They bridge racial differences through daily life—shared homes, children, and experiences that demand mutual understanding.

The prevalence of interracial marriage has increased dramatically over the past four decades. In 1980, only 3% of married couples were interracial; by 2015, that figure had grown to 10%.⁹⁶ The largest gains have occurred among white and Black Americans, rising from 4% to 11% for the former and from 5% to 18% for the latter. Asian Americans remain the most likely to marry across racial lines (29% in 2015), followed by Latino or Hispanic Americans (27%).⁹⁷

Approval of interracial marriage has also transformed. In 1958, when Gallup first asked about marriages between Black and white Americans, only 4% of adults approved. By 2021, 94% expressed approval.⁹⁸ People’s actual choices reflect these changes.

Interracial marriages or partnerships are not uncommon among Americans today. About one-fourth of all adults report having a spouse or romantic partner who has a different racial or ethnic background than themselves. Asian American adults report the most interracial romantic relationships, and white adults report the least.

Reflecting the percentages of interracial romantic partnerships, about one-fourth (23%) of Americans report having children in the household from a different racial or ethnic background than themselves. This share is notably higher than the percentage reporting any parents of a different race or ethnicity (11%).

Rising rates of interracial marriage affect the contacts of extended family members. If, for example, someone’s biological aunt or uncle marries someone from a different race, then the respondent will have an aunt or uncle from another race and cousins if there are new children born through that marriage. In our survey, 19% of respondents report having an aunt or uncle from a different race, and 25% report having a cousin. This will likely increase in future generations as a result of higher numbers of interracial marriages.

Indeed, there are large age differences in exposure to people from different races at the family and extended-family levels. Specifically, younger people are much more likely to report being in an interracial romantic partnership: about one-third of each of the younger groups (18- to 29-year-olds and 30- to 39-year-olds) versus only about 1 in 7 among Americans who are age 70 or older. Similarly, the youngest group (18- to 29-year-olds) is twice as likely to report having a parent of a different race than the oldest group (age 70 and older). These differences speak to the rising prevalence of cross-race romantic partnerships.

Reports of interracial children in the household reveal a more complex pattern. There are no consistent age differences in the percentage of people who report having interracial children at home; across all age groups, 23% do. Among older adults, those who report interracial children in their households (for instance, about 25% of individuals age 70 or older) likely reflect grandparenting roles. Note that these data are based on children currently living in the home, so the children themselves would have been born in 2007 or later to be under age 18 when the survey was conducted.

The differences in reported interracial family relationships are very small between Democrats and Republicans. Across all party affiliations, 20% to 29% report having interracial romantic partners or children in the household. However, Democrats are twice as likely as Republicans to have interracial parents (16% versus 8%).

Intimate relationship satisfaction and interracial marriage

Like other forms of interracial contact, interracial partnerships bring both enrichment and complexity. Findings from the current survey show that individuals in interracial relationships report only slightly lower relationship satisfaction than those in same-race unions.⁹⁹ Less than 1% of the variation in relationship quality is explained by whether the partner is of a different race.

When observing this association within each racial and ethnic group, there are no significant associations between relationship quality and interracial partnership or marriage among any of the single race or ethnic groups. This suggests that satisfaction in intimate relationships might have little to do with racial or ethnic differences.

Dating

Interracial dating provides another context for cross-racial connection. Having dated a diverse pool of people is associated with ending up in an interracial marriage.¹⁰⁰ As it happens, 52% of all adults report having gone on a date with someone from a different race. Remarkably, this varies little across racial groups. A slightly smaller percentage of white Americans report having gone on a date with someone from a different race compared to Asian, Latino or Hispanic, and Black Americans. If dating were random across race, one would expect white adults to have fewer interracial dates than other groups, given that they have a smaller pool of potential dates outside their race. For example, 21% of adults are nonwhite in the county of the average white respondent in our sample. For the other large racial groups, out-group populations comprise at least 60% of the local population.

There are larger differences in interracial dating by age. A majority of adults under the age of 60 report having had a date with someone from a different race, whereas only 37% of those 70 or older report having done so. Slightly fewer Republicans than independents or Democrats report having gone on dates with people from a different race.

Dating interracially appears to be associated with one’s level of comfort in interracial interactions more broadly. Lower levels of anxiety in these interactions may lead to more natural and authentic interactions in which both parties can approach the interaction with curiosity and openness. Among those who have dated interracially, 53% disagree with the statement “I feel more comfortable interacting with people from my own racial group than those from other groups”; only 40% of those who have not dated others of a different race or ethnicity from themselves disagree with this statement. This association seems to be especially strong among white, Latino or Hispanic, and Asian American adults.

