Race and Ethnicity | Custom PHD Thesis

Using the Sapien and Moscou articles, discuss the following:

The Profiling by Appearance and Assumption – SapienDr. Sapien is responding to an article about the dangers of subjectively determining a patient’s race and ethnicity in case presentations.
1) What are the dangers of making assumptions about a patient’s race, ethnicity, and culture?

2) Do these assumptions lead to differences in treatment and inequalities in health care?

Conceptualizing Race/Ethnicity – Moscou

How does Dr. Moscou address the use of race and ethnicity in research?

Nursing Inquiry 2008; 15(2): 94–105

Feature

The conceptualization and
operationalization of race and
ethnicity by health services researchers
Blackwell Publishing Ltd

Susan Moscou
Mercy College, Dobbs Ferry, NY, USA
Accepted for publication 29 October 2007

MOSCOU S. Nursing Inquiry 2008; 15: 94–105
The conceptualization and operationalization of race and ethnicity by health services researchers
Racial and ethnic variables are routinely used in health services research. However, there is a growing debate within nursing
and other disciplines about the usefulness of these variables in research. A qualitative study was undertaken (July 2004 – November
2004) to ascertain how researchers conceptualize and operationalize racial and ethnic data. Data were derived from interviews
with 33 participants in academic health centers in differing geographic regions. Content analyses extracted manifest and latent
meanings to construct categories depicting respondents’ understandings of race and ethnicity in research. Race and ethnicity
held several meanings but the subtext was often not clear because these terms were not operationalized. Measuring race and
ethnicity quantitatively necessitated uniform classifications thus it was often necessary to impose a single racialized identity.
Respondents recognized the problems with racial and ethnic variables but the majority still believed these variables were
necessary and useful. Several researchers understood that racial and ethnic variables were used in ways that may stigmatize the
populations studied. These respondents collected data on variables other than race and ethnicity to ascertain the causes of
health differentials. The policy recommendation calls for a shift in thinking about how to use racial and ethnic variables in
research.
Key words: critical race theory, ethnicity, qualitative research, race, theoretical framework.

‘Race’ and ‘ethnicity’ are routinely included as demographic
variables in health services research. Health services researchers
use racial and ethnic identifiers to track and document
disparities in health outcomes (LaVeist 1996; Williams 1999),
explain treatment variations (Gonzalez-Burchard et al. 2003),
describe populations (Hahn 1992; OMB 1997b), denote
risk markers for diseases (Waldenstrom 1990; Bhopal and
Donaldson 1998), and develop policy (Alland 2002).
Racial and ethnic classifications also function as the
language for racial discourse and serve as the ‘intellectual
products, social markers, and policy tools’ for society
(Nobles 2000, 1738; Omi and Winant 1994). The concept of
categorizing individuals by race and ethnicity, moreover, is

Correspondence: Susan Moscou, Mercy College, Nursing Programs, 555
Broadway, Dobbs Ferry, NY 10522, USA
E-mail: smoscou@mercy.edu

so widely assumed and strongly acknowledged that these
variables have become our normative lens in which we
view the world and impose discussion about human variation
(Stanfield 1993, 3; Fausto-Sterling 2004).
Racial and ethnic definitions (and connotations) depend
on location, class and nationality thus collecting and
analyzing racial and ethnic data pose conceptual and
practical problems (Goldberg 1992; Zuberi 2001; Marks 2005).
Moreover, racial and ethnic categories as used in health
services research lack precision, often fail to capture the
diverse backgrounds of research subjects (McKenney
and Bennett 1994; Culebras 1995; Hahn 1999; Perot and
Youdelman 2001; Graves 2001; Buescher, Gizlice and JonesVessey 2005), and often function as attributes of socioeconomic
status and other proxy variables. Because race and ethnicity
are rarely operationalized in research (Schulman et al. 1995;
Drevdahl, Phillips and Taylor 2006; Ma et al. 2007), these
variables are fraught with contradiction and confusion for

Race and ethnicity in health services research

both the researcher using them and those interpreting the
clinical findings and policy recommendations.
Drevdahl, Phillips and Taylor published a feature
article in Nursing Inquiry about racial and ethnic variables in
nursing research. Their article contributed to the growing
critique as well as vibrant discussions within nursing
(Drevdahl, Phillips and Taylor 2006), epidemiology ( Jones,
LaVeist and Lillie-Blanton 1991; LaVeist 1996; Cooper,
Kaufman and Ward 2003; Krieger 2005), public health
(Fullilove 1998; Buehler 1999; Bhopal 2002), medicine
(Anderson et al. 2001; Rivara and Finberg 2001; Moscou
et al. 2003), health services (Hahn, Mulinare and Teutsch
1992; Williams 1994; Culebras 1995; Bhopal and Rankin
1999; Laws 2001), and journal editors (Kaplan and Bennett
2003; Winker 2006) about the usefulness of racial and
ethnic variables in research, the problems engendered by
using racial and ethnic variables in research and better ways
to utilize race and ethnicity in research.
Although discussion about racial and ethnic variables
exists within these disciplines, there has been a scarcity of
empirical studies investigating how health services researchers
in the USA operationalize and conceptualize race and
ethnicity or the researchers’ role in advancing the ‘production of knowledge and the politics of doing research on
race and ethnicity’ (Gunarathnam 2003, 3). Examining how
researchers and professionals define race and ethnicity has
largely been the purview of sociologists and anthropologists.
These disciplines have explored concepts of race with
professors and scientists and how their definitions may affect
ways in which race was taught, written about and perceived
by the public.
Morning explored the definitions and meanings of
race with biology professors, anthropology professors and
college students (Morning 2004). Lieberman, Hampton,
Littlefield and Hallead asked anthropology and biology
professors to define race (Lieberman et al. 1992). Ellison
and Outram interviewed geneticists and editors of genetic
journals about the social and scientific meanings of race
as well as the reliability and validity of these classifications in
research (Outram and Ellison 2005). To date, no such inquiry
within the health services research sector exists in the USA.
Knowing the meaning(s) researchers ascribe to race and
ethnicity is essential because science establishes the salience
of these concepts (Omi and Winant 1994; Alland 2002).
Furthermore, how researchers construct and operationalize race and ethnicity (biological vs cultural vs social)
reveal the questions they pose in their hypothesis, their
research design, the data they collect (or do not collect), their
analysis of the data, and their interpretation of racial and
ethnic health differentials. How researchers interpret these
© 2008 The author. Journal compilation © 2008 Blackwell Publishing Ltd

