Background Despite lower alcohol drinking rates of UK ethnic minority people (excluding Irish) compared with those of the white majority, events of racial discrimination expose ethnic minorities to unique stressors that elevate the risk for escapist drinking. Studies of ethnic density, the geographical concentration of ethnic minorities in an area, have found racism to be less prevalent in areas of increased ethnic density, and this study hypothesises that ethnic minority people living in areas of high ethnic density will report less alcohol use relative to their counterparts, due to decreased experienced racism and increased sociocultural norms.
Methods Multilevel logistic regressions were applied to data from the 1999 and 2004 Health Survey for England linked to ethnic density data from 2001 census.
Results Respondents living in non-White area types and areas of higher coethnic density reported decreased odds of being current drinkers relative to their counterparts. A statistically significant reduction in the odds of exceeding sensible drinking recommendations was observed for Caribbeans in Black area types, Africans in areas of higher coethnic density and Indian people living in Indian area types.
Conclusion Results confirmed a protective ethnic density effect for current alcohol consumption, but showed a less consistent picture of an ethnic density effect for adherence to sensible drinking guidelines. Previous research has shown that alcohol use is increasing among ethnic minorities, and so a greater understanding of alcohol-related behaviour among UK ethnic minority people is important to establish their need for preventive care and advice on safe drinking practices.
- Ethnic density
- alcohol use
- k-means clustering
- alcohol and health
- cluster analysis
- ethnic minorities SI
Statistics from Altmetric.com
- Ethnic density
- alcohol use
- k-means clustering
- alcohol and health
- cluster analysis
- ethnic minorities SI
Three-quarters of men and over half of women in England drank an alcoholic drink at least once a week in 2005.1 Pooled data from the 2001 to 2005 General Household Survey indicate that of the 66% of Great Britain's adults who reported drinking in the past week, 8% were from an ethnic minority background.2 It is well known that alcohol-drinking rates of ethnic minorities (excluding Irish) are lower than those of the white majority,3–8 and that variations in alcohol intake exist between ethnic groups. For example, Black-Caribbean people are more likely to drink than South Asian people,8 9 and within the South Asian group, Muslim people are less likely to drink than their Sikh and Hindu counterparts.4 10–13
One explanation for differences in drinking behaviour is attitude to alcohol consumption, with the British white majority being relatively permissive in contrast with the reduced use or abstinence of non-white ethnic minority people. Belonging to an ethnic minority group has been stated to provide its members with some protective effects on health through social norms specific to certain behaviours, including breastfeeding14 and substance use, which in this case characterise alcohol consumption as an undesirable activity.11 These protective effects have been reported to diminish as migrants acculturate into the mainstream sociocultural norms. Research in the USA has found an escalation in reported alcohol and other drug use among migrants, particularly Hispanic people, as time in the USA increases.15 This has been suggested to be a result of the deterioration of cultural and family values, and an increased exposure to substance use in the host country.16 Unfortunately, effects of the process of acculturation on health risk behaviours has been little studied in the UK,17 and research conducted in the USA cannot be directly translated to the UK context for several reasons, including different countries of origin of the predominant minority groups, differing reasons for migration, differing timing of migration throughout the last three centuries, and differing cultural, economic and demographic profiles of the ethnic groups represented in the two countries. Nonetheless, some UK studies have found increased alcohol consumption among second-generation ethnic minority people.18 19 For example, a study assessing alcohol consumption among second-generation ethnic minorities in two English Midlands cities found increased alcohol consumption to be correlated with a reduction in both strong religious adherence and membership of exclusively kinship and/or ethnic minority social networks, as well as with a change in sociocultural norms and manners of leisure, which included drinking patterns.18
This study hypothesises that ethnic minority people who live in areas with greater proportions of coethnics, hereby referred to as areas of high coethnic density, will receive a greater benefit from these normative protective effects, as they will be in greater contact with other people of the same ethnic group, and are thus expected to display greater adherence to protective social norms.
