September 7, 2011

The United States has long encompassed a large number of ethnicities, races, and cultures. This appears to be increasingly true. The country’s recognition of this diversity and its recognition of the considerable strength that such variety brings to the country and its peoples also appear to be increasing. Many other nations are also having similar experiences of increasing diversity among people and a movement away from earlier tendencies to devalue cultural differences and to keep cultural and ethnic groups segregated. Diversity, and indeed even celebration of diverse ethnicities and other human characteristics, enriches a given culture, bringing a greater breadth of perspectives, ideas, values, attitudes, and contributions overall. The world over the later half of the 20th century has increasingly recognized, although certainly not fully accepted, the value of cultural and ethnic differences. This is true in spite of the fact that so-called ethnic strengths historically have been built upon by maintaining similar values and characteristics within a given ethnic or cultural group. However, increasing diversity may also lead to a blending of ethnic differences, thus risking the loss of unique aspects of various cultures and ethnicities over time.

This said, the complexity of the concept of ethnicity and its closely related concepts of race and culture are briefly explored below. First, ethnicity is “defined” and related to other concepts. Then, the observed relationship of ethnicity to some basic health status variables in the United States is outlined. Finally, some of the difficulties in the measurement of ethnicity and problems associated with the concept itself are discussed. At the end, key references are provided from which much of this material has been drawn and which will provide the interested reader many more details.

“Ethnic” usually refers to characteristics of a people, especially a group of people who share a distinct and common set of characteristics, such as culture, religion, language, race, or nationality. In comparison, “race” generally refers to biological factors that are the basis of group differences, especially observable physical features. “Culture” usually refers to shared elements that provide a basis for perceiving, believing, communicating, evaluating, and behaving within a common context. However, no consensus exists regarding precise definitions of ethnicity or race or culture, and it is very common (unfortunately) for people in general, and researchers in particular, to use the terms ethnicity, race, and culture interchangeably. The extensive list of possible central characteristics that may be used to define an ethnic group hints at the complexity of the concept itself and the varied ways in which it is used.

For many in the United States, ethnic group also connotes “minority group,” that is, a smaller group within a larger dominant group. This, of course, is an ethnocentric application of the concept—wherein it is used to refer to those outside the larger group, but is not used to refer to those among the majority group. Furthermore, ethnicity often connotes “race” or “nationality” to many people, particularly so-called races or nationalities that are not among the majority population. Thus, by this usage in the United States, whites or Caucasians or non-Hispanic whites are not ethnic groups in the minds of most people, but Asians, blacks or African Americans, Hispanics or Latinos, American Indians and Alaskan Natives, and Pacific Islanders are— primarily because none of those latter groups represent a majority. In parts of various U.S. states, whites are not in the majority—for example, in the state of New Mexico, no traditionally labeled group represents a majority (Hispanics 44%, non-Hispanic whites 44%, Americans Indians 9%, and other groups about 3%). The absurdity of this ethnocentric usage and meaning of ethnicity as commonly applied in the United States thus becomes apparent.

More objectively, any group sharing a distinct and common set of characteristics would be considered an ethnic group, irrespective of its numbers relative to other groups, including so-called whites. But this suggests an additional problem in assessing and utilizing ethnicity/race as a means by which to predict or classify groups of people—whites, Hispanics, blacks, Native Americans, and other large groups are not homogeneous regarding culture, religion, language, or nationality— in fact, it is clear that each “group” is quite diverse in all these respects. Thus, although ethnicity is usually measured by simple broad self-labels, it is actually based upon a complex concept that reflects many dimensions.

In the United States, the assessment of ethnicity is usually done by simply asking a person in an interview or survey how they describe themselves. Historically, a variety of lists of ethnic groups have been used and have changed somewhat over time. Currently, the following list is increasing in use: Asians, blacks or African Americans, Native Americans (or American Indians) and Alaskan Natives, Pacific Islanders, white or Caucasian, and “Mixed” or “Other,” this latter category recognizing that many people do not distinctly fall into one of the others. Note that among that list, Hispanic is omitted because it is now recognized that “Hispanic” is a separate dimension relative to the larger list, which largely reflects what is usually described as “race.” In this fairly recent assessment scheme, Hispanic status is measured with a separate question that precedes the one above, thus recognizing that Hispanics or Latinos or those of Spanish descent may be among white or black “races” and implicitly that Hispanics are not a homogeneous group. Of course, Hispanics are not homogeneous in many nonracial senses as well because Hispanics who “originate” from Spain, Mexico, the Caribbean, South America, and the United States (where some count their heritage back hundreds of years) often have quite distinct cultures and other characteristics.