differences guide their suggested clinical interventions
(individual patient behavior vs neighborhood’s ability to
promote health) and health policy initiatives (genetic
screening vs. wealth redistribution).
This study describes how health services researchers in
the USA conceptualize and operationalize race and ethnicity.
The theoretical frameworks that guided this study were
sociology of knowledge and critical race theory. Sociology
of knowledge informs the production of knowledge within
health services research, its outward transmission (policy
and clinical journals) and its reception by non-researchers.
Critical race theory informs the discourse around race
and ethnicity providing insights into racial and ethnic
metaphors used to explain disparities.

STUDY METHODS AND ANALYSIS
A qualitative study was undertaken ( July 2004 through
November 2004) to ascertain how health services researchers
conceptualize and operationalize race and ethnicity.
Data were derived from semistructured interviews with 33
researchers affiliated with urban academic health centres in
southern, western and north-eastern USA. Several nonprobabilistic sampling strategies were used to determine the
sample size and the selection criteria for researchers.
Quota sampling was used to establish the sample size and
purposeful and reputational sampling strategies were used
to select respondents. These strategies ensured that the
sample was representative of the population studied, that
the sample was consistently drawn from the population of
interest, and that study participants were capable of providing
information about racial and ethnic data collection practices
in health services research (Henry 1990; Bernard 2000).
Establishing the minimum sample size for health services
researchers was carried out by quota sampling. Quota sampling
is used by anthropologists to choose ‘key informants’ who
are knowledgeable about specific ‘domains of life’ in a culture
(Trost 1986; Bernard 2000, 176). Using this sampling design
maximized the selection of key informants who represented
the research environment as well as characterizing variations
among researchers and the research culture.
Quota sampling determined that a minimum sample
size of 30 health services researchers was needed to ensure
that the open-ended interviews would produce relevant
data. The principal independent variables used to stratify
the sample were two genders (male and female) and 15
variable strata consisting of age (five category ranges),
discipline (clinical, non-clinical), diversity (person of colour,
white), region (east, west, south), and research level (junior,
mid-level, senior). Multiplying the number of genders by
95

S Moscou

15 variable strata equalled interviewing a minimum of 30
health services researchers.
Once the necessary sample size was established, study
participants were chosen using purposeful and reputational
sampling methods. Health services study respondents
were identified by several means: (i) poster presentations
viewed by this author at the 2003 Academy Health Services
conference in San Diego (abstracts were available in the
programme); (ii) Health Services Research Projects in Progress
(online database of nationally funded projects) provided
information about funded research projects at academic
health centres in selected city interview sites; and (iii) recommendations of researchers currently interviewed and
previously interviewed in a pilot study carried out in 2002
and 2003.
Health services researchers considered for participation
in this study were the principal investigator (PI) or one of
the PIs on a research project and worked at an academic
health centre in an urban setting. Prospective researchersubjects were contacted in person (Academy Health
conference), by e-mail, or by telephone and asked to
participate in the study. Respondents who agreed to be
interviewed were included in this study.
Choosing health services respondents affiliated with
urban academic health centres allowed the recruitment
of researchers with diverse educational backgrounds and
research topics. Furthermore, the chosen geographical
areas were heavily saturated with academic health centres
and provided an excellent opportunity to find out if racial
and ethnic classifications differed by region.
Researchers were interviewed in person at their offices.
Each audiotaped interview was approximately 30–60 minutes
long. The audiotaped interviews were transcribed verbatim
into Microsoft Word and then entered as electronic documents
into ATLAS.TI version 5, a qualitative data management
software program.
Transcripts were reviewed, coded and then evaluated by
a set of categories that emerged from the data. Descriptive
statistics were acquired from several ‘yes/no’ responses to
questions. Univariate data were generated in SPSS version
11.5.
Content analyses extracted manifest and latent
meanings to create a category system capable of depicting
respondents’ conceptualization and operationalization of
race and ethnicity in research designs.

ASSURING DATA QUALITY
Assuring the quality of the data entailed developing a
codebook to check for intercoder reliability. A sociologist
96

and journal editor were given the coding instructions,
the unit of analysis and the codebook that defined the
thematic categories. Each worked independently applying
codes to the same three transcripts.
The mean intercoder reliability for thematic classifications
was 61%. Agreement in the individual thematic categories
ranged from 50–100%. Coding disagreements may have
occurred because the thematic classifications were similar
and had overlapping meanings and coding three transcripts
was not enough for each coder to synthesize the concepts.

STUDY FINDINGS
We inherit a methodology — we don’t get to make one up.
So, for example in the United States most hospitals use discharge data to collect data on race. Very few, not very few,
none — collect data on some form of socioeconomic class
position. So, we are always confounding and confusing race
variables [that are] hard for us disentangle. That makes us
over emphasize certain aspects of the minority experience
and not understand well, the experience of poverty. MD,
California.

Self-reported demographic characteristics of the study
sample are found in Table 1. Respondents were asked to
identify their research level (senior, mid-level and junior)
but one respondent commented she was ‘brilliant.’
Respondents were asked if they always included racial
and ethnic variables, sometimes included these variables, or
never collected data on race and ethnicity. Participants were
also asked if they considered race and ethnicity as separate
entities (Table 2).
The majority of respondents (82%) always included racial
and ethnic variables in their research designs. A sociologist
in Oregon commented:
[I use race] 100% of the time. I don’t think I’ve ever done
a [research] project where we didn’t collect data [on race].
In fact, I turned away [several] projects recently because it
was one of those white, non-white [classifications]. [Also]
there were so few minorities included [in the project] there
[would be] no way to analyze the data by using race and
ethnicity. Also, I thought, journals and funding agencies
[would] say, ‘Forget it’ [and not publish or fund the
project]. But, also [I rejected this project] because I always
want to ask about [race and ethnicity] as part of the
questions I ask. [Without racial data] — it, would limit the
value of the data — to me.