Although social norms of ethnic minority groups are suggested to protect their members against alcohol use, events of racial discrimination experienced by ethnic minorities in the UK20–24 expose ethnic minority people to unique stressors that elevate the risk for ‘escapist’ drinking,25 referred to as drinking as a means of reducing and coping with stress. It is not surprising, then, that experiences of racial discrimination have been associated with increased levels of alcohol consumption.26–28 Studies have found that ethnic minority people living in areas of high ethnic density report fewer experiences of interpersonal racism compared with their counterparts living in areas of less ethnic density,20 29 30 so it is possible that ethnic minority people living in areas of high ethnic density will report less alcohol use than those living in less ethnically dense areas, not only because they are more likely to show greater adherence to social norms that label alcohol consumption as undesirable, but also because of a decreased likelihood of experiencing racial attacks and thus utilising alcohol drinking as a means to cope with stress.
The two mechanisms mentioned above (sociocultural norms of drinking behaviour and decreased racial attacks) are part of the ethnic density effect, which proposes that ethnic minority people who reside in areas where they form a greater proportion of the population have better health than ethnic minority residents in areas of lower ethnic minority concentration.29 31 To date, no study has explored the ethnic density effect on alcohol use in the UK, and so no evidence exists on whether ethnic minority people benefit from an ethnic density effect on alcohol consumption.
The present study examines the ethnic density effect on the alcohol use of ethnic minorities in the UK, and hypothesises that ethnic minority people living in an area of high ethnic density will report less alcohol use than their counterparts living in an area of less ethnic density.
This study uses data from two different sources: the adult sample, aged 18 and over, from a merged dataset of the 1999 and 2004 Health Survey for England (HSE), and the 2001 UK Census, which was used to provide data on ethnic density for HSE respondents.
The HSE is an annual survey commissioned by the UK Department of Health since 1991, designed to provide regular information on various aspects of the nation's health.32 The 1999 survey was the first to increase the representation of ethnic minority adults and children from Black Caribbean, Indian, Pakistani, Bangladeshi, Chinese and Irish communities, followed by the 2004 survey, which also focused on ethnic minority people and included the same ethnic minority groups as the 1999 survey, together with an increase in the representation of Black African informants. Sampling points for the ethnic minority samples were identified using information from the UK Census, which allowed areas to be selected on the basis of the concentration of ethnic minority people within them, including areas with low concentrations of ethnic minority residents (for further information on the HSE methodology, see Erens et al32 and Sproston and Mindell33).
Alcohol use was examined as both current drinking and engaging in sensible drinking. Current drinking was measured by asking respondents whether they drank any alcohol nowadays. Among those who reported current drinking, we then considered how many people followed daily sensible drinking limits, measured by the number of units drunk in the heaviest day of the last week. Sensible drinking, defined by the Department of Health as ‘drinking in a way that is unlikely to cause oneself or others significant risk or harm’34 (p. 3), stipulates that men should not drink more than 3–4 units per day, whereas women should not exceed 2–3 daily units. For the purpose of this study, we defined exceeding sensible drinking guidelines as drinking more than 4 units per day for men, and more than 3 units per day for women.
The respondent's ethnicity was measured as a self-report variable and was categorised into White, Indian, Pakistani, Bangladeshi, Black Caribbean and Black African. Due to the low prevalence of alcohol drinking (1.6%), Bangladeshi respondents were not included in the analytical models. Other ethnic groups covered too few respondents to be considered in the analyses presented here.
Analyses controlled for differences in age, gender, socio-economic position of the head of household, nativity (UK vs foreign born), marital status, education, employment status and area deprivation. Socio-economic position was measured using the National Statistics Socio-economic Classification of occupation (NS-SEC) from the household reference person. Marital status was categorised as being single, married or cohabiting, and separated, divorced or widowed. Education was measured by highest educational attainment and was recoded into four categories: none, national vocational qualification (NVQ)1–2, NVQ3–5 and foreign qualifications. Employment status was analysed with a measure that categorised respondents into 0: unemployed, retired or economically inactive, and 1: employed. Area deprivation was captured by deciles of the 2004 Index of Multiple Deprivation (IMD) summary score.
