Fear of racism, employment and expected organizational racism: their association with health

Abstract

Background: Racism has been argued to be a focal element of larger societal inequalities which generate ethnic health disparities. Despite suggestions that socio-demographic characteristics of the victim may influence the impact of racism on health, little is known in the United Kingdom about how self-reported experiences of racism vary by socio-demographic characteristics, whether racism contributes to ethnic differences in health and whether there is a differential association between racism and health for certain socio-demographic groups. Methods: Multilevel logistic regression models were conducted using data from the 2005 Citizenship Survey to identify the demographic characteristics associated with reporting experienced racism; explore the association between health, racism and its contribution to ethnic inequalities in health; and explore the moderating role that gender, age, ethnicity and socio-economic position (SEP) have in the relationship between racism and health. Results: Females were significantly more likely to report fear of racial and religious attacks, but reported lower odds of experiencing employment and expected organizational discrimination. A trend was observed for decreasing employment discrimination as SEP decreased. A reverse association was found for SEP and expected organizational discrimination, where people in the lowest employment categories reported lower odds of experiencing discrimination. Conclusion: This study highlights variations in the types of racial discrimination most commonly reported across different socio-demographic characteristics. Despite substantial differences in the experience of racial discrimination, the detrimental impact of racism on health was the same across socio-demographic groups.

Introduction

It is now well established that discrimination and experienced racism harm health.^1–10 Racism has been argued to be a focal element of ethnic inequalities in health,^11–13 impacting on the health of ethnic minority people through differential exposure to socio-economic, environmental, psychosocial and health care-related pathways.^12,14,15 Exposure to these mechanisms has been suggested to vary depending on socio-demographic characteristics of the victim, including age, sex and socio-economic position (SEP).¹¹ However, despite indications that socio-demographic characteristics may influence the exposure to, and effect of experienced racism on health, little is known about how self-reported experiences of racism vary by socio-demographic characteristics,⁷ and whether there is a differential association between racism and health for certain groups. Although possible moderating effects of factors such as SEP, age or gender have been hypothesized, a systematic review has shown findings to be inconsistent in the United States and largely unexplored in the United Kingdom (UK).⁷ An examination of how and whether racism impacts differently across socio-demographic characteristics, and a substantiation of its hypothesized contribution to ethnic health inequalities would provide timely and novel information to the fast developing research area of racism and health, still in its infancy in Europe and the United Kingdom.

This study aims to fill this gap in the UK literature by analysing a nationally representative dataset to: (i) identify the demographic characteristics associated with an increased risk of reporting experienced racism; (ii) explore the association between racism and the health of ethnic minorities, and its contribution to ethnic inequalities in health; and (iii) explore the moderating role that gender, age, ethnicity and SEP have in the relationship between racism and health.

Methods

Study population

This study involves secondary analysis of the 2005 Citizenship Survey (CS), the third in a series of biannual surveys conducted to inform the work of the Communities and Local Government Department (formerly the Department for Communities and Local Government).¹⁶ The survey consists of two separate components: a core representative sample of the general adult population of England and Wales of around 10 000 individuals, and an ethnic minority boost sample of approximately 4000 individuals, which was selected from wards where >1% of the population was from an ethnic minority group. The study sample was selected using a two-stage sampling approach. At the first stage, a random sample of Census Area Statistics (CAS) wards was selected; at the second stage, addresses were sampled within the selected wards. Response rates were 63% for the core sample, and 54 and 62% for the ethnic boost sample (focused enumeration screening and direct screening, respectively). For further details on the sampling methodology and population weighting, see the 2005 CS Technical Report.¹⁷

Measures

Three different dimensions of racism were explored in the present study: fear of racial/religious attacks, expected organizational discrimination, and experienced employment discrimination. Fear of racial/religious attacks was measured by asking the respondent how worried he or she was about being subject to a physical attack because of skin colour, ethnic origin or religion (dichotomized in the analyses as ‘not very worried or not worried at all’ and ‘fairly or very worried’, but fully shown in table 1 for descriptive purposes). Organizational discrimination was analysed as a dichotomous variable that measured whether the respondent felt that he/she would be treated better, worse or the same as other races by members of any of the following organizations (categorized into ‘expects to be treated better or same’, and ‘expects to be treated worse’): a local doctor's surgery, a local hospital, the health service generally, a local school, the education system generally, a council housing department or housing association, a local council, a private landlord, the Courts, the Crown Prosecution Service, the Home Office, the Police, the local police, the immigration authorities, the Prison Service and the Probation Service.