Eighteen per cent of researchers indicated they ‘sometimes’ included racial and ethnic variables. A physician from
Birmingham, Alabama, noted that racial and ethnic variables
were only included when they were available. However,
this physician also noted, ‘The IRB requires, in our reports
(the) expected racial breakdown (of study participants)
for the initial approval (of the research project).

Race and ethnicity in health services research

Table 1 Health services researchers
self-reported descriptive data (N = 33)

Gender
Male
Female
Age
25– 29
30– 39
40– 49
50– 59
60– 65
Education
Master’s
PhD
EdD
DrPH
MD/PhD
MD/MPH
MD

Percentage

Region

15
18

45%
55%

11
11
11

33%
33%
33%

1
9
10
10
3

3%
27%
30%
30%
9%

East
South
West
Research year
0–10
11–20
21–35

12
12
9

36%
36%
27%

1
16
1
1
3
5
6

3%
48%
3%
3%
9%
15%
18%

Mean year
Research level
Junior
Mid-level
Senior
Brilliant

Table 2 Race/ethnicity in research (N = 33)
N

Percentage

10
6
16
1

30%
18%
48%
3%

Table 3 Essentialist conceptualization (N = 33)
Percentage

Race/ethnicity use
Always
Sometimes

27
6

82%
18%

Race/ethnicity distinct
Yes
No

20
13

61%
39%

When asked if race and ethnicity were treated as distinct
concepts or entities, 61% indicated ‘yes’ and 39% indicated
race and ethnicity were treated as the same entity. A physician
in New York remarked:
I don’t like separating the two of them. I like thinking of
[race and ethnicity] as one [variable]. [Further] I think it is
hard to separate [these terms] like we do [and at times], I
do. But, I just don’t like [the separation because] it doesn’t
feel right.

However, this same physician indicated that even though
participants were permitted to select a race or an ethnicity,
these identifications were ‘lumped’ together when analyzing
the data. The researcher reasoned, ‘I think that [race and
ethnicity] should be lumped [because] they are not separate
issues. I think they are one big issue.’
Many researchers combined the terms race and ethnicity
and created the variable race/ethnicity. One researcher (MD
© 2008 The author. Journal compilation © 2008 Blackwell Publishing Ltd

Percentage

Race is biological
Yes
No
Yes/no
Open question
Struggling
Maybe

12
13
5
1
1
1

36%
39%
15%
3%
3%
3%

Ethnicity is biological
Yes
No
Yes/no
NA (choice)

4
25
3
1

12%
76%
9%
3%

in Alabama) reported, ‘I try to always use the term race/ethnicity because I think they are so intermingled.’
Researchers were asked whether they believed that
race and ethnicity were a biological construct, a cultural construct, social construct and a political construct (Table 3).
The theme essentialism was identified to denote those
researchers that held a biological or genetic understanding
of race and ethnicity. The essentialist definition of race was
held by 36% of the respondents.
Respondents equating race with biology accepted
that race approximated the genetic make-up in groups
97

S Moscou

Table 4 Constructionist conceptualizations of race and ethnicity (N = 33)
Yes

Percentage

Race
Cultural
Social
Political

26
30
28

79%
91%
85%

Ethnicity
Cultural
Social
Political

31
32
18

94%
97%
55%

Percentage

Yes/No

Percentage

DK/NA

Percentage

6
2
3

18%
6%
9%

0
0
0

0%
0%
0%

1
1
2

3%
3%
6%

1
0
14

3%
0%
42%

0
0
0

0%
0%
0%

1
1
1

3%
3%
3%

identified as black, white, Hispanic, Asian, Native American
and so forth. A physician in Atlanta observed that race is biology,
if you understand that race represents ‘how one looks’:
Race in its purest form is a biological construct because you
know people are phenotypically different. So, if you base
[race] on phenotype [then race] is a biological construct.
But, I think how [race is] used [in research] has a lot of
social overlay.

On the other hand, 15% believed that race had both
biological and social meanings, thus answered ‘yes’ and ‘no’
to the question ‘is race biological’. Another group of
researchers (9%) gave comments such as ‘I’m struggling
with this concept’, ‘It’s still an open question’, and ‘Maybe.’
The researcher ‘struggling’ remarked:

I’m looking at [race] as a social/cultural concept. I also
have a measure of discrimination in the study. I think that’s
how I approached [race]. The differences [noted] could be
coping [strategies] between the two groups [because] they
have different life histories, different perceptions of people
in their communities and outside of [their communities].
So, I think, I really was conceptualizing [race] as social/
cultural variable.

I think the definitions of race sit on the fact that [race is]
a social and political construct [and not] a biological construct. [But] I have to be honest, I struggle with [this] a little
bit because I think that the Human Genome Project and
some of the [research] we are doing [and are part of ] is a
hotly contested debate. Does race have biological markers
that are important for the development of medications,
responses to medications, and development of diseases?
I think right now, we [can] say, ‘No. Race is a social construct.’ I think that’s fair. I just say this is something that I’m
struggling with personally in my mind.