Ethnic density was measured in two different ways: first as a typology of areas based on their ethnic minority composition; second, following the method commonly used in studies of ethnic density, operationalising ethnic density as percentage ethnic minority residents in an area, separately for Caucasians (including White British, White Irish and Other White), Indian, Pakistani, Black Caribbean and Black African people. The former method considers the combination of ethnicities of all residents in the area, whereas the latter considers only own ethnicity versus all other minority and majority ethnicities. Middle super output areas were used to define area boundaries; these have a minimum population of 5000 and an average population of 7200, and were the lowest level of disaggregation that was permissible given identifiability constraints. Permission to link the 2001 census data on ethnicity to the HSE was approved by the ethics committee of the data holder (the National Centre for Social Research) with the constraint that 5% random error be added to each ethnic density variable to minimise risk of disclosure.
Area typologies were created using k-means clustering analyses,35 which partitions the data into a set of clusters, the number of which has been previously specified. The k-means algorithm assigns each data point to a cluster whose centroid is nearest, making the centre of each cluster the average of all its points. Thus, areas with similar levels of ethnic density (based on the 2001 census) across all ethnic groups were placed into the same cluster. Since Bangladeshi people were not included in the analyses, and due to their cultural and residential similarity to Pakistani people, Bangladeshi and Pakistani densities were merged into one variable; so were Black Caribbean and Black African populations, whose density variables showed a high correlation (0.67). In order to determine the number of clusters appropriate for the data, we first started by conducting the analysis with four clusters. However, we found that five clusters yielded a better fit for the data, and the addition of a sixth group had only a marginal impact on the within-group variability.36 Results yielded five different area types (White area type, Mixed area type, Black area type, Indian area type, and Pakistani and Bangladeshi area type) with different distributions of ethnic minority residents.
Multilevel logistic regression models, which account for the sampling of individuals within neighbourhoods (level 2), were conducted to ascertain the impact of ethnic density on current drinking and exceeding sensible drinking guidelines. Due to their low prevalence of current drinking (5%), Pakistani people were not included in the analyses examining sensible drinking.
Separate regression models were conducted for each ethnic group. Analyses were conducted for all respondents over 18 years of age, and were adjusted for age, sex, socio-economic position, nativity, marital status, education, employment status and area deprivation. All analyses were conducted using STATA 9.37
Black African and Pakistani respondents tended to be younger and, together with Black Caribbean people, tended to occupy lower socio-economic positions and live in the most deprived areas. White and, to some extent, Indian respondents tended to be older and have a higher socio-economic status. Indian, Pakistani and White people tended to be married or cohabiting, whereas most of the Black Caribbean and Black African sample were single, married or cohabiting. Pakistani people had the lowest educational attainment, and Indian and Black African people had the highest (see table 1).
Areas classified by cluster analysis as ‘Pakistani and Bangladeshi area types’ were the poorest, with over 80% of these areas in the most deprived decile of the country. Black and Indian area types tended to be less deprived than Pakistani and Bangladeshi area types, but more deprived than Mixed and, especially, White area types.
Table 2 shows the distribution of HSE respondents in the different area types. Almost half of Black African people, and 44% of Black Caribbean respondents lived in Black area types. Only a quarter of Indian people lived in Indian area types, and over a third of Pakistani respondents lived in Pakistani and Bangladeshi area types. White people were the most ethnically concentrated, with only 11% of their population living in non-White area types. Across all ethnic groups, people in lowest socio-economic positions tended to live in Pakistani and Bangladeshi area types, whereas those in higher socio-economic positions tended to live in White area types. For example, among Caribbean respondents, those living in White area types tended to be younger, in higher SES categories, and less likely to have been born overseas, while Caribbean people living in Pakistani and Bangladeshi area types were older, in lower SES categories and born overseas.