Experienced employment discrimination was measured by combining two variables that asked whether the respondent had been refused/turned down for a job, or had been discriminated against at work with regard to a promotion, due to race or colour in the last five years (coded ‘yes’ or ‘no’). Only those who reported ever having had a job were included in the analyses of employment discrimination.

Health was evaluated using limiting long-term illness (LLTI), one of the most common measures of chronic ill health, which is frequently used as a morbidity index in national health surveys¹⁸ including the census, and as a predictor of mortality and health service utilization.^19–21 LLTI was analysed using a derived measure based on two questions: ‘do you have any long-standing illness, disability or infirmity? By long-standing I mean anything that has troubled you over a period of time or that is likely to affect you over a period of time?’ Those participants who responded affirmatively were then asked, ‘does this illness or disability limit your activities in any way?’

SEP was measured using the National Statistics Socio-economic Classification of occupation, categorized into ‘higher and lower management’, ‘intermediate, small employers and lower supervisory’, ‘semi-routine and routine’ and ‘other’ which included never worked and long-term unemployed.

Ethnicity was measured as a self-reported variable, and was categorized into White, Indian, Pakistani, Bangladeshi, Black Caribbean and Black African. Other ethnic groups covered too few respondents to be considered in the analyses presented here. Age (analysed as a continuous variable) and sex were also included.

Statistical analysis

Multilevel logistic regression models were used to account for the clustering of individuals within Middle Super Output Areas (MSOAs; minimum 5000 residents, mean 7200) to ascertain the demographic characteristics of ethnic minority people who report racism, with separate regression models used for each type of discrimination.

To examine the contribution of racism to ethnic inequalities in LLTI, over and above socio-economic factors, three-step multilevel logistic regression models were conducted entering age, sex and ethnicity first, then SEP, and then adding racism in the last step. The model examining the contribution of employment discrimination to ethnic inequalities in LLTI was conducted only amongst respondents who reported ever having had a paid job, and included an additional control for current economic activity (categorized into: ‘in paid employment’ or ‘not economically active’) in order to account for a healthy worker effect.

The moderating effect of age, gender, ethnicity and SEP in the relationship between racism and LLTI was investigated by including an interaction between racism and the variable of interest in the model. All analyses were conducted using STATA 9.²²

Results

White and Indian respondents tended to be older and in higher employment categories than people in all other ethnic groups, whereas Pakistani and Bangladeshi respondents were found to be in the lower employment classes (table 1). About half of the Bangladeshi, Indian and Pakistani sample reported being fairly or very worried about experiencing racial/religious attacks. Unadjusted analysis show that among those who reported ever having had a job, Black African people reported more experiences of employment discrimination and, in conjunction with Black Caribbean respondents, a greater percentage of expected organizational discrimination.

Table 2 shows the mutually adjusted socio-demographic characteristics of people who reported experiencing racism, which varied greatly by type of discrimination. Females were significantly more likely to report fear of racial and religious attacks, but reported lower odds of expected organizational discrimination and employment discrimination. Decreased odds of experiencing expected organizational discrimination and discrimination in employment were reported by older people.

Reports of fear of experiencing interpersonal racism significantly increased as social class decreased, with respondents in semi-routine and routine occupations being one and a half times as likely to report fear of racial/religious attacks as those in the highest employment category. In contrast, expected organizational discrimination and employment discrimination were less likely for people in the lowest employment categories (table 2).

