In defence of ‘race’ and ethnicity as analytical variables in epidemiological research George TH Ellison St George’s – University of London.

In defence of ‘race’ and ethnicity

as analytical variables in epidemiological

researchGeorge TH Ellison

St George’s – University of London

Overview

Overview

• What do ‘race’ and ethnicity actually mean?• What are the problems with ‘race’ and

ethnicity?• Should we use ‘race’ and ethnicity? If so,

when?

• How should we measure ‘race’ and ethnicity?

Overview

• What do ‘race’ and ethnicity actually mean?• What are the problems with ‘race’ and

ethnicity?• Should we use ‘race’ and ethnicity? If so,

when?

• How should we measure ‘race’ and ethnicity?

Overview

• What do ‘race’ and ethnicity actually mean?• What are the problems with ‘race’ and

ethnicity?• Should we use ‘race’ and ethnicity? If so,

when?

• How should we measure ‘race’ and ethnicity?

Overview

• What do ‘race’ and ethnicity actually mean?• What are the problems with ‘race’ and

ethnicity?• Should we use ‘race’ and ethnicity? If so,

when?

• How should we measure ‘race’ and ethnicity?

Where do the concepts of ‘race’ and ethnicity come

from?

Where do the concepts of ‘race’ and ethnicity come

from?• ‘Race’ – is traditionally used by biologists as a

synonym for ‘subspecies’ (isolated populations of a species that contain individuals who are genetically more closely related to one another than to the rest of the species).

• Ethnicity – is traditionally used by anthropologists to classify human populations on the basis of shared ancestry, culture and customs.

Where do the concepts of ‘race’ and ethnicity come

from?• ‘Race’ – is traditionally used by biologists as a

synonym for ‘subspecies’ (isolated populations of a species that contain individuals who are genetically more closely related to one another than to the rest of the species).

• Ethnicity – is traditionally used by anthropologists to classify human populations on the basis of shared ancestry, heritage, culture and customs.

Are ‘race’ or ethnicity genetic variables?

Are ‘race’ or ethnicity genetic variables?

• ‘Race’ – captures little of the modest geographical and sociocultural gradients in genetic variation and there are no genetically discrete human ‘races’.

But, ironically…

• Ethnicity – is routinely measured using aspects of appearance, ancestry and/or culture which are linked to phenotypic, geographical and reproductive factors which reflect/are responsible for genetic variation.

Are ‘race’ or ethnicity genetic variables?

• ‘Race’ – captures little of the modest geographical and sociocultural gradients in genetic variation and there are no genetically discrete human ‘races’.

Yet, ironically…

• Ethnicity – is routinely measured using aspects of appearance, ancestry and/or culture which are linked to phenotypic, geographical and reproductive factors which reflect/are responsible for genetic variation.

Are ‘race’ or ethnicity genetic variables?

• ‘Race’ – captures little of the modest geographical and sociocultural gradients in genetic variation and there are no genetically discrete human ‘races’.

Yet, ironically…

• Ethnicity – is routinely measured using aspects of appearance, ancestry and/or culture which are linked to phenotypic, geographical and reproductive factors which reflect or are responsible for genetic variation.

What have the concepts of ‘race’ and ethnicity come to

mean?

What have the concepts of ‘race’ and ethnicity come to

mean?

• ‘Race’ – is seen as a socially-constructed and arbitrary classification based on discredited theories of innate genetic difference between human populations.

• Ethnicity – has become conflated with ‘race’ and is increasingly used as a (sanitised) synonym for ‘race’, hence the biosocial term ‘race/ethnicity’.

What have the concepts of ‘race’ and ethnicity come to

mean?

• ‘Race’ – is seen as a socially-constructed and arbitrary classification based on discredited theories of innate genetic difference between human populations.

• Ethnicity – has become conflated with ‘race’ and is increasingly used as a (sanitised) synonym for ‘race’, hence the biosocial term ‘race/ethnicity’.

Why have ‘race’ and ethnicity been criticised as analytical

variables?