Respondents living in non-White area types from all ethnic groups (including white) reported decreased odds of being current drinkers when compared with their counterparts living in White area types (see table 3). For ethnic minority respondents, residence in area types other than White and increasing coethnic density were associated with lower odds of current drinking. This was the case for 17 out of the 20 comparisons made (five types of areas/measurements of ethnic density and four ethnic minority groups) and was statistically significant for 10 comparisons. For three comparisons, a detrimental effect of ethnic density was found whereby Indian people living in Pakistani and Bangladeshi area types, and Indian and Pakistani respondents living in Black area types reported increased odds of current drinking. These associations were not statistically significant.
White respondents were also less likely to drink alcohol if they lived in a non-White area type, although this was only significant in the case of Mixed and Black area types. White people were found to be more likely to drink as coethnic density increased (OR per 10% increase in density: 1.10; 95% CI 1.05 to 1.15; p<0.001).
Table 4 summarises the association between area types and coethnic density and sensible drinking for those who drank alcohol in the past week. ORs show a less consistent pattern here, but a clear statistically significant reduction in the odds of exceeding recommended drinking levels can be seen for Caribbean people in Black area types (OR 0.56; 95% CI 0.37 to 0.84; p<0.01), African people in areas of higher coethnic density (OR per 10% increase: 0.43; 95% CI 0.21 to 0.87; p<0.01) and Indian people living in Indian area types (OR 0.43; 95% CI 0.21 to 0.87; p<0.05; see table 4).
This study set out to examine the ethnic density effect on alcohol use among ethnic minority people in the UK, analysing ethnic density as both a variable of area typologies, and as the percentage of ethnic minority people in an area. Results confirmed a protective ethnic density effect for current alcohol consumption but showed a less consistent picture of an ethnic density effect for adherence to sensible drinking guidelines among those who drank alcohol.
The present study is the first UK study to explore the ethnic density effect on alcohol use. To the best of our knowledge, there is only one previous study that has examined the impact of ethnic density on adult alcohol use. In their classic ecological study, Faris and Dunham31 found alcoholism rates for African–Americans to be lower in Black areas of Chicago. Although Faris and Dunham's results were similar to those found in our study, their research cannot be directly contrasted to our study, due to the differing UK and USA contexts mentioned in the Introduction, and different measures of alcohol use and abuse.
A protective ethnic density effect was found for current alcohol consumption. A consistent picture of decreased likelihood of reports of current drinking was observed across ethnic minority groups for respondents living in all ethnic minority area types, and in areas of higher coethnic density. Only in the case of Pakistani people living in Black area types and white people living in white area types and areas of coethnic density did ethnic density have a detrimental impact on current alcohol use. Previous studies have found Pakistani people to be negatively impacted by the ethnic density effect20 29; however, this is not fully the case in this study, since Pakistani people living in Pakistani and Bangladeshi area types were 75% less likely to report current drinking as compared with their counterparts living in White area types. Moreover, a strong association was found between increased Pakistani ethnic density and decreased likelihood of being a current drinker. It is unclear why living in Black area types was detrimental for Pakistani respondents. An examination of the characteristics of Pakistani people living in all area types yielded no statistically significant differences across areas, although Pakistani people living in Black area types tended to be in higher SES categories. It is possible that Pakistani people living in Black area types were influenced by more tolerant and alcohol-promoting social norms. It would then appear that Pakistani people living in White or Black area types are more likely to be in an environment where drinking alcohol is more normative as compared with Pakistani and Bangladeshi area types. This suggests that a cultural rather than stress explanation underlies the findings.