Ethnicity was not seen as biological (76%) but seven
researchers (21%) indicated that ethnicity had biological
components. One researcher chose not to answer.
The theme constructionism was identified to explain the
non-biological meanings ascribed to race and ethnicity.
Respondents were asked if they considered race and ethnicity
a cultural construct, a social construct and a political construct (Table 4). A cultural construct was operationalized as
shared characteristics (e.g. diet and cultural practices)
attributed to specific racial and ethnic groups whereas a
social construct was defined as society’s role in the naming
and assigning of racial and ethnic categories.
98

Most researchers considered race to be culturally constructed (79%), most thought race was socially constructed
(91%), and most felt that race had political implications
(85%). Similarly, most respondents considered ethnicity
to be a cultural construct (94%), most saw ethnicity as a
social construct (97%) and most found that ethnicity had
political aspects (55%). A physician in Atlanta chose not
to answer the questions of whether race and ethnicity are
social, cultural and political constructs because ‘I think
that’s a question that doctors aren’t well set up to answer.’
The cultural and social meanings ascribed to race
and ethnicity by study participants provided a glimpse
into what they believed their research was examining. A
psychologist in Tampa, Florida explained:

Respondents used standardized racial and ethnic
categories in their research designs in order to generalize
their findings and compare health outcomes between or
among racial and ethnic groups. Most of the respondents
used some form of the standardized racial and ethnic
classifications promulgated by the Office of Management
and Budget (OMB). The Office of Management and Budget
Directive 15 specifies five minimum racial categories
(black or African-American, white, Hawaiian or Other Pacific
Islander, American Indian or Alaska Native, Asian) and two
ethnic categories, Hispanic or Latino and Not Hispanic or
Latino (OMB 1997a).
Imposing uniform racial and ethnic classifications
illustrated the acceptance of these racial and ethnic
groupings as well as the malleability of these classifications.

Race and ethnicity in health services research

When a respondent’s study participant reported a racial or
ethnic identity outside of the standardized classifications,
these participants were asked (when possible) to select a racial
or ethnic identity that conformed to a list of standardized
classifications.
Study participants indicating multiple racial and ethnic
identities (e.g. white and black, Puerto Rican and black,
Asian and white) were asked (when possible) to choose a
race or ethnicity they identified with most thus negating a
part of their identity. If study participants did not or could
not choose another racial or ethnic identity, researchers’
either excluded these participants from the study or, in some
cases, the principal investigator or project statistician assigned
a racial or ethnic identity. A researcher in Seattle remarked:
when I get my surveys back, I look at the data. For those who
check off more than one [racial or ethnic identity], we have
group discussions in the research team. I say, ‘Okay, how do
we categorize this person?’

Some researchers rejected dual racial and ethnic identities. These researchers either forced their study participants
to decide on a single racial or ethnic identity or, again, made
the choice for them. A researcher in Portland commented:
We assume that people who identify with any kind of minority
background — let’s say somebody [self-reports] as both
African American and white or Latino and white — we will
assume that being part of a minority group will have some
impact on their [health] experiences. We will put [those
people] in that [minority] group [because we] assume that
the part identifying with the minority group will impact
their experiences [in the health care system].

An epidemiologist in Atlanta provided a rationale for how
this practice originated:
What we’ve done, what our data person has done — per my
instructions — for those individuals who classify themselves
as Black [and something else], then they’re [classified as]
Black. I pretty much use White and Black as [the] predominant
races with Black being the most predominant. So, that took
a slavery mentality an ounce of Black you’re Black.

Quantitative methods predominated (67%) in both
medical and non-medical disciplines. Quantitative methodology,
in particular, requires that category cell sizes are large
enough to generalize about the studied populations, thus,
race or ethnicity become a categorical variable in which binary
relationships exist (e.g. black/white, Hispanic/Non-Hispanic,
white/non-white). A researcher in Nashville stated:
I’m a sociologist … I really believe [that] sociologists like
behavioral scientists get [race and ethnicity in a broader
context] but MDs and epidemiologists do not. They understand black = 0 and white = 1. I hate that. That is not race.
That’s a binary variable.

Quantitative methodology also requires selecting a
reference group in order to make comparisons between or
among racial and ethnic groups. The reference group
most used by researchers was white (45%). A physician from
Boston provided this rationale:
I think the nature of disparities research, for better of
for worse, the white population [is used] as the reference
group. There are plenty of reasons why you could debate
[this practice] and people have argued this in a variety of
ways. I think there are significant sociologic and historical
reasons why, in this country, we use ‘white’ as the reference
group. Issues related to [the] history of our nation [including] power and privilege [of a particular group]. [There
are] a whole set of things even socioeconomic class and the
over-representation of certain groups in certain socioeconomic levels that allow us, at a rough cut, to look at the
white population as a benchmark

Although many respondents used ‘white’ as the reference
group, a physician in San Francisco never considered selecting a different referent group:
Good question. We see [white as] the larger sample. I think
[this practice] probably reflects a convention that has its
own value and assumptions that [white] is a majority population. [This is] a provocative question. [This practice] is
like these things we make assumptions about [but do not]
think [about] the vis-à-vis standard to which you are comparing other things.

The majority of researchers (64%) stated that race and
ethnicity served as a proxy for other variables of interest
such as socioeconomic status, discrimination, marginalization and social stratification. Other respondents (36%)
indicated they had never used racial and ethnic variables
to suggest other concepts. Race and ethnicity, most commonly, served as markers for culture and socioeconomic
factors such as poverty and class. Race and ethnicity were
accepted as legitimate variables to convey socioeconomic
status information when that information was not available.
For those acknowledging that race and ethnicity were
proxy variables, social and cultural meanings were attributed
to them. A sociologist in Los Angeles stated that race and
ethnicity are ‘sometimes a proxy for culture and sometimes
a proxy for social stratification.’ A physician in San Francisco
declared, ‘I think [race] is measuring a status in society
that is defined by one’s culture, one’s language, one’s skin
color, one’s social network.’
Of note, 12 respondents stated that race and ethnicity
were not being used as proxy variables. However, reasons
given for racial and ethnic differences revealed that race and
ethnicity functioned as surrogate markers for social variables
(e.g. prejudice) or cultural variables (e.g. diet). Additionally,
there was a presumption of cultural homogeneity among
racial and ethnic groups thus race and ethnicity did serve as
99

