Discrimination in the United States: Experiences of Latinos

Abstract Objective To examine experiences of racial/ethnic discrimination among Latinos in the United States, which broadly contribute to their poor health outcomes. Data Source and Study Design Data come from a nationally representative, probability‐based telephone survey including 803 Latinos and a comparison group of 902 non‐Hispanic white US adults, conducted January—April 2017. Methods We calculated the percent of Latinos reporting discrimination in several domains, including health care. We used logistic regression to compare the Latino‐white difference in odds of discrimination, and among Latinos only to examine variation by socioeconomic status and country of birth. Principal Findings One in five Latinos (20 percent) reported experiencing discrimination in clinical encounters, while 17 percent avoided seeking health care for themselves or family members due to anticipated discrimination. A notable share of Latinos also reported experiencing discrimination with employment (33 percent applying for jobs; 32 percent obtaining equal pay/promotions), housing (31 percent), and police interactions (27 percent). In adjusted models, Latinos had significantly higher odds than whites for reporting discrimination in health care visits (OR: 3.18, 95% CI: 1.61, 6.26) and across several other domains. Latinos with college degrees had significantly higher odds of reporting discrimination in multiple domains than those without college degrees, with few differences between foreign‐born and US‐born Latinos. Conclusions Latinos in the United States report experiencing widespread discrimination in health care and other areas of their lives, at significantly higher levels than whites. Being born in the United States and earning a college degree are not protective against discrimination, suggesting that further health and social policy efforts to eliminate discrimination are needed.


| INTRODUC TI ON
In recent years, there has been a growing national debate about the seriousness of racial and ethnic discrimination in the United States. 1 Despite this increased national attention, institutional discrimination against Latinos-the largest minority group in the United States 2 -has not been studied extensively with nationally representative data, nor across public policy domains. Because discrimination is associated with a wide range of adverse health outcomes and is therefore a prominent risk factor for Latinos' health, [3][4][5][6][7] understanding the extent of discrimination against Latinos across different areas of their lives may partially explain observed variations in their health, particularly along relevant demographic factors affecting health.
Previous research suggests that nationality and socioeconomic status moderate the relationship between race/ethnicity and discrimination, but their effects depend on the racial or ethnic group of inquiry (eg, whites with higher education and income are less likely to report experiencing discrimination, while blacks and Asians with higher education and income are more likely to report discrimination). [8][9][10][11][12][13][14] Additionally, 48 percent of Latino adults in the United States are foreign-born, 15 and previous studies suggest Latino groups have cultural differences related to ethnic identity, nativity, accent, and language that may be important correlates of discrimination. 13,16,17 To date, the vast majority of existing studies on discrimination against Latinos have either relied on samples that cannot be generalized to the US Latino population (eg, small sample sizes, convenience samples, only Mexican Americans, only Latino immigrants), 8,17,18 or have focused on everyday discrimination (unfair treatment) 13 or discrimination within a single policy domain (such as health care) 11,17 and have not simultaneously examined the extent of discrimination Latinos face across multiple institutions or policy domains.
This study brings a public health perspective to the complexity and pervasiveness of discrimination in the United States today, alongside complementary articles in this issue of Health Services Research. It was conducted as part of a larger nationally representative survey fielded in 2017 in response to a growing national debate about discrimination in the United States today, 1,19 to understand experiences of discrimination against several different groups in America including blacks, Latinos, Asian Americans, Native Americans, women, and LGBTQ people. Specifically, the purpose of this study was threefold: (1) to document the prevalence of ethnic discrimination against Latinos adults across multiple institutional domains, including health care, education, employment, housing, political participation, police, and the criminal justice system, as well as interpersonal domains that affect health outcomes, including slurs, microaggressions, harassment, and violence; (2) to document disparities by comparing Latinos' experiences to whites; and (3) to examine the variation in Latinos' experiences of discrimination by socioeconomic status (income and education) and country of birth (US-born vs foreign-born).

