Relation Between Racial Discrimination, Social Class, and Health Among Ethnic Minority Groups

Abstract

Objectives. This study explored associations between racism, social class, and health among ethnic minority people in England and Wales.

Methods. We conducted a series of regression analyses on cross-sectional data from the Fourth National Survey of Ethnic Minorities to explore the relation between different indicators of racism and health and household occupational class.

Results. Marked independent associations existed between reported experience of racism and perceptions of Britain as a “racist society,” household social class, age, sex, and various mental and physical health indicators. These associations showed reasonable consistency across the different ethnic groups.

Conclusions. The different ways in which racism may manifest itself (as interpersonal violence, institutional discrimination, or socioeconomic disadvantage) all have independent detrimental effects on health, regardless of the health indicator used.

In the opening sentence of a paper discussing the mechanisms that may produce ethnic inequalities in health, Cooper stated that, unlike the consequences of being in a disadvantaged social position, “Lynching makes no contribution to the excess mortality among blacks in the United States today.”1^(p137) We would argue, however, that the contemporary equivalent of lynching—racially motivated verbal and physical attacks—may have an important effect on the health experience of ethnic minority groups in industrialized countries. The way in which this has been ignored in research in general, and health research in particular, means that an important element of social disadvantage has been inadequately explored.

Genetic explanations for ethnic differentials in social position and health persist,2 despite a considerable lack of evidence and more than 100 years of research evidence exposing the limitations of such assumptions. On the whole, genetic or cultural factors are generally alluded to after other potential confounders have been controlled for, rather than themselves being the focus of explicit investigation.3,4 Such explanations therefore assume that all other confounders have been both recognized and accurately accounted for, such that the remaining unexplained component of ethnic difference can only be attributable to unmeasured “innate” (i.e., cultural or genetic) characteristics.

There are 2 potential problems with this assumption. First, the measurement of these potential confounders may be inaccurate. For example, recent research suggests that current measures of socioeconomic status are too insensitive for the exploration of ethnic variations in social position. Thus, although different measures consistently show that people from ethnic minority groups experience socioeconomic disadvantage compared with the ethnic majority, the full extent of this disadvantage cannot be easily captured by traditional measures.5,6 The extent to which any residual effect in a statistical model can be assigned to other (particularly unmeasured) factors, when social position is incompletely measured, is questionable.

Second, aspects of the relation between ethnicity and health have been generally ignored. Nazroo5 suggested that 3 aspects of the structural context of ethnicity are unaccounted for in current research: (1) the effect of the accumulation of disadvantage over the life course; (2) the role of ecological effects, produced by the concentration of ethnic minority groups in deprived residential areas; and (3) what we discuss here, the effects of living in a racist society.

Discrimination can vary in form, depending on how it is expressed, by whom, and against whom, and can occur in all aspects of life. Discrimination can be divided into 2 main, but not mutually exclusive, types: interpersonal and institutional. Interpersonal discrimination refers to discriminatory interactions between individuals, which usually can be directly perceived. The little research undertaken in Britain and the United States has suggested that the experience of interpersonal discrimination among people from ethnic minority groups is widespread. A study investigating the experiences of people from ethnic minority groups from 4 areas of the United Kingdom found that, for many, the experience of interpersonal racism was part of everyday life, and being made to feel different was largely seen as routine and even expected.7 Eighty percent of the respondents to a US study reported having experienced racial discrimination at some time in their lives.8 About a third of the respondents to a UK-based study said that the way they led their lives was constrained by the fear of being racially harassed.9

Interpersonal discrimination has been shown to be associated with health. A US study found that differences in rates of hypertension between Black and White respondents were substantially reduced by taking into account reported experiences of and responses to racial harassment.10 Also in the United States, self-reported experience of interpersonal racism has been shown to be associated with raised blood pressure; increased psychological distress, depression, and stress; poorer self-rated health; and more reported days spent unwell in bed.8,10–12 It also has been shown to be associated with increased prevalence of cigarette smoking and low birthweight among the children of those discriminated against.13,14

Institutional discrimination typically refers to discriminatory policies or practices embedded in organizational structures; therefore, it tends to be more invisible than interpersonal discrimination. Research has repeatedly shown that people from ethnic minority groups have lower incomes and are concentrated in environmentally and economically poorer geographic areas, in poorer quality and more overcrowded accommodations, in less desirable occupations, and in longer periods of unemployment than their ethnic majority counterparts.5,15 However, although the relation between social position and health is widely documented,16–18 little evidence is available on the extent to which this ethnic concentration of disadvantage is a product of institutional racism.