S Moscou

a proxy for a shared experience. A researcher in Seattle did
not believe that race and ethnicity were acting as proxy
variables but when asked what race and ethnicity measured
this respondent disclosed: ‘I think [race] is measuring social
justice. I think [race] is measuring how we feel about each
other, how we have treated each other, and the impact of our
actions towards each other’.
Researchers may hold assumptions about a particular
racial and ethnic group based on their previous research or
from other studies. Presumptions about racial and ethnic
groups (language, insurance status, socioeconomic status
characteristics, biological and physical traits) might be
carried into newer research projects even though the new
participants may be vastly different from previous study
participants.
A researcher in the northeast presupposed that study
participants who self-identified as Hispanic or Latino would
require bilingual services because the researcher’s previous
study participants were older diabetic patients who either
only spoke Spanish or felt more comfortable communicating
in Spanish.
The study participants in this respondent’s current
research project were members of a hospital union. The
respondent remarked, ‘I have my bilingual staff lined up
and ready to go. But, we’re just not finding the Spanish
speakers.’ This researcher’s supposition, however, ignored
that English is required in most employment settings and
that Hispanic or Latino study populations might be quite
different.
Standardized racial and ethnic classifications signify an
internal sameness of culture, biology and behaviour within
the group identified with that label. Racial and ethnic
labelling thus permits assumptions about the prevalence
of particular diseases in specific racial and ethnic groups.
A physician in San Francisco conveyed:
When I’m teaching (medical residents) and a student
presents a case (of) a 50-year-old African-American
with hypertension and whatever — I say, ‘We better check
for diabetes because those tend to go hand in hand.
And, in African-American populations there is a high
prevalence.’

This same physician further noted: ‘Understanding
comorbidity is [an] important cognitive shorthand that is
important to teach. [But] on the other hand, it essentializes
race by saying that — this is something I’m struggling with a
lot in my teaching’. This physician believed that linking
hypertension to other comorbid diseases is a necessary
lesson for medical residents but also recognized there was
a problem with equating a racial identity with a particular
disease. This conflation often leads to generalizations and
100

unwarranted assumptions about genetic backgrounds
and disease processes within a specific racial or ethnic
group.
Multiple studies have shown that racial and ethnic
data in administrative databases are not reliable particularly for non-white individuals and repeatedly fail to
capture the full range of ways in which people selfidentify (Hahn, Mulinare and Teutsch 1992; Pan et al.
1999; Baumeister et al. 2000; Boehmer et al. 2002;
Moscou et al. 2003; Blustein 2005). Although most
respondents acknowledged the problems of using
secondary data, several indicated that there were observable
physical characteristics and cultural indicators such as
language that could lead someone to select the correct
or appropriate racial or ethnic identity. A sociologist in
Oregon observed:
Administrative data collected by the Medicaid program in
Oregon is collected by the caseworker simply selecting a
race. Who knows what they put [down]? I think, they ask
sometimes but I don’t think they always do. Sometimes [the
person’s race and ethnicity] might be really obvious [to the
clerk]. If you know someone speaks Spanish and moved
from Mexico about 2 months ago, sure, that is easy. But,
I think there are other cases when it’s not [obvious]. In
particular, Native Americans or one of those groups that
don’t always appear ethnic.

A physician in Birmingham, Alabama also conceded that
the accuracy of racial and ethnic identities was questionable
but accepted that classifications of African-American were
likely accurate:
We are not entirely sure how accurate (administrative
data) is (in their racial and ethnic classifications). Although
for African-American, for example, I work with a lot of
Medicare claims data and the validity of African-American
(classification) is pretty high. But, the validity for Hispanic
or Native American is not high at all … we do the best we can.

All of the respondents believed that examining racial
and ethnic differences was important; therefore, collecting
racial and ethnic data were necessary. Still, some researchers
questioned the traditional ways that race or ethnicity is
conceptualized and interpreted in health services research.
These respondents acknowledged that their research
findings might play a role in racial and ethnic stereotyping.
A medical anthropologist in Arkansas noted:
I worry about [how my research is interpreted]. In fact, I’m
afraid [that] I’m stereotyping. I’m afraid people will sometimes interpret, for example, if I say African-Americans [did
this] or in this study [African-Americans] were this then
everybody says, ‘Oh, all African-American women are like
this.’ Even in the South with the populations that we have,
I can’t really generalize. I can’t say all African-Americans in
East Arkansas are poor even though a good proportion of

Race and ethnicity in health services research

them are either poor or live in a poor culture. I do worry
that folks are being stereotyped and that [my findings] lead
to these stereotypes. I really try to provide a cultural variable
[to explain the disparity] rather than the ethnic variable
because a lot of this stuff is particular to the South.

Rather than routinely include race and ethnicity in
research designs, some respondents attempted to tease out
other variables that might be responsible for the disparity.
These researchers used alternative variables to examine
the studied problem, thus, eliminating race or ethnicity as
the explanatory variable. A researcher in Santa Monica,
California was able to tease out a physical trait that contributed to birth outcome differentials. The researcher
commented that using the variable ‘stature’ instead of
race did not perpetuate unwarranted presumptions about a
particular racial group. ‘We actually had a different way of
being able to look at the mom’s stature, which was the thing
we were trying to control for instead of assuming that all
Asians were small’.
A researcher in New York City recognized the secondary
consequences of using racial and ethnic variables in
ways that reinforce assumptions and stereotypes about
communities of colour. When race and ethnicity are ‘in the
equation’, minorities are often viewed in unflattering ways
and the social analysis of why the disparity exists or persists
is frequently missing.
When another researcher re-analyzed this respondent’s
data from an earlier study, this respondent found the reanalysis more informative and instructive. The respondent
remarked:
The damage that had been done by identifying the problem
with [a specific] racial/ethnic group was bad enough. We
wrote pretty much the following [after the study appeared]:
good research and regret the fallout that it caused. [We]
feel that [the reanalysis of our study] published in this issue
of [ journal X] does a much better job of moving the analysis
forward because it takes race/ethnicity out of the equation
and focuses much more on neighborhood dynamics, which
is at this point in our research efforts, we feel is really going
on [and causing the differentials seen in disease prevalence].