The current U.S. categories used to assess “ethnicity” emanate from the U.S. Federal Office of Management and Budget, which announced this system in 1998 and encouraged other federal agencies to follow it for the sake of consistency in reporting. The Department of Health and Human Services and the U.S. Census Bureau among others have adopted this twoquestion approach to ethnicity. Those who use it, often to meet federal standards, are encouraged by governmental guidelines to consider making such ethnicity assessments at a finer grained level, but they are not required to do so. This too suggests that the sociopolitical systems that have evolved this scheme also recognize that ethnicity is more complex than what the simple categories or labels listed above imply.

Assessing ethnicity (or race) in a population usually serves one of two major purposes. The first is descriptive, wherein the percentage of people that can be attributed to a particular broad ethnic group is simply provided as information to characterize a population. The second is analytic, wherein differences in nonethnic characteristics or behaviors are reported to vary as a function of membership in an ethnic group. Below we will discuss some of the many problems with the analytic use of ethnicity, but first we will turn to a brief and simple review of the known general relationships of ethnicity to several major health status measures in the United States to illustrate the analytic use of ethnicity.


Health Status Indicators (HSIs) were developed as a part of Healthy People 2000, a set of Department of

Health and Human Services objectives for the United States. A central goal for these objectives was to help reduce disparities in health care among various ethnic groups in the United States and to encourage significant improvement in the health indicators for the population overall. In addition to reporting the rates of various illnesses, an index of disparity was used to summarize ethnic/racial differences in the HSIs. Examination of trends for the period 1990-1998 showed that most of the 17 HSIs improved for most ethnic/racial groups, although the differences between ethnic groups did not change very much.

Infant mortality is often used as a principal measure of health status among groups and nations worldwide. In the United States, rates of infant mortality have long been much lower for non-Hispanic whites, Hispanics, and Asian/Pacific Islanders than for blacks and American Indians/Alaskan Natives, roughly by a factor of two times. During the 1990s, infant mortality decreased notably for all groups to 6.0/1,000 for whites, 5.8/1,000 for Hispanics, and 6.6/1,000 for Asians, and although declining by roughly 25%, remained much higher for black non-Hispanics (13.9) and American Indians/Alaskan Natives (9.3). Rates for low birthweight, however, showed a different trend, increasing by as much as 18% for some groups over the decade—whites (6.6/1,000), Hispanics (6.4), Asians (7.4), blacks (13.2), and American Indians/Alaskan Natives (6.8). The percentages of women with no prenatal care during their first trimester of pregnancy in 1998 showed marked declines of 24-35% compared to 1990.

The total death rate is also used as a prime HSI nationally and globally. The total death rate decreased over the 1990s roughly 10% for all groups except American Indians/Alaskan Natives, who showed a 4% increase in death rate between 1990 and 1998—whites (453/100,000), Hispanics (343), Asians (265), blacks (711), and American Indians/Alaskan Natives (458). The ratio of rates between highest and lowest groups (the disparity ratio) was 2.7 in 1998, the same as it was in 1990, indicating no reduction in overall health differences among groups.

Death rates by “violent” means showed substantial overall declines during the decade of the 1990s—homicide (28%), suicide (10%), and motor vehicle crash (15%). Percentages of decline across ethnic groups were roughly similar with two major exceptions: (a) the rate of suicide for American Indians/Alaskan Natives actually increased 8%, and the rates of decline of death by homicide (11%) and motor vehicle crashes (4%) were much less than for the other groups; and (b) Asians showed a much lower decline in suicide rate (2%) than other groups, probably because their rate was already the lowest by far. The disparity ratio actually increased over the period for motor vehicle crash (2.6-3.7) and suicide (2.1-2.3), but it declined for homicide (9.7-8.2).

Death rates by major diseases are also used as important indicators of health status in the United States. For example, rates of death from heart disease declined overall by 16% over the 1990s, and this decline was uniform for all ethnic groups, except it was notably lower for blacks (11%) and American Indians/Alaskan Natives (8%). The disparity ratio among groups actually increased slightly from 2.7 to 2.8. The decline in stroke death rate was much lower overall (9%) than for heart disease and was not uniform among groups, with American Indians/Alaskan Natives actually increasing 3%. The disparity ratio was 2.2, down from 2.5 in 1990.