Just as the experience of discrimination may vary, so may its health effect; for example, the physical and mental health consequences of a racially motivated assault may be quite different from those associated with the usually more subtle racism experienced by people in their daily lives, what Essed calls “everyday discrimination.”19^(p45) And such racist verbal or physical attacks may have a more direct effect on health than that produced by the way racism leads to the socioeconomic disadvantage of ethnic minority groups. Other forms of racism also may influence the relation between ethnic minority status and health20: for example, disparities in access to health services21,22 or the targeting of ethnic minority communities by the tobacco industry.23–25

The sociodemographic characteristics of the victim, such as age, sex, and social position, also may influence how racism affects health. This may be a result of variations in incidence or type,7,26 or there may be variations in responses to racism. Research suggests, for example, that the perception or reporting of discrimination may be associated with sex (with women reportedly more likely than men to internalize their experiences by accepting their subordinate status and unfair treatment as in some way “deserved” and therefore not explicitly recognize or subsequently report them as being discriminatory27), social class (with more underreporting occurring among those with fewer socioeconomic resources12,28), or particular historical cohorts (with those coming of age during or after the civil rights and women's movements of the 1960s more likely to identify discrimination than older cohorts19,29). Also, people may experience multiple forms of discrimination, which cannot “simply be reduced to the ‘sum’ of each type.”12^(p41)

One problem with establishing the connection between experience of racial discrimination and health is the difficulty associated with measuring the extent to which individuals experience racism, a problem discussed in more detail in the Discussion. Several studies suggest that the way in which someone responds to the discrimination he or she experiences and, consequently, how he or she reports it will influence its health effect.8,10,11,28,30 Studies have shown that Black Americans who said that they would report and challenge racism had lower blood pressure than did those who said that they would tolerate racism and not report their experience.8,10 These authors suggest that this negative health effect is a result of the internalization of the experience among those who would tolerate racism. The existence of social networks to allow a victim to recognize and discuss experiences of racism with others also may mediate the relation between racism and health.

Findings suggest that racism may have severe health consequences, but so far this has not been explored in any detail, and hardly at all in the United Kingdom. We used multivariate analysis to explore the relation between racism, social position, and health among ethnic minority groups, using a range of health outcomes and data from England and Wales.

METHODS

The ethnic minority groups that tend to be the focus of research in Britain are largely made up of postwar labor migrants and their families and descendants from the Caribbean and Indian subcontinent (which includes those who were expelled from East Africa in the late 1960s and early 1970s). In 1993 and 1994, the Fourth National Survey of Ethnic Minorities was undertaken by the Policy Studies Institute and Social and Community Planning Research (now the National Centre for Social Research) to examine the circumstances of these groups. The Fourth National Survey of Ethnic Minorities contained a nationally representative sample of 5196 people of Caribbean, South Asian (Indian, Pakistani, and Bangladeshi), and Chinese origin who were interviewed in detail, together with a comparison sample of 2867 White people.

The sampling procedures were designed to select probability samples of both individuals and households. Sampling points were identified with information from the 1991 British census, which allowed areas to be selected on the basis of the concentration of ethnic minority people within them. White respondents were identified with a straightforward stratified sampling process, in which areas, then addresses, and then individuals within addresses were identified to be included in the study. Screening for ethnic minority respondents was carried out with focused enumeration, with recruiters identifying households containing ethnic minority people by visiting, for example, every sixth address in a defined area and asking about the ethnic origin of those living at both the visited address and the 5 addresses on each side.

In addition to physical and mental health, the questionnaire covered a comprehensive range of information on both ethnicity and other aspects of the lives of ethnic minority persons, including demographic and socioeconomic factors. (For further details on the methods, the demographic details of the sample, and other findings, see Modood et al.31) Respondents were assigned to an ethnic group on the basis of their family origins, a measure that had close correlation with a question very similar to that used in the 1991 British census.32 This analysis involved only those defined as being from an ethnic minority group.