Why have ‘race’ and ethnicity been criticised as analytical

variables? Reliability

Why have ‘race’ and ethnicity been criticised as analytical

variables? Reliability

Validity

Why have ‘race’ and ethnicity been criticised as analytical

variables? Reliability

Validity

Sensitivity

Why have ‘race’ and ethnicity been criticised as analytical

variables? Reliability • There is little consensus on

how ‘race’ or ethnicity should be defined or measured.

• ‘Race’ and ethnicity are fluid, subjective and context-specific characteristics that are therefore inherently unstable.

Why have ‘race’ and ethnicity been criticised as analytical

variables? Reliability • There is little consensus on

how ‘race’ or ethnicity should be defined or measured.

• ‘Race’ and ethnicity are fluid, subjective and context-specific characteristics that are therefore inherently unstable.

Why have ‘race’ and ethnicity been criticised as analytical

variables? Reliability • There is little consensus on

how ‘race’ or ethnicity should be defined or measured.

• ‘Race’ and ethnicity are fluid, subjective and context-specific characteristics that are (therefore) inherently unstable.

Why have ‘race’ and ethnicity been criticised as analytical

variables? Validity • ‘Racial’ and ethnic categories

rarely reflect genetically or socioculturally discrete and homogeneous populations.

• ‘Racial’ and ethnic categories are rarely viewed as the consequences of the structures which make classification possible.

Why have ‘race’ and ethnicity been criticised as analytical

variables? Validity • ‘Racial’ and ethnic categories

rarely reflect genetically or socioculturally discrete and homogeneous populations.

• ‘Racial’ and ethnic categories are rarely viewed as the consequences of the structures which make classification possible.

Why have ‘race’ and ethnicity been criticised as analytical

variables? Validity • ‘Racial’ and ethnic categories

rarely reflect genetically or socioculturally discrete and homogeneous populations.

• ‘Racial’ and ethnic categories are rarely viewed as the consequences of the structural forces which make classification possible.

Why have ‘race’ and ethnicity been criticised as analytical

variables? Sensitivity• ‘Race’ and ethnicity tend to

essentialise any differences observed as intrinsically genetic or sociocultural rather than the consequence of extrinsic structural forces.

• Using ‘race’ and ethnicity as if they were precise markers of difference tends to generalise, stereotype and stigmatise.

Why have ‘race’ and ethnicity been criticised as analytical

variables? Sensitivity• ‘Race’ and ethnicity tend to

essentialise any differences observed as intrinsically genetic or sociocultural rather than the consequence of extrinsic structural forces.

• Using ‘race’ and ethnicity as if they were precise markers of difference tends to generalise, stereotype and stigmatise.

Why have ‘race’ and ethnicity been criticised as analytical

variables? Sensitivity• ‘Race’ and ethnicity tend to

essentialise any differences observed as intrinsically genetic or sociocultural rather than the consequence of extrinsic structural forces.

• Using ‘race’ and ethnicity as if they were precise markers of difference tends to generalise, stereotype and stigmatise.

When should we use ‘race’ and ethnicity in

epidemiology?

When should we use ‘race’ and ethnicity in

epidemiology? Reliability

Some definitions of ‘race’ and ethnicity can be reliably measured in some populations, in some contexts, at some times.

ValidityThe characteristics used to classify ‘race’ and ethnicity (appearance, ancestry and culture) can be correlated with determinants of health status: genes; behaviour; and class.

When should we use ‘race’ and ethnicity in

epidemiology? Reliability

Some definitions of ‘race’ and ethnicity can be reliably measured in some populations, in some contexts, at some times.

ValidityThe characteristics used to classify ‘race’ and ethnicity (culture, appearance and ancestry) can be correlated with determinants of health status: culture; class; and genes.

Health

Race/ethnicity

Taken from: Nazroo (1997) The Health of Britain’s Ethnic Minorities.

When should we use ‘race’ and ethnicity in

epidemiology?

Health

Culture

Class

Genetics

Race/ethnicity

Taken from: Nazroo (1997) The Health of Britain’s Ethnic Minorities.

When should we use ‘race’ and ethnicity in

epidemiology?

Access to services

Health

Culture

Class

Genetics

Migration

Discrimination

Race/ethnicity

Taken from: Nazroo (1997) The Health of Britain’s Ethnic Minorities.

When should we use ‘race’ and ethnicity in

epidemiology?