DISCUSSION
The goal of this descriptive study was to ascertain how
health services researchers conceptualize and operationalize
race and ethnicity. The broad findings showed that race and
ethnicity held several meanings for respondents but the
subtext of race and ethnicity (biological, social, cultural or
political) was often unclear because race and ethnicity were
seldom operationalized in the original research design.
The interviews illustrated that respondents believed that
race and ethnicity held biological and social meanings.
© 2008 The author. Journal compilation © 2008 Blackwell Publishing Ltd

Although one-third of the respondents stated that race was
biological, eight respondents acknowledged dual definitions
of race (biological and social). Ethnicity, on the other hand,
was overwhelmingly accepted as non-biological and was seen
to represent the cultural aspects of specific racial and ethnic
groups. Most of the respondents indicated that ethnicity
held cultural, social and political meanings.
The large number of respondents who believed that
racial and ethnic variables denoted biological and
genetic properties was surprising and such views contribute
to the often unwavering belief that race and ethnicity
(particularly in clinical medicine) are useful in treatment
regimens and necessary variables in health services and
clinical research.
Racial and ethnic labelling did carry assumptions about
the groups studied because these classifications implied a
commonality within a given group in terms of culture, cultural
practices and biology. Additionally, biased beliefs or
assumptions about racial and ethnic groups were embedded,
often uncritically, in the limited number of racial and
ethnic labels used, the analytical framework used (quantitative or qualitative), and the educational discipline of the
researcher (clinician or non-clinician).
Researchers’ tendency to imbue race and ethnicity
with social or biological characteristics depended on the
researcher’s discipline, whether or not they were clinicians,
and their areas of interest. Non-clinicians, for example, were
much clearer that race signified culture whereas clinical
researchers, for the most part, appeared reluctant to dismiss
that race captured on some level meaningful biological
differences (this belief might grow stronger as funding
increases for pharmacogenomic research).
The majority of respondents were cognizant that
quantitative methods identified the problem but not
why it exists (or persists). As one respondent noted, if class,
neighbourhood characteristics or other social variables
were examined instead of racial and ethnic variables,
explanations for racial and ethnic differences in health
outcomes would become clearer. Additionally, health
policies would begin to correct the identified social
problem (e.g. housing, education, transportation, clinical
facilities and economic deprivation) often responsible for
the reported disparity.
Race and ethnicity, although this was not always recognized,
did serve as proxy measurements for such powerful social
variables as discrimination, marginalization and economic
resources. Moreover, race and ethnicity were often used to
convey class factors (income groupings), social behaviours
(risk factors), and social conditions (advantages and
disadvantages).
101

S Moscou

Race and ethnicity for many respondents were substitutes for social standing (e.g. lower income, upper income
and educated), residential locations (poor or wealthy
communities), markers for different life experiences, access
and barriers to health-care and treatment, and social stratification (economic advantages or disadvantages). Because
race and ethnicity have become synonymous with socioeconomic factors, health services researchers accepted
that racial and ethnic variables, although imperfect, could
stand-in for class or socioeconomic status. However, social
variables (often unstudied but speculated about) were driving
many of the racial and ethnic health disparities noted.
Measuring race and ethnicity quantitatively imposed
racial and ethnic stratification thus reinforced the belief
that groups assigned a racialized identity were different.
Enforcing a standardized racial and ethnic identity obliged
the researchers’ study participants to conform to classification schemes that rendered some racialized identities
invisible, negated some racialized identities or excluded
some racialized identities from the research project.
Negating, changing or removing study participants’
racialized identities skews research findings towards a
homogenous and false conclusion. Moreover, this practice
maintains the hegemony of the researcher to ignore how
study participants conceptualize race and ethnicity and
the importance or non-importance of racial and ethnic
identities for these participants.
Quantitative methods required choosing a referent
group to make comparisons between or among racial and
ethnic groups. The reference group most often used by
researchers in this study was white. Reasons given for
using this group were (i) whites were more likely to be
insured, (ii) whites were more likely to have access to healthcare, (iii) whites were less likely to encounter access barriers,
and (iv) whites as the comparison group are standard
research practice.
Whereas this is often true for well-insured white study
participants, it does not necessarily hold for uninsured or
underinsured whites who are more likely to have health outcomes similar to others in their class. By ignoring poor or
lower-income whites, researchers often rendered this group
invisible in health services studies. Furthermore, defining
and accepting a particular class of whites as the default
referent group maintains the structural hierarchy of the
dominant class, subordinates minority group members, and
constructs a framework that obscures the reality that poor
whites have similar health outcomes to poor non-whites.
Respondents did recognize the inherent problems with
racial and ethnic variables (measurement concerns, impreciseness, misclassifications and narrow racial and ethnic
102

categories) but the majority still believed these variables
were necessary and therefore useful to their research. There
was an a priori assumption that racial and ethnic variables
could discern the reasons for health disparities. This belief
reinforced the conviction that racial and ethnic variables
had explanatory power despite the reality that race and
ethnicity often served as proxies for social factors such as
inequality and marginalization and socioeconomic factors
such as poverty and class.
Several researchers understood that racial and ethnic
variables were often used in ways that contributed to
reductive and flawed interpretations of racial and ethnic
differences. These respondents noted that racial and ethnic
variables often engendered simplistic comparisons, ignored
the vast differences within the populations studied, and
did not address the social problems responsible for many
disparities in health outcomes.
These researchers by using other variables of interest
had begun to think about factors other than race and ethnicity
that might create differentials in health outcomes. Additionally, shifting the analysis beyond racial and ethnic labelling
held the possibility that policy solutions and clinical interventions would be addressed within the broader societal
milieu rather than in the realm of the ‘problematic’
individual racial or ethnic group.
The findings demonstrated that a better paradigm is
needed to recognize how a racialized identity may or may not
contribute to health disparities. The existing paradigms
that race and ethnicity equal biology or race and ethnicity
equal culture and lifestyle is incapable of advancing
knowledge about the nature of racial and ethnic disparities
because the effect of social stratification is missed in these
models.