This analysis used 2 indicators of racism to explore the different ways it may affect health (Table 1 ▶). The first question asked respondents about experiences of interpersonal racism within the last year, coded as no experience, experience of racist verbal abuse, and experience of an attack on the respondent or his or her property that was perceived by the respondent to be racially motivated. The second question asked respondents what proportion of British employers they believed would refuse someone employment on the grounds of race, color, religion, or cultural background. Full details of the questions included in the questionnaire can be found in Smith and Prior33 or at http://qb.soc.surrey.ac.uk/surveys/nsem/nsem94.htm.

TABLE 1.

—Perceived and Experienced Racial Harassment

In the last 12 months, have you been attacked for reasons to do with race or color?
No	87% (4523)
Experience of verbal abuse	10% (518)
Experience of a physical attack on the person or the destruction or vandalism of his or her property	3% (156)
Do you think there are employers in Britain who would refuse a job to a person because of his or her race, color, religion, or cultural background? If so, do you think that this is true of most employers, about half, fewer than half, or hardly any?
No/hardly any	31% (800)
Fewer than half	33% (856)
About half	21% (540)
Most	16% (419)

Source. Modood et al.31

Because of the length of the questionnaire, some of the questions were asked of only half the sample. This was the case regarding the question asking respondents what proportion of British employers they believed would discriminate against someone and regarding some of the health indicators. The question exploring employer discrimination was unavailable for 2 of the health indicators used here (respiratory illness and estimated weekly prevalence of depression) because the questions were asked of opposite halves of the sample.

We performed a series of regression tests to explore the relation between experienced or perceived racism and several different indicators of health to determine whether there was any variation by condition. Other variables included were household social class, sex, and age (entered as a continuous variable). Household social class was assigned on the basis of the head of the household's occupation. When there was more than 1 working adult, class was assigned on the basis of sex and age (i.e., men's over women's and father's over son's). It was divided into nonmanual- and manual-headed households and households with no full-time worker.

Details of the health outcomes included in the analysis are shown in Table 2 ▶. The physical conditions explored in the analysis were self-assessed fair or poor health, long-standing illness that limits work, diagnosed high blood pressure, diagnosed diabetes, possible ischemic heart disease (severe chest pain, diagnosed heart attack, or diagnosed angina), and respiratory illness (wheezing, shortness of breath, or bringing up phlegm). In general, these questions were the same as those used in the Health Survey for England.34 The mental health outcomes used were estimated weekly prevalence of depression and annual prevalence of psychosis,35 which were calculated with the revised Clinical Interview Schedule36 and the Psychosis Screening Questionnaire37; responses to these instruments were validated with the Present State Examination.38 (For further exploration of these indicators, see Nazroo.32,35) Preliminary analysis showed great similarity between Pakistani and Bangladeshi people in terms of health outcomes and sociodemographic profiles.31,32 To overcome the limitations of small sample sizes, these groups were combined for this analysis. Because the associations were similar across the different ethnic minority groups for the findings reported here, we elected to maximize statistical power by combining all of the ethnic minority groups for health outcomes other than self-assessed fair or poor health.

TABLE 2.

—Health Outcomes: All Ethnic Minority Groups Combined

	%	n	N
Self-assessed fair or poor health (fair/poor/very poor combined)	34	1780	5182
Do you have any long-standing illness, disability, or infirmity that limits the kind of paid work that you can do?	13	680	5156
Have you ever had high blood pressure?	11	589	5173
Have you ever had diabetes?	6	315	5185
Have you ever had severe chest pain or experienced a heart attack or been diagnosed with angina?	16	326	1989
Have you had attacks of wheezing and/or shortness of breath in the last 12 months? Do you usually bring up phlegm for as much as 3 months each year?	16	426	2609
Estimated weekly prevalence of depression	4	104	2579
Estimated annual prevalence of psychosis	1	43	5196

Note. “n” = number of people responding “yes” when asked whether they had the given condition. N = total number of people asked.