When should we use ‘race’ and ethnicity in

epidemiology? ReliabilitySome definitions of ‘race’ and ethnicity can be reliably measured in some populations, in some contexts, at some times.

ValidityThe characteristics used to classify ‘race’ and ethnicity (appearance, ancestry and culture) can be correlated with determinants of health status: culture; class; and genes.

Sensitivity‘Race’ and ethnicity are crucial markers of the tendency to generalise, stereotype and stigmatise.

Three ‘tests’ when using ‘race’

Three ‘tests’ when using ‘race’

• Reliability – Can it be reliably measured?

• Validity – Is it the most reliable proxy?

• Sensitivity – Is it acceptable?

Three ‘tests’ when using ‘race’

• Reliability – Can it be reliably measured?

• Validity – Is it the most reliable proxy?

• Sensitivity – Is it acceptable?

Three ‘tests’ when using ‘race’

• Reliability – Can it be reliably measured?

• Validity – Is it the most reliable proxy?

• Sensitivity – Is it acceptable?

Summary and recommendations

Summary and recommendations

• Using ‘race’ or ethnicity remains contentious.• There have been exaggerated claims about

their unreliability, invalidity and (in)sensitivity. • They are useful (if not crucial) in some

contexts:- as a proxy when no others are available; and - for measuring its (potential) consequences.

• Their principal utility is a consequence of their past/present use to stigmatise and discriminate.

• To reduce these effects the reasons for their utility should always be explained and they should only be used when absolutely necessary.

Summary and recommendations

• Using ‘race’ or ethnicity remains contentious.• There have been exaggerated claims about

their unreliability, invalidity and (in)sensitivity. • They are useful (if not crucial) in some

contexts:- as a proxy when no others are available; and - for measuring its (potential) consequences.

• Their principal utility is a consequence of their past/present use to stigmatise and discriminate.

• To reduce these effects the reasons for their utility should always be explained and they should only be used when absolutely necessary.

Summary and recommendations

• Using ‘race’ or ethnicity remains contentious.• There have been exaggerated claims about

their unreliability, invalidity and (in)sensitivity. • They are useful (if not crucial) in some

contexts:- as a proxy when no others are available; and - for measuring its (potential) consequences.

• Their principal utility is a consequence of their past/present use to stigmatise and discriminate.

• To reduce these effects the reasons for their utility should always be explained and they should only be used when absolutely necessary.

Summary and recommendations

• Using ‘race’ or ethnicity remains contentious.• There have been exaggerated claims about

their unreliability, invalidity and (in)sensitivity. • They are useful (if not crucial) in some

contexts:- as a proxy when no others are available; and - for measuring its (potential) consequences.

• Their principal utility is a consequence of their past/present use to stigmatise and discriminate.

• To reduce these effects the reasons for their utility should always be explained and they should only be used when absolutely necessary.

Summary and recommendations

• Using ‘race’ or ethnicity remains contentious.• There have been exaggerated claims about their

unreliability, invalidity and (in)sensitivity. • They are useful (if not crucial) in some contexts:

- as a proxy when no others are available; and - for measuring their (potential) consequences.

• Their principal utility is a consequence of their past/present use to stigmatise and discriminate.

• To reduce these effects the reasons for their utility should always be explained and they should only be used when absolutely necessary.

Summary and recommendations

• Using ‘race’ or ethnicity remains contentious.• There have been exaggerated claims about their

unreliability, invalidity and (in)sensitivity. • They are useful (if not crucial) in some contexts:

- as a proxy when no others are available; and - for measuring their (potential) consequences.

• Their principal utility is a consequence of their past/present use to stigmatise and discriminate.

• To reduce these effects the reasons for their utility should always be explained and they should only be used when absolutely necessary.

Summary and recommendations

• Using ‘race’ or ethnicity remains contentious.• There have been exaggerated claims about their

unreliability, invalidity and (in)sensitivity. • They are useful (if not crucial) in some contexts:

- as a proxy when no others are available; and - for measuring their (potential) consequences.

• Their principal utility is a consequence of their past/present use to stigmatise and discriminate.

• To reduce these effects the reasons for their utility should always be explained and they should only be used when absolutely necessary.