CONCLUSION
Racial classifications have a long history of negative social
consequences for those identified as non-white (segregation,
economic deprivation, racism and inequality), are without
biological merit and provide inconsequential information.
Little is accomplished by using these classifications other
than legitimizing unscientific categories ‘full of evil social
import’; therefore, some have suggested it is time to
abandon racial variables (Fullilove 1998, 1297; Buehler
1999; Cooper, Kaufman and Ward 2003).
Given that the collection of racial and ethnic data will
continue, a paradigmatic shift in the knowledge production of race and ethnicity is needed. Additionally, learning
the consequences of racialization (identity assignment)
calls for a theoretical framework that deconstructs (and

Race and ethnicity in health services research

reconstructs) the mechanisms, practices and social relations
responsible for producing and reproducing racial and
ethnic inequalities in health outcomes.
Critical race theory (CRT) offers a conceptual framework in which to examine the social relations of race and
ethnicity and their effects on health outcomes seen in
groups assigned a particular racial or ethnic identity.
CRT identifies the interplay (or hidden correlation) of race
and ethnicity in relation to one’s economic status, living
conditions (e.g. neighbourhood characteristics) and social
status (e.g. educational level).
CRT emerged from legal studies as an approach to
examine race within the economic, social and political
dimensions of the legal system (Ladson-Billings 1998).
Using the lens of perspectivism (contextualization of oppression for a particular person at a particular time and place),
critical race theorists analyze the myths, presuppositions
and conventional wisdom about race by collecting stories
and narratives of those with limited power and privilege
(Matsuda 1992; Delgado and Stefanic 2001).
CRT moves the prevailing scientific paradigm of
studying (and controlling for) phenotypic characteristics
(racial categories) and the dominant methodological
lens (quantitative) towards a scientific paradigm that can
analyze the structural dimensions (racism, classism and
sexism) that are contributory factors in racial and ethnic
disparities in health.
CRT shifts the existing scientific paradigm (predictive
explanation models) by applying a qualitative methodological
lens that can bring about a contextualized understanding of
the interplay of race and health or ethnicity and health that
is currently missing from disparity research.
Integrating CRT into the study of racial and ethnic
disparities provides a rigorous conceptual framework that
will allow health services researchers to investigate the
effects of a racially stratified society (socially structural
arrangements) on an individual, particular racial or ethnic
group, or a community’s health. Furthermore, applying
a critical race theoretical framework when investigating
racial and ethnic disparities encourages the researcher to
consider the scientific, clinical, social and ethical ramifications inherent in the design and implementation of
race-based research.
Race and ethnicity have a role in health services research
because as long as racial inequality exists, the research
community is obligated to bear witness to and redress
health disparities. However, by ignoring the complicated
relationship between race and health or ethnicity and
health, we continue with a one-dimensional analysis that
limits our knowledge about the existence and persistence
© 2008 The author. Journal compilation © 2008 Blackwell Publishing Ltd

of disparities in health outcomes. Without contextualizing
research participants’ racialized identities, race and ethnicity
continue to be proxies for biological or cultural variations.
Examining the effect(s) of race or ethnicity on health within
a critical race theoretical framework broadens our analysis
(as well our discussion) of racial and ethnic disparities
beyond biological and cultural reductionism.

ACKNOWLEDGEMENTS
This article would not have come to fruition without the
contribution of the following individuals: Sarita Bhalotra,
PhD, Jon Chilingerian, PhD (Heller School at Brandeis
University), Vanessa Calderon-Rosado, PhD (IBA, Boston, MA),
Barbara Katz-Rothman, PhD (Baruch College & Graduate
Center at CUNY), Judith B. Kaplan, MS, and Sue Pfefferle, PhD.

REFERENCES
Alland A Jr. 2002. Race in mind: Race, IQ, and other racisms.
New York: Palgrave Macmillan.
Anderson MR, S Moscou, C Fulchon and DR Neuspiel. 2001.
The role of race in the clinical presentation. Family
Medicine 33: 430–4.
Baumeister L, K Marchi, M Pearl, R Williams and P
Braveman. 2000. The validity of information on ‘race’
and ‘Hispanic ethnicity’ in California birth certificate
data. Health Services Research 35: 869–83.
Bernard HR. 2000. Social research methods. Thousand Oaks,
CA: Sage.
Bhopal R and L Donaldson. 1998. White, European, Western,
Caucasian, or what? Inappropriate labeling in research
on race, ethnicity, and health. American Journal of Public
Health 88: 1303–7.
Bhopal R and J Rankin. 1999. Concepts and terminology in
ethnicity, race and health: Be aware of the ongoing
debate. British Dentistry Journal 186: 483–4.
Bhopal R. 2002. Revisiting race/ethnicity as a variable in
health research. American Journal of Public Health 92:
156–7.
Blustein J. 2005. The reliability of racial classifications in
hospital discharge abstract data. American Journal of
Public Health 84: 1018–21.
Boehmer U, NR Kressin, D Berlowitz, C Christiansen, L
Kazis and J Jones. 2002. Self-reported vs. administrative
race/ethnicity data and study results. American Journal of
Public Health 92: 1471–3.
Bonilla-Silva E. 1996. Rethinking racism: Toward a structural interpretation. American Sociological Review 62:
465–80.
103