RESULTS

As reported by Virdee,26 3% of the respondents said that they believed that they or their property had been physically attacked for reasons to do with their ethnicity in the past year, and 12% of the respondents reported experiencing racially motivated verbal abuse in the past year (with 10% reporting at least 1 verbal, but no physical, racist attack). Of the respondents, 64% believed that some British employers would refuse someone a job on the grounds of race, color, religion, or cultural background, whereas 37% believed that at least half of British employers would do so.31

Table 3 ▶ shows a logistic regression model with self-assessed fair or poor health as the outcome and experienced and perceived racism, social class, age, and sex as the independent variables for each ethnic minority group separately and for all ethnic minority groups combined. The findings suggested that after including the other variables in the model, statistically significant associations between self-assessed fair or poor health, the experience or perception of racism, social class, age, and sex were found for the combined ethnic minority groups. Respondents who had experienced verbal abuse were approximately 50% more likely than those who reported no racially motivated attack to describe their health as fair or poor, and those who reported being physically attacked or having their property vandalized were over 100% more likely. Those who believed that the majority of British employers are racist were approximately 40% more likely to describe their health as fair or poor, compared with those who believed that this characterization applied to fewer than half of employers, if any.

TABLE 3.

—Odds Ratios (ORs) From Regression Analysis of Self-Assessed Fair or Poor Health, by Social Class and Experience and Perceptions of Racism

	Ethnic Group
	Caribbean (n = 582), OR (95% CI)	Indian (n = 973), OR (95% CI)	Pakistani and Bangladeshi (n = 848), OR (95% CI)	Chinese (n = 104), OR (95% CI)	Total (n = 2507), OR (95% CI)
Experience of racial harassment
No attack	1.00	1.00	1.00	1.00	1.00
Verbal abuse	1.29 (0.70, 2.39)	1.94 (1.00, 3.74)	2.17 (1.06, 4.45)	0.62 (0.12, 3.12)	1.54 (1.07, 2.21)
Physical attack	1.02 (0.27, 3.86)	3.88 (1.39, 10.86)	2.53 (0.98, 6.52)	1.23 (0.14, 11.00)	2.07 (1.14, 3.76)
Perception of discrimination
Fewer than half of employers racist	1.00	1.00	1.00	1.00	1.00
Most employers racist	1.37 (0.90, 2.10)	1.20 (0.81, 1.76)	1.76 (1.11, 2.78)	1.18 (0.34, 4.10)	1.39 (1.10, 1.76)
Social class
Nonmanual	1.00	1.00	1.00	1.00	1.00
Manual	0.96 (0.54, 1.72)	1.86 (1.20, 2.90)	1.43 (0.78, 2.63)	0.82 (0.18, 3.71)	1.44 (1.07, 1.94)
No full-time worker	1.41 (0.82, 2.44)	2.58 (1.63, 4.09)	2.84 (1.62, 4.98)	1.69 (0.49, 5.82)	2.42 (1.82, 3.22)
Age	0.98 (0.88, 1.09)	1.08 (1.01, 1.16)	1.12 (1.04, 1.21)	0.93 (0.73, 1.18)	1.05 (1.00, 1.10)
Sex
Male	1.00	1.00	1.00	1.00	1.00
Female	1.62 (1.02, 2.58)	1.62 (1.13, 2.32)	1.85 (1.24, 2.76)	1.13 (0.43, 2.98)	1.61 (1.28, 2.01)

Note. CI = confidence interval.

Respondents from households classified as manual were approximately 40% more likely than those from nonmanual households to describe their health as fair or poor, and those from households with no full-time worker were approximately 150% more likely. Analyses not presented here suggested that no interaction occurred between reported experience of racism and household social class in predicting self-assessed fair or poor health. Being female was associated with a 60% greater likelihood of reporting fair or poor health, compared with being male.

When models were calculated for particular ethnic groups, findings were similar for the 2 South Asian groups, although there was some variation for the Chinese and Caribbean groups. Significant associations also emerged from the logistic regression models that used the other health indicators, associations that followed patterns very similar to those for self-assessed fair or poor health (Table 4 ▶). For example, there was an 85% increase in risk of respiratory illness and a 150% increase in estimated rates of psychosis and depression among those who experienced verbal abuse, compared with those reporting no experience of racism. Reporting experience of racially motivated assault or property damage was associated with between 3 and 5 times the risk associated with reporting no experience of racism for these indicators. Among respondents with hypertension and for measures of estimated rates of psychosis, perceiving more than half of British employers to be racist was associated with an increased risk of around 60%, in addition to any change in risk associated with reporting actual experience of racism. Compared with living in a nonmanual household, living in a manual-headed household and, particularly, living in a household with no full-time worker was associated with an increased risk of ill health for almost all of the indicators explored. There was also the expected increased risk of ill health associated with increasing age and being female.

TABLE 4.