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?• Data are not routinely collected using traditional

‘racial’ or ethnic categories (such as ‘Caucasian’ or ‘Black’).

• Danish populations are likely to be unfamiliar with these traditional ‘racial’ or ethnic categories.

• Few analyses have therefore been conducted on inequalities in health related to traditional ‘racial’ or ethnic categories.

But…• Data are available on two of the characteristics

that are traditionally used to classify ‘race’ and ethnicity: Ancestry – place of birth and country of origin. Culture – language, religion and various behaviours.

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?• Data are not routinely collected using traditional

‘racial’ or ethnic categories (such as ‘Caucasian’ or ‘Black’).

• Danish populations are likely to be unfamiliar with these traditional ‘racial’ or ethnic categories.

• Few analyses have therefore been conducted on inequalities in health related to traditional ‘racial’ or ethnic categories.

But…• Data are available on two of the characteristics

that are traditionally used to classify ‘race’ and ethnicity: Ancestry – place of birth and country of origin. Culture – language, religion and various behaviours.

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?• Data are not routinely collected using traditional

‘racial’ or ethnic categories (such as ‘Caucasian’ or ‘Black’).

• Danish populations are likely to be unfamiliar with these traditional ‘racial’ or ethnic categories.

• Few analyses have therefore been conducted on inequalities in health related to traditional ‘racial’ or ethnic categories.

But…• Data are available on two of the characteristics

that are traditionally used to classify ‘race’ and ethnicity: Ancestry – place of birth and country of origin. Culture – language, religion and various behaviours.

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?• Data are not routinely collected using traditional

‘racial’ or ethnic categories (such as ‘Caucasian’ or ‘Black’).

• Danish populations are likely to be unfamiliar with these traditional ‘racial’ or ethnic categories.

• Few analyses have therefore been conducted on inequalities in health related to traditional ‘racial’ or ethnic categories.

But…• Data are available on two of the characteristics

that are traditionally used to classify ‘race’ and ethnicity: Ancestry – place of birth and country of origin. Culture – language, religion and various behaviours.

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?• Data are not routinely collected using traditional

‘racial’ or ethnic categories (such as ‘Caucasian’ or ‘Black’).

• Danish populations are likely to be unfamiliar with these traditional ‘racial’ or ethnic categories.

• Few analyses have therefore been conducted on inequalities in health related to traditional ‘racial’ or ethnic categories.

But…• Data are available on two of the characteristics

that are traditionally used to classify ‘race’ and ethnicity: Ancestry – place of birth and country of origin. Culture – language, religion and various behaviours.

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?• Data are not routinely collected using traditional

‘racial’ or ethnic categories (such as ‘Caucasian’ or ‘Black’).

• Danish populations are likely to be unfamiliar with these traditional ‘racial’ or ethnic categories.

• Few analyses have therefore been conducted on inequalities in health related to traditional ‘racial’ or ethnic categories.

But…• Data are available on two of the characteristics

that are traditionally used to classify ‘race’ and ethnicity: Ancestry – place of birth and country of origin. Culture – language, religion and various behaviours.

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?• Data are not routinely collected using traditional

‘racial’ or ethnic categories (such as ‘Caucasian’ or ‘Black’).

• Danish populations are likely to be unfamiliar with these traditional ‘racial’ or ethnic categories.

• Few analyses have therefore been conducted on inequalities in health related to traditional ‘racial’ or ethnic categories.

But…• Data are available on two of the characteristics

that are traditionally used to classify ‘race’ and ethnicity: Ancestry – place of birth and country of origin. Culture – language, religion and various behaviours.

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?A review of the literature on ‘race’ and ethnicity in Denmark suggests that there is evidence for many of the pathways within Nazroo’s model linking race/ethnicity to inequalities in health:

• Genetics – existing sociocultural stratification might facilitate genetic stratification, as would migration from geographical regions where specific disease alleles are more/less prevalent than in Denmark.

• Culture – socioculturally isolated and ‘acculturated’ populations may have differing prevalence of risk.

• Class – prior migrant disadvantage and subsequent discrimination are likely to create populations with different health risks.

Access to services

Health

Culture

Class

Genetics

Migration

Discrimination

Race/ethnicity

Taken from: Nazroo (1997) The Health of Britain’s Ethnic Minorities.