A number of other trends in major HSIs can be found that illustrate the substantial differences in health among the various major ethnic/racial groups in the United States. Comparison of disparity indices in 1990 to those in 1998 reveal that for 11 of 17 HSIs, ethnic difference has shown a decline, but statistically significantly so for only 6 of the indicators, while 5 indicators showed increases, 3 of them significantly. Thus, overall little decrease in health differences seems to have occurred over the decade of the 1990s, and for some groups on some measures, notable increases were revealed, particularly for American Indians/Alaskan Natives.


Turning back to ethnicity/race as a means to measure a concept, much debate has focused on the appropriateness of using self-reported ethnicity or race as a variable to predict or explain differences in health and other outcomes. A substantial scientific literature over the past two decades discusses the problems surrounding the use of “ethnicity” as an independent variable (i.e., a characteristic that influences other characteristics, but is not itself affected by those characteristics). A consensus among researchers is not apparent, but ethnicity (and gender) continue to be used as if they are potential causes in a wide range of health outcomes. This, in spite of the fact that many experts in scientific research methods agree that measuring and using ethnicity for this purpose may produce nothing more than descriptive results at best, and onereous ones at worst. Some argue that ethnicity and race are simply sociopolitical concepts that have little, if any, basis in scientific reality. Others, however, argue that underlying genetic differences exist among ethnic and racial groups (and other groups) and that these genetic differences may well be important factors that contribute to risks for both mental and physical illnesses. Knowledge about such factors might contribute to more effective diagnosis and better treatments. Still others argue that even if group genetic differences do matter, cultural and social differences between ethnic groups contribute greatly to behaviors that are causes of or associated with a variety of health factors.

It has often been argued that although genetic, behavioral, social, and cultural differences among ethnic groups may be predictors of average health status, it is much more important to examine differences within the ethnic group than to examine the differences between ethnic groups to gain an understanding of the health and illness and to increase effectiveness of diagnosis and treatment. Additional problems with using ethnicity to explain or predict health are substantial. Researchers often assume that individuals in an ethnic group all share some common characteristic associated with culture, and that the cultural characteristic is associated with mental or physical ill health. Two major problems are apparent with this thinking. First, as previously discussed, there is considerable variation among individuals within an ethnic group on almost all characteristics. Second, ethnicity usually serves as a substitute for some other concept of true interest, such as culture, and in particular specific features of culture. Thus, ethnicity is often used as a substitute measure for culture or attitudes or behaviors. These are usually much more difficult to measure, and thus researchers simply assess ethnicity instead.

The measurement of the variable that is directly associated with the outcome is easier to defend scientifically. For example, if a researcher were studying use of birth control and determined that those who identified themselves as “Hispanic” were significantly less likely to use birth control pills, it would be scientifically imprecise (some methodologists and theoreticians would say flat wrong) to say that being a member of an ethnic group “causes” use of specific birth control methods. A variable that might be closer to the outcome (here, use of a specific birth control method) might be “religious preference” because Hispanics predominantly identify themselves as Catholics, and the Catholic formal doctrine forbids use of birth control pills. But “religious preference” would still be a substitute variable since many Hispanics may express a Catholic religious preference, but not equally hold to the church doctrine about birth control, and thus use birth control pills. Probably closer characteristics of pill use would be attitudes about birth control or acceptance of Catholic religious doctrine or prior experiences with birth control or the lack of experience with birth control methods. Measures of these might well show much better ability to predict birth control use than simply having the status of “Hispanic ethnicity.” Hence, rather than just using “ethnicity,” researchers should carefully think about the likely causes of the outcome they are studying and measure those characteristics that are the most directly associated with the outcome, if possible. Commonly, that means measuring past behaviors or current attitudes or beliefs fairly directly rather than simply assessing ethnic group status.

Another serious problem in the assessment of ethnicity involves how to categorize individuals who are of “mixed” ethnicity or individuals who are not aware of their “full ethnicity.” If the characteristic that was most closely associated with the outcome under study was assessed instead of the substitute characteristic ethnicity, this problem would disappear. Frequently the problem of great diversity within ethnic groups is compounded by researchers who “homogenize” so-called minority groups by comparing the responses of all minority ethnic groups to whites, as if all members of all minority ethnic groups share something in common.