—Odds Ratios (ORs) From Regression Analysis of Health Outcomes, by Social Class and Experience and Perceptions of Racism: All Ethnic Minority Groups Combined

	Experience of Racial Harassmenta		Perception of Discriminationb	Social Classc
Health Outcome	Verbal Abuse, OR (95% CI)	Physical Attack, OR (95% CI)	Most Employers Racist, OR (95% CI)	Manual, OR (95% CI)	No Full-Time Worker, OR (95% CI)	Age, OR (95% CI)	Female Sex, OR (95% CI)d
Self-assessed poor health	1.54 (1.07, 2.21)	2.07 (1.15, 3.76)	1.39 (1.10, 1.76)	1.44 (1.07, 1.94)	2.42 (1.82, 3.21)	1.05 (1.00, 1.10)	1.61 (1.28, 2.01)
Long-standing limiting illness that limits work	1.82 (1.10, 3.01)	1.66 (0.62, 4.44)	0.90 (0.66, 1.24)	2.04 (1.31, 3.18)	3.80 (2.50, 5.78)	1.18 (1.11, 1.25)	1.36 (1.01, 1.85)
High blood pressure	1.45 (0.86, 2.42)	1.71 (0.65, 4.46)	1.66 (1.19, 2.32)	1.71 (1.07, 2.75)	2.25 (1.42, 3.57)	1.18 (1.09, 1.28)	1.24 (0.89, 1.73)
Diabetes	1.52 (0.73, 3.13)	0.97 (0.31, 2.98)	1.12 (0.73, 1.73)	1.37 (0.72, 2.60)	1.76 (0.97, 3.19)	1.40 (1.28, 1.53)	1.03 (0.68, 1.56)
Myocardial infarction risk	1.69 (0.79, 3.60)	1.61 (0.47, 5.58)	0.95 (0.60, 1.49)	1.58 (0.80, 3.10)	2.51 (1.40, 4.50)	1.32 (1.13, 1.55)	1.05 (0.69, 1.60)
Respiratory illness	1.85 (1.20, 2.84)	3.64 (1.83, 7.27)	. . .	1.35 (0.95, 1.93)	1.86 (1.34, 2.59)	0.96 (0.92, 1.00)	1.05 (0.80, 1.37)
Estimated weekly prevalence of depression	2.45 (1.55, 3.88)	2.89 (1.19, 7.03)	. . .	1.27 (0.82, 1.97)	2.16 (1.45, 3.22)	1.07 (1.02, 1.13)	1.26 (0.91, 1.75)
Estimated annual prevalence of psychosis	2.86 (1.69, 4.83)	4.77 (2.32, 9.80)	1.57 (1.02, 2.42)	0.80 (0.45, 1.45)	1.60 (0.94, 2.75)	0.99 (0.94, 1.05)	1.77 (1.16, 2.72)

Note. CI = confidence interval.

^aNo attack = 1.

^bLess than half of employers are racist = 1.

^cNonmanual = 1.

^dMale = 1.

DISCUSSION

This analysis suggests that the experience of interpersonal racism and perceptions of racism in wider society both have independent negative health consequences. These findings support earlier research that found a positive association between racism and ill health.8,10,12 They also would seem to be supported by other studies suggesting a positive association between experiences of assault in general and heart disease39 and psychological distress.40 But the small number of quite varied studies that have been undertaken to explore the relation between racism and health make it difficult to make precise comparisons. In addition, studies have shown that socioeconomic disadvantage contributes heavily to ethnic inequalities in health.5,15,31,41–43 Our findings suggested that such an effect is independent of the direct experience of racism.

In regard to interpreting these findings, some issues should be borne in mind. First, only 1 in 8 respondents to the Fourth National Survey of Ethnic Minorities reported experiencing some interpersonal racism, and this would seem to be an underestimation, compared with other studies. One reason for this may be that this question was restricted to experiences occurring only in the last year: 80% of the respondents in Krieger's study reported having experienced racial discrimination at some time.8,10 Or, as we suggested above, there may be reasons that people may not perceive the discrimination they experience. A British-based study, for example, found that people who initially stated on a questionnaire that they had not experienced racial discrimination later, during an in-depth interview, said that they had experienced such discrimination but found it too difficult to discuss.44 One problem with measuring experience of racial discrimination is that people's interpretations also will vary: whether an experience is seen to be a function of an individual's social category or something else will be a consequence of his or her own history of intergroup interactions, as well as a response to the “objective” experience.