When should we use ‘race’ and ethnicity in

epidemiology?

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?A review of the literature on ‘race’ and ethnicity in Denmark suggests that there is evidence for many of the pathways within Nazroo’s model linking race/ethnicity to inequalities in health:

• Culture – socioculturally isolated and ‘acculturated’ populations may have differing prevalence of risk.

• Class – prior migrant disadvantage and/or social discrimination are likely to create populations with different socioeconomic status and related risks.

• Genetics – existing sociocultural stratification might facilitate genetic stratification, as would migration from geographical regions where specific disease alleles are more/less prevalent than in Denmark.

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?A review of the literature on ‘race’ and ethnicity in Denmark suggests that there is evidence for many of the pathways within Nazroo’s model linking race/ethnicity to inequalities in health:

• Culture – socioculturally isolated and ‘acculturated’ populations may have differing prevalence of risk.

• Class – prior migrant disadvantage and/or social discrimination are likely to create populations with different socioeconomic status and related risks.

• Genetics – existing sociocultural stratification might facilitate genetic stratification, as would migration from geographical regions where specific disease alleles are more/less prevalent than in Denmark.

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?A review of the literature on ‘race’ and ethnicity in Denmark suggests that there is evidence for many of the pathways within Nazroo’s model linking race/ethnicity to inequalities in health:

• Culture – socioculturally isolated and ‘acculturated’ populations may have differing prevalence of risk.

• Class – prior migrant disadvantage and/or social discrimination are likely to create populations with different socioeconomic status and related risks.

• Genetics – existing sociocultural stratification might facilitate genetic stratification, as would migration from geographical regions where specific disease alleles are more/less prevalent than in Denmark.

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?• The data are available to conduct preliminary

analyses exploring inequalities in health that are associated with ancestry and culture.

• These analyses are likely to illuminate the potential contribution of ‘race’ and ethnicity to health inequalities in Denmark.

• It may not be necessary or desirable to collect data using traditional ‘racial’ or ethnic categories to assess the contribution of ‘race’ and ethnicity to health inequalities.

• Beware of collecting data on ‘racial’ or ethnic identity that might create the potential for discrimination where none currently exist.

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?• The data are available to conduct preliminary

analyses exploring inequalities in health that are associated with ancestry and culture.

• These analyses are likely to illuminate the potential contribution of ‘race’ and ethnicity to health inequalities in Denmark.

• It may not be necessary or desirable to collect data using traditional ‘racial’ or ethnic categories to assess the contribution of ‘race’ and ethnicity to health inequalities.

• Beware of collecting data on ‘racial’ or ethnic identity that might create the potential for discrimination where none currently exist.

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?• The data are available to conduct preliminary

analyses exploring inequalities in health that are associated with ancestry and culture.

• These analyses are likely to illuminate the potential contribution of ‘race’ and ethnicity to health inequalities in Denmark.

• It may not be necessary or desirable to collect data using traditional ‘racial’ or ethnic categories to assess the contribution of ‘race’ and ethnicity to health inequalities.

• Beware of collecting data on ‘racial’ or ethnic identity that might create the potential for discrimination where none currently exist.

What is the relevance of ‘race’ and ethnicity to epidemiology

in Denmark?• The data are available to conduct preliminary

analyses exploring inequalities in health that are associated with ancestry and culture.

• These analyses are likely to illuminate the potential contribution of ‘race’ and ethnicity to health inequalities in Denmark.

• It may not be necessary or desirable to collect data using traditional ‘racial’ or ethnic categories to assess the contribution of ‘race’ and ethnicity to health inequalities.

• Beware of collecting data on ‘racial’ or ethnic identity that might create the potential for discrimination where none currently exist.

Further reading:Ellison GTH. ‘Population profiling’ and public health risk: when and how should we use race/ethnicity? Critical Public Health 2005; 15: 65-74.

Ellison GTH, Rees Jones I. Social identities and the ‘new genetics’: scientific and social consequences. Critical Public Health 2002; 12: 265-282.

Ellison GTH, De Wet T, IJsselmuiden CB, Richter LM. Desegregating health statistics and health research in South Africa. South African Medical Journal 1996; 86: 1257-62.