S Moscou

Buehler JW. 1999. Abandoning race as a variable in public
health research. American Journal of Public Health 89: 783.
Buescher PA, Z Gizlice and KA Jones-Vessey. 2005. Discrepancies between published data on racial classification
and self-reported race: Evidence from the 2002 North
Carolina live birth records. Public Health Reports 120:
393–8.
Cooper RS, JS Kaufman and R Ward. 2003. Race and
genomics. New England Journal of Medicine 348: 1166–70.
Culebras A. 1995. Hispanic: An epidemiologically meaningless term. Archives of Neurology 52: 533–4.
Delgado R and J Stefanic. 2001. Critical race theory: An
introduction. New York: New York University Press.
Drevdahl DJ, DA Phillips and JY Taylor. 2006. Uncontested
categories: The use of race and ethnicity in nursing
research. Nursing Inquiry 13: 52–63.
Fausto-Sterling A. 2004. Refashioning race: DNA and the
politics of health care. Differences 15: 1–37.
Fullilove MT. 1998. Comment: Abandoning ‘race’ as a
variable in public health research: An idea whose time
has come. American Journal of Public Health 88: 1297–8.
Goldberg D. 1992. The semantics of race. Ethnic and Racial
Studies 14: 543–67.
Gonzalez-Burchard E, E Ziv, N Coyle, SL Gomez, H Tang,
AJ Karter, JL Mountain, EJ Perez-Stable and N Risch.
2003. The importance of race and ethnic background
in biomedical research and clinical practice. New England
Journal of Medicine 348: 1170–5.
Graves JLJ. 2001. The emperor’s new clothes: Biology theories of race
at the millennium. New Brunswick, NJ: Rutgers University
Press.
Gunarathnam Y. 2003. Researching ‘race’ and ethnicity: Methods,
knowledge, and power. London: Sage.
Hahn RA. 1992. The state of federal health statistics on racial
and ethnic groups. Journal of American Medical Association
267: 268–71.
Hahn RA. 1999. Why race is differentially classified on US
birth and infant death certificates: An examination of
two hypotheses. Epidemiology 10: 108–11.
Hahn RA, J Mulinare and SM Teutsch. 1992. Inconsistencies in coding race and ethnicity between birth and
death in US infants: A new look at infant mortality
1983–85. Journal of American Medical Association 267:
259–63.
Henry GT. 1990. Practical sampling. Thousand Oaks, CA: Sage.
Jones CP, TA LaVeist and M Lillie-Blanton. 1991. ‘Race’ in
the epidemiologic literature: An examination of the
American Journal of Epidemiology, 1921–90. American Journal of Epidemiology 134: 1079–84.
Kaplan JB and T Bennett. 2003. Use of race and ethnicity in
104

biomedical publication. Journal of American Medical Association 289: 2709–16.
Krieger N. 2005. Stormy weather: Race, gene expression,
and the science of health disparities. American Journal of
Public Health 95: 2155–60.
Ladson-Billings G. 1998. Just what is critical race theory and
what’s it doing in a nice field like education? Qualitative
Studies in Education 11: 7–24.
LaVeist TA. 1996. Why we should continue to study race but
do a better job: An essay on race, racism, and health.
Ethnicity and Disease 6: 21–9.
Laws MB. 2001. Race and ethnicity in biomedical and health
services research. Archives of Pediatric and Adolescent
Medicine 155: 972.
Lieberman L, RE Hampton, A Littlefield and G Hallead.
1992. Race in biology and anthropology: A study of
college texts and professors. Journal of Research in Science
Teaching 29: 301–21.
Ma IWY, NA Khan, A Kang, N Zalunardo and A Palepu.
2007. Systematic review identified suboptimal reporting
and use of race/ethnicity in general medical journals.
Journal of Clinical Epidemiology 60: 572–8.
Marks J. 2005. The realities of races. Social science research council.
http://raceandgenomics.ssrc.org/Marks/ (accessed
7 January 2006).
Matsuda MJ. 1992. When the first quail calls: Multiple
consciousness as jurisprudential method. Women’s Rights
Law Reporter 14: 297–300.
McKenney NR and CE Bennett. 1994. Issues regarding data
on race and ethnicity: The census bureau experience.
Public Health Reports 109: 16–20.
Morning A 2004. The nature of race: Teaching and learning
about human difference, PhD Sociology, Princeton
University.
Moscou S, MR Anderson, JB Kaplan and L Valencia.
2003. Validity of racial/ethnic classifications in medical
records data: An exploratory study. American Journal of
Public Health 93: 1084–6.
Nobles M. 2000. History counts: A comparative analysis
of racial/color categorization in US and Brazilian
censuses. American Journal of Public Health 90: 1738–45.
Office of Management and Budget. 1997a. Revisions to the
standards for classification of federal data on race and ethnicity.
http://www.whitehouse.gov/OMB/fedreg/ombdir15.
html (accessed 20 November 2005).
Office of Management and Budget. 1997b. Standards for
maintaining, collecting, and presenting federal data on race and
ethnicity. http://www.doi.gov/diversity/doc/racedata.htm.
Omi M and H Winant. 1994. Racial formation in the United
States: From the 1960s to the 1990s. New York: Routledge.

Race and ethnicity in health services research

Outram S and G Ellison. 2005. Anthropological insights into
the use of race/ethnicity to explore genetic contributions to disparities in health. Journal of Biosocial Science
38: 1–20.
Pan CX, RJ Glynn, H Mogun and I Choodnovskiy. 1999.
Definition of race and ethnicity in older people in
Medicare and Medicaid. Journal of American Geriatric
Society 47: 730–3.
Perot RT and M Youdelman. 2001. Racial, ethnic and
primary language data collection in the health system:
An assessment of federal policies and practices. Commonwealth Fund (Report No. 9/01).
Rivara FP and L Finberg. 2001. Use of the terms race and
ethnicity. Archives of Pediatric and Adolescent Medicine 155:
119.
Schulman KA, E Rubenstein, FD Chesley and JM Eisenberg.
1995. The roles of race and socioeconomic factors in
health services research. Health Services Research 30: 179–
95.

Stanfield JH. 1993. Methodological reflections: An introduction and epistemological considerations. In Race and
ethnicity in research methods, eds JH Stanfield II and RM
Dennis, 3–35. Beverly Hills, CA: Sage.
Trost JE. 1986. Statistically nonrepresentative stratified
sampling: A sampling technique for qualitative studies.
Qualitative Sociology 9: 54–7.
Waldenstrom J. 1990. Disease, race, geography, and genes.
Journal of International Medicine 228: 419–24.
Williams DR. 1994. The concept of race in health services
research: 1966–90. Health Services Research 29: 261–74.
Williams DR. 1999. Race, socioeconomic status, and health:
The added effects of racism and discrimination. Annals
of the New York Academy of Sciences 896: 173–88.
Winker MA. 2006. Race and ethnicity in medical research:
Requirements meet reality. Journal of Law, Medicine, and
Ethics 34: 520–5.
Zuberi T. 2001. Thicker than blood: How racial statistics lie.
Minneapolis: University of Minnesota Press.

105