A US study suggested that there may have been a change in the nature of racial prejudice over time, such that experiences of racism may be more difficult to recognize today.45 Dovidio and Gaertner described the rise of “aversive racism,” characterized by people who “endorse egalitarian values, who regard themselves as non-prejudiced, but who discriminate in subtle rationalizable ways.”45^(p315) Thus, in addition to more overt, traditional forms, discrimination may be expressed in indirect and rationalizable ways that, while potentially having a similar health effect, will be more difficult to measure. As Cooper put it: “The lynch mob was an effective instrument of social policy in its day, but too clumsy for a time when appearances count for more than reality.”1^(p137)The rise of aversive racism, Dovidio and Gaertner45 argued, has led to a decline in self-reported experiences of discrimination.

Research has repeatedly shown that people report perceiving greater discrimination directed toward their group as a whole than toward themselves personally as members of that group; this has been called the personal/group discrimination discrepancy.30 That the individual may consciously not wish to discuss, or simply not recognize, the discrimination he or she experiences is one possible explanation for this phenomenon. Alternatively, it may result from unconscious reactions to personal experiences of discrimination.

Ruggiero and Taylor28 described several theories that suggest that effective coping is achieved through an internal sense of control over one's experiences, maintenance of which requires minimizing the role of external forces of control on them, which leads to the denial of influences such as discrimination but also may limit their negative health effect. However, other studies have suggested that the health effects of such internalization may vary; although it may have self-protective qualities under some circumstances, it has been shown to be related to hypertension.8,10,11 In relation to the analysis presented here, the effects of any such underreporting are unclear. It is possible that general underreporting of experiences of racism would introduce a conservative bias and so would dilute the observed differences between those with and without experience of racism. However, disproportionate underreporting by specific social or health groups might magnify the observed differences.

We suggest that some of any such underreporting should be picked up by our indicator exploring perceptions of discrimination in wider society. Preliminary findings from the Fourth National Survey of Ethnic Minorities31 suggested that, in keeping with the personal/group discrimination discrepancy theory described earlier,30 perceptions of Britain as a “racist society” (determined, here, as responses to the question exploring whether respondents believed that British employers would racially discriminate against someone) are more widely reported than actual experiences of interpersonal discrimination. However, the discrepancy between the 2 indicators of racism used here also may exist because this first question referred to experiences within the last year, whereas a sense of institutional or societal racism is likely to be developed over a longer period, in response to repeated institutional and interpersonal experiences of racism. Alternatively, our indicator of societal racism may explore a “sense” of being a victim of discrimination, which may not develop from direct, reportable experiences. Our findings suggest that both have negative consequences for health.

Another issue to be borne in mind is that the small sample sizes made it necessary to combine heterogeneous ethnic groups. Although the analysis for each ethnic group separately suggested similarity in associations between racism and health across the different ethnic minority groups, there may have been some variation for the Chinese and Caribbean groups, which could not be explored because of small sample sizes.

Unmeasured factors related both to people's reports of racism and to their selfreported health could explain our findings, particularly because our measure of household occupational status as a surrogate for socioeconomic position may appear crude. However, we repeated this analysis with different indicators of social class (data not shown) and found that the effects were consistent.

Also, earlier analysis of this data set suggested that traditional measures of social class (the Registrar General's occupational measure of class,46 for example) were problematic for analyses drawing comparisons across different ethnic groups because the internal heterogeneity of such traditional class groupings masked the concentration of ethnic minority people in lower-income occupations, poorer-quality housing, and longer periods of unemployment compared with White people in the same class.32 However, this was not found to be a problem for analyses within specific groups, as conducted here.32 Although this earlier analysis showed that self-reported general health varies by social position (such that much of the variation in health between different ethnic groups may be explained by differences in socioeconomic position32), it also suggested that no relation existed between reported experience of racism and socioeconomic position.26 Of course, reported racism and reported health also may be affected by several other potential confounders, such as migration status, ethnic identity, household structure, and the availability of social support. The inclusion of such factors is beyond the scope of this paper, but we have explored the relation between ethnic identity and health elsewhere.47,48

Regardless of these issues, reported experience of racism would appear to be strongly associated with a wide variety of health indicators, independent of age, sex, and those aspects of socioeconomic position measured by household occupational class. On the surface, the consistently higher odds ratios for both verbal and physical harassment do not appear to support Krieger's hypothesis that respondents reporting no harassment are internalizing or denying their experience and are therefore more likely to have poorer health, at least in terms of hypertension, than those reporting some harassment.8,10 One reason for this result may be that the 2 sets of findings may not be directly comparable: respondents to Krieger's study described how they believed they would respond in a fictitious situation (which may not correspond to how they would respond in actuality), whereas in the Fourth National Survey of Ethnic Minorities, respondents were reporting actual experiences or the absence of such experiences.