The introductory section of this article presented the hypothesised causes behind an expected ethnic density effect (protective social norms, decreased experiences of racism). However, it is also important to consider social causation and acculturation bias as possible explanations behind the ethnic density effect found.29 As this and other studies show, the UK white majority reports higher drinking levels than ethnic minority populations.3–8 If individuals who are less acculturated to the majority culture tend to colocate, and if drinking norms are culturally specific such that decreased drinking is more likely to be observed in the less acculturated, then the observed association between ethnic density and alcohol intake may be due to individual level of acculturation rather than a true collective or contextual effect of ethnic density. We tried to control for possible acculturation biases by including a measure of nativity in our analyses, but it is likely that residual effects still remain. It can also be the case that social selection is partly responsible for the results, so that respondents who are more inclined to partake in drinking are less likely to choose to live in an area with other coethnics. The cross-sectional nature of the data means that we were not able to assess this possibility, although it relates to the previous point regarding acculturation.
Although a strong ethnic density effect was found for current drinking, this was not the case for sensible drinking, where ethnic density was protective only in certain cases. Analyses on sensible drinking were conducted among respondents who reported being current drinkers, and so the sample size was significantly reduced. Considering that analyses were stratified by ethnic group, lack of a significant ethnic density effect is possibly due to insufficient analytical power. Nevertheless, a total of 10 out of 15 possible combinations (five types of areas/ethnic density measurements and three ethnic minority groups) showed an OR below 1 indicating a possible (though not always statistically significant) protective effect of ethnic density on exceeding sensible drinking guidelines.
Besides the small sample size in some ethnic groups for the analysis of sensible drinking levels, this study presents other limitations, including the use of a self-report measure of alcohol use. Although self-report measures have been found to offer a reliable and valid approach to measuring alcohol consumption,38 39 due to the stigmatised nature of alcohol consumption in certain ethnic minority communities,40 it is possible that ethnic minority people living in areas of greater ethnic density under-reported their alcohol use. Finally, as previously mentioned, the HSE is a cross-sectional dataset, and thus causality cannot be inferred from our results.
Despite its limitations, this study presents an important contribution to the fields of ethnic density and alcohol consumption among ethnic minorities in the UK. Previous research has shown that alcohol use is increasing among ethnic minorities,4 12 and so a greater understanding of alcohol-related behaviour among UK ethnic minority people is important to establish their need for preventive care and advice on safe drinking practices.6
What is already known on this subject
Alcohol drinking rates of ethnic minorities (excluding Irish) are lower than those of the white majority, and variations in alcohol intake exist between ethnic groups. However, events of racial discrimination experienced by ethnic minorities in the UK expose ethnic minority people to unique stressors that elevate the risk for drinking as a means of reducing and coping with stress. It is not surprising, then, that experiences of racial discrimination have been associated with increased levels of alcohol consumption.
Studies have found that ethnic minorities living in areas of high ethnic density report fewer experiences of interpersonal racism compared with their counterparts living in areas of less ethnic density, so it is plausible that ethnic minority people living in areas of high ethnic density will report less alcohol use than those living in less ethnically dense areas.
What this study adds
This is the first study to examine the impact of ethnic density on adult alcohol use in the UK. Results confirmed a protective ethnic density effect for current alcohol consumption but showed a less consistent picture of an ethnic density effect for adherence to sensible drinking guidelines among those who drank alcohol.
Alcohol use is increasing among ethnic minorities, and so a greater understanding of alcohol-related behaviour among UK ethnic minority people is important to establish their need for preventive care and advice on safe drinking practices.
The authors would like to acknowledge J Laurence for the idea of creating a set of area typologies using k-means clustering analyses, utilised in his working paper, ‘The Effect of Diversity and Community Disadvantage on Social Cohesion: A Multi-Level Study of Social Capital and Inter-Ethnic Relations in UK Communities’ and presented at the 2008 Harvard–Manchester Summer Workshop on Immigration and Social Change in Britain and the USA.
Funding This work was supported by the UK Economic and Social Research Council [RES-163-25-0043].
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.