| Study design and sample
Data were obtained from an original, nationally representative, probability-based telephone (cell and landline) survey of US adults, conducted from January 26 to April 9, 2017. The survey was jointly designed by Harvard TH Chan School of Public Health, the Robert Wood Johnson Foundation, and National Public Radio. SSRS administered the survey. Because Harvard researchers were not directly involved in data collection and de-identified datasets were used for analysis, the study was determined to be "not human subjects research" by the Harvard TH Chan School of Public Health Office of Human Research Administration.
The full sample included 3453 US adults aged 18 years and older, and this paper examines the subsample of 803 Latinos or Hispanics and 902 non-Hispanic whites. Throughout the paper, we use shorthand descriptors of Latino or white. To identify Latinos and whites, screening questions regarding both racial and ethnic identity were asked at the beginning of the survey, and all questions about racial/ethnic identity were based on respondents' self-identification. If respondents identified as Latino and another race, interviewers asked whether they identified more with being Hispanic/Latino (coded as Latino) or more with the other race (coded as the other race). In all follow-up questions for Latino respondents, question wording used the term "Latino," following language most commonly used. This method of screening also allowed interviewers to use the appropriate language in survey questions to describe or refer to the respondent's own identity.
For example, this allowed questions to be read as "Did you experience [form of discrimination] because you are ['Latino']?" rather than "because of your race or ethnicity?" Interviewers were trained to identify Spanish-speaking only respondents, and adults who preferred being interviewed in Spanish were interviewed by bilingual interviewers. Respondents were allowed to switch between Spanish and English according to their comfort level, and 38 percent of Latino surveys were conducted in Spanish (weighted percent).
The completion rate for this survey was 74 percent among respondents who answered initial demographic screening questions, with a 10 percent overall response rate, calculated based on the American Association for Public Opinion Research's (AAPOR) RR3 formula. 20 Because data from this study were drawn from a probability sample and used the best available sampling and weighting practices in polling methods (eg, 68 percent of interviews were conducted by cell phone, and 32 percent were conducted via landline), they are expected to provide accurate results consistent with surveys with higher-response rates 21,22 and are therefore reliably generalizable to the broader population, within a margin of error of ±4.5 percentage points at the 95% confidence interval. See Benson et al (2019) 23 for a further description of the survey methodology.

| Survey instrument
The poll asked about adults' experiences of discrimination. We conceptualized racial/ethnic discrimination as differential or unfair treatment of individuals based on self-identified race or ethnicity, whether by individuals (based on beliefs, words, and behavior) or social institutions (based on laws, policies, institutions, and related behavior of individuals who work in or control these laws, policies, or institution). 4,8,24 We analyzed 18 questions from the survey, covering six institutional and six interpersonal areas of discrimination (question wording in Appendix S1). Institutional areas included were health care; employment; education; housing; political participation; and police and courts. Interpersonal areas included were racial/ethnic slurs; microaggressions (ie, negative assumptions or insensitive or offensive comments about you); racial/ethnic fear; sexual harassment; being threatened or nonsexually harassed; and violence. We also examined two areas in which concerns about discrimination might prevent adults from taking needed action: seeking health services and protection from the police. We examined discrimination in domains previously demonstrated to be associated with health, as well as some that were not (eg, political participation), in order to capture a wide range of possible discriminatory experiences across adults' lives. We also examined general beliefs about the existence of discrimination against one's own racial/ethnic group (Latinos or whites) in America today. Questions about experiences were only asked among a random half sample of respondents to maximize the number of questions while limiting respondent burden. Questions were only asked of relevant subgroups (eg, college questions only asked among adults who had ever applied to college). Questions on harassment, violence, and avoiding institutions for fear of discrimination were asked about yourself or family members because of the sensitive nature of the topic and prior literature demonstrating that vicarious experiences of stress (eg, through discrimination experienced by family members) can adversely affect individuals. 25