That there is some underreporting in our indicator of experienced interpersonal racism may be one explanation for the independent health effect shown by the indicator exploring perceptions of discrimination in wider society. Alternatively, there may be direct effects. For example, the belief that you are living in a society that will discriminate against you on the grounds of your ethnicity may itself be detrimental in terms of health. Exposure, susceptibility, and responses to socially inflicted trauma, of a mental or physical nature, is one of the “five pathways of embodying discrimination” described by Krieger.12^(p42) However, the course of the pathway between socially inflicted trauma and health, beyond any immediate physical injury, is little understood. In general, it has been suggested that long-term exposure to inferior treatment and a devalued status is damaging to self-esteem, invalidates self-worth, and may block aspirations.12,49 Such exposure may shape the content and frequency of stressful life events and may limit the range of feasible responses to them, as well as the social support available. All of these will have severe consequences for health.

More specifically, the discriminatory act may produce a sense of threat within the victim that may cause various reactions, including fear, distress, anger, and denial. These reactions could produce a physiological response (be it cardiovascular, endocrine, neurological, immunological) that subsequently affects health.12

Several studies have described the reactions of those living with the threat of discrimination. It is also important to acknowledge here that racism, unlike other criminal acts, need not be personal to produce a threat, because it is often viewed as “an attack on the community as a whole.”26^(p284) This may be seen in findings that suggest that those living with the threat of, or in fear of, racism are more numerous than those reporting actual personal experience of racism.9,26 Thus, the nature of the response to the racist incident, as well as producing potential measurement artifacts, may determine the health consequences of the experience itself. For example, people living in a climate of fear and insecurity may adapt by constraining their lives to avoid vulnerable situations, a response that may lead to stress,9,26,49 and those who explicitly recognize the racist nature of their experiences have been reported to retain higher levels of self-esteem and self-efficacy, compared with those who internalize or deny them.12

The differential health effect of such responses is perhaps more easily understood if we imagine the mental and physical health consequences for those who resort to the use of psychoactive substances to alleviate their reaction to the racist experience, compared with those who would respond by participating in some form of active resistance or community organization.12 Again, the availability of sources of social support will influence the health effect of racism.

In addition to the health consequences of experienced and perceived interpersonal racism, it is argued that institutional racism leads to the concentration of ethnic minority groups in the conditions of social and economic disadvantage described earlier.5,15,41 Institutional racism promotes the identification of ethnic minority groups, their reification as biologically and culturally different, and their consequent social and economic exclusion,41 such that “racism is the primary reason that Blacks are disproportionately concentrated in the poorest sectors of the working class and face restricted class mobility.”15^(p87) We have shown elsewhere that socioeconomic disadvantage makes a major contribution to ethnic inequalities in health,5,31,42,43 and our findings reported here suggest that socioeconomic effects are independent of experiences of racial harassment and discrimination. This would suggest an important role for institutional racism in the relation between ethnicity and health.

In conclusion, these and earlier findings suggest that racism harms health—in terms of actually experienced attacks, perceived discrimination, and the concentration of ethnic minority groups in lower social classes and in unemployment, regardless of the health indicator used. Reported experienced interpersonal racism and perceived institutional discrimination have each been shown to have independent health effects, which, based on their consistency across a range of quite different health indicators, would seem to persist over and above any immediate physical injury caused by an incident itself. The role of racism in the relation between ethnicity and health has been ignored for too long. Regardless of this health effect, unjustly denying people fair treatment, ignoring their human rights, and constraining their possibilities for living fully expressed and dignified lives are unacceptable. A lack of theoretical and analytic creativity has prevented earlier recognition of the role of racism in the development of ethnic inequalities in health. This paper adds further support to the argument that we can no longer use assumptions about the biological and cultural basis of ethnic inequalities either to limit the search for the underlying mechanisms producing them or to justify our inaction in reducing them.