Variations in reporting discrimination were observed across ethnic groups. Compared with Indians as the reference group, Black Caribbean and Black African people were less likely to report fear of interpersonal discrimination and more likely to report expected organizational discrimination and experienced employment discrimination.

When compared with White people, Pakistani, Bangladeshi and Black Caribbean respondents were more likely to report having an LLTI (table 3, step 1). Tables 3 and 4 examine the contribution of racism to ethnic inequalities in LLTI after adjusting for socio-economic factors. Results show an association between fear of racial/religious attack and expected organizational discrimination with increased reporting of LLTI (β = 0.24, 95% CI 0.11–0.37 and β = 0.30, 95% CI 0.20–0.41, respectively). Employment discrimination was not associated with reports of LLTI (β = −0.18, 95% CI −0.75 to 0.38). The log odds of having an LLTI for each ethnic minority (compared with whites) was reduced when fear of racial/religious harassment was included in the model (table 3, step 3a). Inclusion of expected organizational discrimination (table 3, step 3b) resulted in reduced log odds of LLTI for Black Caribbean and Black Afican people. When employment discrimination was included, the log odds for LLTI were reduced for all ethnic minorities (compared with whites). In other words, socio-economic characteristics explained some of the variations in LLTI between ethnic groups. Experiences of racism further explained these ethnic variations in LLTI.

Finally, the last set of analyses examined the moderating effect of gender, age, ethnicity and socio-economic status on the relationship between racism and LLTI (table 3, steps 4a and 4b; table 4, step 4). No significant interactions were found between socio-demographic factors and any of the three discrimination types.

Discussion

This study set out to identify the demographic characteristics associated with an increased risk of reporting experienced racism; to explore the association between racism and the health of ethnic minority people and its contribution to ethnic inequalities in health; and to explore the moderating role that gender, age, ethnicity and SEP have in the relationship between racism and health.

Results show interesting variations in socio-demographic characteristics of racism victims across different types of discrimination. For example, females reported greater odds of experiencing fear of interpersonal racism, but were less likely to report expected organizational discrimination or employment discrimination. Perhaps women are more prone to verbalize fear of experiencing physical assaults, but less likely to report other types of discrimination. This has been suggested by previous studies that have shown women to be more likely to say that they have not been discriminated against,²³ and more likely to internalize their negative experiences, keeping quiet about it and accepting unfair treatment.²⁴

Older people reported less organizational and employment discrimination than younger respondents. Although other studies have found similar findings,^25–27 the association between age and reports of experienced discrimination remains inconclusive, with some studies reporting opposite results, or no association.⁷

The effect of SEP differed between types of discrimination. Whereas people in lower SEP reported greater odds of experiencing fear of interpersonal racism, the relationship was reversed for expected organizational and employment discrimination, although only significant in the former, possibly because of low statistical power to detect differences in less frequently experienced employment discrimination. Limiting analyses to respondents who reported ever having had a paid job might underestimate the level of employment discrimination for those who have been seeking work but never had a paid job. We decided to take this analytical approach in order to select out those respondents who had never sought work, and were therefore not at risk of experiencing discrimination in employment.