These patterns are consistent with decades of research showing that sustained, personal contact breaks down stereotypes and fosters empathy, although it is also possible that people who already feel more comfortable around others who don’t share their racial or ethnic background seek interracial dates more often. Still, interracial dating, while private, may be a powerful avenue for reducing bias through the desire to learn about and get to know each other, and potentially be friends or partners in life.

Across every domain of American life, including workplaces, friendships, and families, this report reveals both encouraging progress and persistent limits in racial cooperation. Explicit racial bias in people’s self-reports is rare, yet interracial contact, as reported by the number of friends and other relationships, is often below what would be expected from population share estimates.

The paradox of contact

Most Americans today report some level of interracial interaction, especially in the workplace. The workplace is an important context where individuals from different racial and ethnic backgrounds regularly collaborate toward shared goals. Findings from the survey show varying patterns of association between workplace diversity and job satisfaction among different racial and ethnic groups. The data here add nuance to prior research showing that workplace diversity may adversely affect comfort or productivity. For some groups (e.g., Black Americans), diversity is associated with more job satisfaction; for others, it is associated with small or no effects.

Friendships and family relationships tell a more personal story. While most Americans say race is not important in choosing friends, the racial homogeneity of social networks suggests otherwise. This is likely because residential segregation and limited exposure continue to confine opportunities for interracial friendship. Those who do form friendships with people of other racial and ethnic backgrounds than themselves are as satisfied with those relationships as they are with relationships with their same-race friends. Only among Black respondents is having friends of a different race slightly associated with lower satisfaction. The data also show that attitudes toward racial equality—such as beliefs about the how anti-Black discrimination affects the lives of Black people today—are related to how individuals think about the importance of race in friendships.

In intimate and familial relationships, the rise in interracial marriage and dating may indicate growing comfort and opportunity for connection. These relationships, while still less common among some groups, are increasingly normalized. Small differences in reported satisfaction between interracial and same-race couples may reflect external social pressures rather than interpersonal conflict.

From attitudes to actions and structures

If explicit bias has declined, as indicated by self-reports, why does segregation still persist across residential areas and in actual patterns of friendships and dating?

There are several explanations. One is that racial groups genuinely differ in their cultural preferences for the things that matter most in relationships, in ways that make it easier to share same-race friends or romantic partners. While this is possible, the data presented here—that relationship satisfaction is largely unaffected by different-race combinations—suggests that these differences are likely to be small, if they exist at all.

A second explanation is that people routinely assume that the divides between races are larger than they actually are. For instance, imagine a young man who would be friends with anyone who shares his favorite interest, without regard to race. He might incorrectly assume that a different-race peer won’t share that interest based on media representation, previous experience, lack of exposure, or some other reason. This young man may not think of himself as racist in any way, but may unintentionally discriminate based on lack of knowledge about out-groups.

Along these lines, this survey focused on explicit measures of bias, but discrimination could also operate with less conscious and more implicit belief patterns. Even though such implicitly biased beliefs are common among people, meta-analyses of years of data on implicit racial and ethnic bias show small effects on actual behavior, and effects do not generalize to real-world decisionmaking.¹⁰¹

An alternative explanation for the persistence of same-race affinity has to do with the structural dimensions of interracial contact and opportunities for cooperation. For much of the 20th century, institutions were set up to limit interracial cooperation.¹⁰² People can only build relationships where opportunity exists. Residential segregation, racial clustering across schools, and differences in wealth and social class all shape the probability of contact and the likelihood of forging significant relationships. Even when individuals express race-neutral preferences, unequal housing, educational, and career opportunities or outcomes may limit exposure based on historic patterns.

Toward a more cooperative future

Whether racial cooperation continues to deepen will depend on more than just individual goodwill, given structural barriers to exposure. Leaders can scrutinize their own practices with a view toward ensuring that they are not indirectly and unnecessarily hindering interracial contact. Positive efforts to make neighborhoods, organizations, schools, and workplaces attractive to all groups can foster additional opportunities for interracial exposure and cooperation.

Ultimately, the findings suggest that interracial cooperation is not only possible, but already unfolding in everyday life—albeit unevenly. Each interaction, friendship, and family that forms across racial or ethnic groups chips away at the legacy of separation that has long shaped the American experience. Given the demographic trajectory of the country, results from this survey offer hope that interracial relationships and diverse groups of work colleagues, friends, and families will increasingly shape the future of American life.