Another problem is that in many research studies that detect differences across various ethnic groups, the ethnic groups vary in many ways other than underlying culture or attitudes or beliefs. For example, they differ in educational level, income level, age distribution, language fluency, general acculturation, and many possibly unknown ways. Researchers often attempt to “control” for such differences using statistics, but serious problems exist for interpreting such analyses that “equate” groups using various covariates. Simply put, real differences in groups cannot be meaningfully eliminated using abstract mathematical “corrections.” Thus, nonequivalent ethnic groups are different in many ways, and statistical controls cannot disentangle such differences in any clear manner.

Recommendations for the application of several guidelines for using ethnicity in research have been made by many authors: (a) make clear the assumptions that are the basis for the use and assessment of ethnicity in a particular context; (b) test specific hypotheses about specific aspects of culture or other characteristics of ethnicity rather than using ethnicity as a substitute variable; (c) consider matching samples of different ethnic groups selected for study while retaining as much diversity within the group; (d) fully report in scientific manuscripts the sample characteristics and sampling methodology used in studies; (e) use sample sizes large enough to adequately detect the differences that are likely to be found in naturally occurring groups; (f) use several measures and several assessment methods, where feasible, to be sure that the concept being measured is actually the causal factor being studied (i.e., convergent validity); (g) work with cultural/ethnic experts to ensure appropriate translation of language and concepts of the measures being used; and (h) use study results to generate further research rather than assume findings are valid. Thus, use of the concept of ethnicity should entail careful thinking and planning to enable the collection of data of the highest quality and that most directly speak to the research questions.

It is clear that various “ethnic groups” as commonly assessed differ on many characteristics, including many measures of health status, but it is also true that such groups differ on many other characteristics, such as income, education, language use, general acculturation, attitudes, beliefs, and values, among many others. It seems unwise to attribute differences, such as differences in health status, purely to characteristics of ethnic status in any simple or direct manner. Different ethnic groups for many reasons experience the world in different ways, which are likely in turn to lead to differences among and within ethnic groups in many complex ways. Indeed, different ethnic groups express different cultures, and it is the more direct study of specific aspects of culture that may lead to greater understanding of the differential ethnic experiences, rather than stereotypically treating all members of the same ethnic group as the same. That is, it is not only important to recognize the diversity among the many ethnic groups in the United States and the world, but also to recognize the immense diversity with each of those groups and to attempt to understand how the greater diversity may or may not contribute to variation in health among individuals.

SEE ALSO: Acculturation, African American, Asian and Pacific Islander, Birth control, Cardiovascular disease, Education, Latinos, Socioeconomic status

Suggested Reading

  • Alvidrez, J., & Arean, P. A. (2002). Psychosocial treatment research with ethnic minority populations: Ethical considerations in conducting clinical trials. Ethics and Behavior, 12, 103—116.
  • Burchard, E. G., Ziv, E., Coyle, N., Gomez, S. L., Tang, H., Karter, A. J., et al. (2003). The importance of race and ethnic background in biomedical research and clinical practice. New England Journal of Medicine, 348, 1170-1175.
  • Cooper, R., Kaufman, J. S., & Ward, R. (2003). Race and genomics. New England Journal of Medicine, 348, 1166-1170.
  • Keppel, K. G., Pearcy, J. N., & Wagener, D. K. (2002). Trends in racial and ethnic-specific rates for the health status indicators: United States, 1990-1998. Healthy People Statistical Notes, 23, 1-16.
  • Nazroo, J. Y. (2003). The structuring of ethnic inequalities in health: Economic position, racial discrimination, and racism. American Journal of Public Health, 93, 277-284.
  • Okazaki, S., & Sue, S. (1995). Methodological issues in assessment research with ethnic minorities. Psychological Assessment, 7, 367-375.
  • Phinney, J. (1996). When we talk about American ethnic groups what do we mean? The American Psychologist, 51, 918-927.
  • Triandis, H. C. (1996). The psychological measurement of cultural syndromes. The American Psychologist, 51, 407-415.
  • U.S. Department of Health and Human Services. (2001). Mental health: Culture, race, and ethnicity. Washington, DC: Author.

Tags: , ,

Category: E