Conversely, people in lower SEP were less likely to report expected organizational and employment discrimination. As with the effect of age and gender, studies have found inconsistent results.⁷ However, most of these studies have been conducted in the United States, and it is difficult to translate and compare results due to the differing historical contexts between countries. A recent study conducted in the United Kingdom by Kelaher and colleagues²⁸ studied the relationship between education, racism and health in a stratified community survey of 692 adults living in Leeds²⁸ and found higher levels of education to be associated with greater reports of experienced discrimination. The authors argue that findings may be the result of greater awareness and reporting of discrimination and/or increased exposure to situations in which discrimination might occur.²⁸ We expand Kelaher and colleagues’ arguments by citing new work in the arena of applied organizational psychology, which reports perceived racial discrimination at work to be less prevalent among those with same ethnicity supervisors.³ It is possible that people in lower SEP are more likely to have a supervisor of the same ethnicity, and thus be less likely to experience work-related discrimination. Unfortunately, the 2005 CS does not provide data on employer ethnicity, so it was not possible to test whether this is the case. However, previous studies have found that ethnic minority people are likely to work amongst co-ethnics, which would, as Kelaher and colleagues suggested, decrease exposure to situation where they can be discriminated against. For example, analysis of the Fourth National Survey on Ethnic Minorities showed Bangladeshi men to be overrepresented in the catering sector, with 60% of them working in waiting and kitchen work in restaurants.²⁹ It has also been shown that ethnic minorities are overrepresented in the self-employment category, which in turn has resulted from discrimination against ethnic minorities in paid employment.³⁰ It is thus important to note that although people in lower SEP categories are less likely to report experiencing discrimination, the fact that ethnic minorities are overrepresented in lower social classes is a consequence in itself of racial discrimination and disadvantage.^12,31,32

Whereas fear of interpersonal racism and expected organizational racism were significantly associated with reports of LLTI, employment discrimination was not. Although we attempted to control for the healthy worker effect by adding an additional employment control, it is possible that the findings of a non-significant protective effect of employment discrimination on LLTI are the result of an unmeasured component not captured in our statistical models. This discrepancy in the findings creates an interesting question: is fear of racism, rather than actual experienced racism, more detrimental to health? It is possible that the fear of an expected physical or verbal assault produces more stress than an occurred event. Studies have found that living with the fear of a racist attack is more frequent than reports of experienced racism,^33,34 and have reported it to be highly detrimental to health. In an analysis of the Fourth National Survey on Ethnic Minorities, Karlsen and Nazroo³⁵ found that independent of the effects of age, gender and household social class, those people who reported being worried about being harassed were 61% more likely to report poor health than those who did not.

Findings from this study show that the detrimental association between racism and LLTI is not an effect of an uncontrolled SEP factor, but rather, an independent effect of racism on health. Although we found a large individual-level mark in increased exposure to racism, particularly for fear of racism and expected organizational discrimination, a small effect at the group level was detected, which contributed partially to ethnic inequalities in LLTI.

Finally, we found no evidence that the impact of racism on LLTI was different across groups defined by gender, age or SEP. Although moderating effects of SEP have been found in the United States,^36,37 such results have not been found in the United Kingdom.³⁸ We conducted additional analyses to examine whether a different measure of SEP, such as education or income, had a moderating effect in the relationship between racism and health, but did not find any significant results, irrespective of the SEP measure used (results available from the authors on request).

Limitations

This study examines a wider range of different types of racism than has previously been investigated. Nevertheless, limitations in the racism measures remain. Only employment discrimination asked directly about experienced racism. Moreover, fear of interpersonal discrimination was worded as fear of racial/religious attacks, and it is thus possible that religion acted as a confounder in our results. However, analyses were conducted to test this by adding religion to the analytical model, which did not produce a change in the results reported.

LLTI was used as the health outcome in this study. Although it has been found to predict mortality and health service utilization,^18–21 it is not the usual health outcome of choice for investigating effects of discrimination on health. However, alternative measures are not available in the CS 2005. Finally, the study suffered from small sample sizes (only 241 Bangladeshi respondents) and low prevalence rates, particularly for employment discrimination, which might have had an impact on the non-significance of the results found.

In conclusion, significant variations exist in the types of racial discrimination most commonly reported across different socio-demographic groups. Despite substantial differences in the experience of racial discrimination, the detrimental impact of racism on health was the same across groups defined by age, gender and SEP. This and prior work¹¹,^39–42 show that socio-economic factors are a major part of the explanation for the poorer health experienced by some ethnic minorities in the United Kingdom. Our results show that, after controlling for socio-economic factors, fear of racism and employment discrimination additionally contribute to ethnic inequalities in health.

Conflicts of interest: None declared.

References