| Statistical analyses
After calculating descriptive statistics, we calculated the prevalence of all Latinos and whites who reported that they had ever experienced racial (white) or ethnic (Latino) discrimination in each of the domains. Using pairwise t tests of differences in proportions, we made uncontrolled comparisons of the percentage of Latino and white adults reporting discrimination across domains. For all analyses, statistical significance was determined at P < .05.
We then conducted logistic regression models to assess whether reporting discrimination remained significantly associated with race/ ethnicity after controlling for the following variables that may be related to variation in experiences of discrimination 8,9,11,13,26,27 : gender, age (18-29, 30-49, 50-64, 65+), household income (<$25 000, $25 000-<$50 000, $50 000-<$75 000, $75 000+), education (less than college degree or college graduate), current health insurance status (used only for the health care question-uninsured, Medicaid insured, non-Medicaid insured), neighborhood racial composition (measured as whether respondents live in a neighborhood that is predominantly their own race/ethnicity or not), metropolitan status (urban, suburban, rural), and region (US Census Bureau 4-region division: Midwest, Northeast, South, West). Among Latino adults only, we estimated logistic regression models to examine variation in experiences of institutional discrimination by socioeconomic status (education and income) and country of birth, while controlling for gender, age, health insurance (for health care outcomes only), neighborhood racial/ethnic composition, and geographic measures. To test the sensitivity of our results to model specifications, we fit alternate models using different measures of discrimination, income, and education. We also tested models interacting age with income, but models showed insignificant results, most likely due to small sample size, and were ultimately dropped from analysis. In addition, among Latino adults only, we ran an ordinal logistic regression model to test characteristics associated with reporting overall institutional discrimination. Because questions were only asked among a randomized half sample of respondents, the scale of this model ran from 0 (no reported discrimination in any domains) to 7 (maximum possible reported discrimination in 7 different institutional domains).
To compensate for known biases in telephone surveys (eg, nonresponse bias) and variations in probability of selection within and across households, sample data were weighted by household size and composition, cell phone/landline use, and demographics (gender, age, education, race/ethnicity, and census region) to reflect the true population distribution of Latino and white adults in the country. 28 Other techniques, including random-digit dialing, replicate subsamples, and random selection of a respondent within a household, were used to ensure that the sample is representative. All analyses were conducted using STATA version 15.0 (StataCorp) and all tests accounted for the variance introduced by weighted data.

| RE SULTS
Weighted characteristics of Latinos and non-Hispanic whites in this study sample are presented in Table 1. Latinos differed from whites on almost every demographic measure. Compared to whites, Latinos were younger, less likely to have a college degree (15 percent vs 34 percent, P < .01), and more likely to live in lower-income households (less than $25 000 per year) (39 percent vs 23 percent, P < .01). Latinos were also more likely to lack health insurance than whites (22 percent vs 9 percent uninsured, P < .01), less likely to live in a neighborhood that was predominantly their own race/ethnicity (44 percent vs 67 percent, P < .01), and more likely to live in the Western United States (37 percent vs 18 percent, P < .01). About half of Latinos in this sample (49 percent) were born in the United States (including Puerto Rico). Table 2 shows unadjusted estimates of Latinos and whites reporting personal discrimination because of their race (whites) or ethnicity   Overall, 78 percent of Latinos reported that "generally speaking, [they] believe discrimination against Latinos exists in America today," compared to 55 percent of whites reporting they believe discrimination exists against whites in America today (P < .01). In the context of personally experiencing discrimination, Latinos were significantly more likely than whites to report experiencing discrimination in most institutional and interpersonal areas, including employment (applying for jobs: 33 percent vs 19 percent, P < .01, and obtaining equal pay or being considered for promotions: 32 percent vs 13 percent, P < .01); housing (trying to rent a room/apartment or buy a house: 31 percent were also more likely than whites to report discrimination in trying to vote or participate in politics (15 percent vs 4 percent, P < .01) and report others have acted afraid of them due to their race/ethnicity (15 percent vs 7 percent, P < .01).
After we controlled for potential sociodemographic confounders in logistic regression models (gender, age, education, household income, neighborhood racial composition, metropolitan status, region, and health insurance status where applicable), all Latino-white disparities in reported institutional discrimination persisted except for job applications, while disparities in interpersonal discrimination were no longer statistically significant. Figure 1 shows the adjusted differences in the odds of Latinos personally experiencing discrimination compared to whites. Among Latinos only, there were some differences in odds of reporting discrimination by socioeconomic status (education and income) and country of birth, as shown in Table 3. Latinos with a college degree had significantly higher odds than those without a college degree for reporting discrimination when applying for jobs (3.

| D ISCUSS I ON
Four key findings emerged from this national survey of Latino adults.
First, we found widespread reported discrimination against Latinos.
A majority of Latinos perceived general discrimination against Latinos in America today, while one in five reported experiencing discrimination in their clinical encounters.
Second, we found major differences in discrimination experienced between Latinos and whites. Regardless of socioeconomic status, Latinos reported experiencing discrimination at significantly higher levels than whites in health care and several other social institutions, including in clinical encounters and avoiding seeking health care due to anticipated discrimination.
Third, our results also showed that education was not protective against discrimination for Latinos in any policy domain, and instead was associated with higher levels of discrimination against Latinos in their jobs, police interactions, health care, and college. This relationship was reversed among whites, where higher socioeconomic status was associated with reporting less discrimination. These results are consistent with prior literature showing that high-SES minorities may experience greater discrimination than their lower-SES counterparts. 8,10-12 However, our results do not explain why this might occur. It is unclear whether this relationship is driven by unequal exposures (eg, high-SES minorities having greater contact with whites than low-SES minorities) or differential reporting (eg, high-SES minorities being more likely to recognize and/or self-report unequal treatment than low-SES minorities). 14,29,30 Future research should seek to explore the reasons for these findings among Latinos.
Fourth, we found few differences in discrimination between foreign-born Latinos and US-born Latinos after controlling for other major sociodemographic characteristics, suggesting that Latinos face discrimination, regardless of their nationality and immigration status.
While it is beyond the scope of these results to recommend specific approaches to ending discrimination, these findings update prior studies showing that Latinos continue to face widespread barriers to equal treatment across public institutions and interpersonally. 13,18 Our results add to the literature by showing that Latinos widely believe discrimination against their ethnic group occurs in the United States today, they continue to report experiencing high levels of discrimination personally, and more than one in six have avoided seeking health care and calling the police to avoid experiencing discrimination or unfair treatment.
Discrimination carries major health consequences for Latinos in the United States. Other research has found that prolonged or repeated discrimination causes major health problems over time, due to progressive wear and tear on the body's systems owing to repeated adaption to stressors (known as allostatic load and overload). 31,32 We expect that on average, discrimination will have a greater effect on the health of high-socioeconomic status and upwardly mobile Latinos, who report experiencing greater discrimination than Latinos with lower-socioeconomic status and those who are not upwardly mobile. 8  Housing question only asked among respondents who have ever tried to rent a room or apartment, or to apply for a mortgage or buy a home.
f Ordinal logistic regression model with experiencing discrimination in 0-7 institutional domains as the outcome; individual questions only asked among a randomized half sample of respondents, so the maximum number of times a respondent could report experiencing discrimination in institutional questions was 7. *Significant at P < .05 (shown in bold font).
Bold denotes statistical significance at P < .05.
been rigorously evaluated for their effects on reducing racial/ethnic health disparities.

| Limitations
Our results should be interpreted considering several limitations.
Interviews were conducted in both English and Spanish to overcome some communication challenges. Nonetheless, some respondents may have been excluded from the study due to language barriers, respondents may have interpreted questions differently based on varying backgrounds and expectations, and they may have reported experiences of cultural or communication problems as discrimination. Despite this, previous research has shown that discrimination affects health through multiple pathways and that perceived, self-reported discrimination is associated with worse health outcomes. 4,6,8 We did not ask about the timing or severity of experiences of discrimination. We also did not examine discrimination experienced as a result of respondents' other social identities (eg, LGBTQ-based or gender-based discrimination), or racial/ethnic identities they had in addition to the one they chose as their primary identity, which may compound experiences of discrimination among some adults. Our low response rate is a notable limitation, though evidence suggests that low response rates do not bias results if the survey sample is representative of the study population. 21,22 Recent research has shown that such surveys, when based on probability samples and weighted using US Census parameters, yield accurate estimates in most cases when compared with both objective measures and higher-response surveys. 21,22,38,39 For instance, a recent study showed that across fourteen different demographic and personal characteristics, the average difference between government estimates from high-response rate surveys and a Pew Research Center poll with a response rate similar to this poll was 3 percentage points. 21 However, it is still possible that some selection bias may remain that is related to the experiences being measured. In addition, large confidence intervals in some logistic regression models (eg, discrimination in college applications/attendance) should be cautiously interpreted, as they may indicate low precision in some estimates. We also did not distinguish between the experiences of different Latino heritage groups (eg, Mexicans vs Cubans).
Despite these limitations, this study design allowed us to closely examine reported experiences of ethnic discrimination among Latinos. Our results highlight the extent of discrimination currently experienced by Latinos across public policy areas. Latinos face significant barriers to equal treatment across public institutions and policies, which carries major consequences for their health and well-being.

| CON CLUS IONS
A sizeable share of Latinos currently report experiencing discrimi-