SEARCH

SEARCH BY CITATION

Keywords:

  • African-American;
  • cancer;
  • race;
  • research

Abstract

  1. Top of page
  2. Abstract
  3. A Brief Historical Perspective
  4. Accurate Racial Categories?
  5. Priorities
  6. Opportunities
  7. An Alternative
  8. Recommendations
  9. Acknowledgements
  10. REFERENCES

This essay questions the appropriateness of racial categories in breast cancer research and recommends the discontinuation of “African-American” as a valid racial category in breast cancer research until better categories can be developed. Cancer 2003;97(1 Suppl):335–41. © 2003 American Cancer Society.

DOI 10.1002/cncr.11024

“Race must be acknowledged as an unstable, shifting, and strategic construction.” Higginbotham EB. J Women in Culture Soc. 1992;17:251–274.

Attendees at the Summit Meeting on Breast Cancer among African-American Women met to discuss several themes.1 Data accuracy (deviation from the actual [true] value) was one of the themes the planners emphasized.2 However, implicit in the majority of discussions were a priori assumptions that “African-American” is a valid racial category and that African-American women with breast cancer have unique, measurable qualities that are identified accurately. This short essay, although it is not definitive, considers several research inferences implicit in these assumptions and is written to stimulate a more detailed examination and assessment of these assumptions for future breast cancer research.

A Brief Historical Perspective

  1. Top of page
  2. Abstract
  3. A Brief Historical Perspective
  4. Accurate Racial Categories?
  5. Priorities
  6. Opportunities
  7. An Alternative
  8. Recommendations
  9. Acknowledgements
  10. REFERENCES

“I existed in the presence of a look.”

Sartre JP. The reprieve. [translated by Eric Sutton.] New York: Alfred A. Knopf, 1947:135.

Colonial America's censuses and the initial United States 1790 Federal census included racial categories. However, these censuses lacked any intrinsic understanding of African slave ancestral diversity. From 1790 to 1860, census takers had little incentive to differentiate African ancestral diversity due in part to the predominantly chattel status of African slaves in the United States.3, 4 Consequently, if you looked African, you were African. From 1800 to 1870, features like skin color, hair texture, facial features, and/or consanguinity became the basis of biologic population categories for black African descendents living in the United States. However, these racial categories evolved from a 17th century European, colonial, capitalist economy that was driven by slave labor.5, 6 In that economy, slave labor costs were far more important than slave origins, and European colonists discovered that African slaves were cheaper than Native American or European slaves. Therefore, it is not surprising that, after 1850, the United States Constitution would codify black African descendents as an economic category without defining race.4 (States developed race statutes: The United States Constitution does not contain the word “slave” or “Negro” but provides for people who owed service or labor. The 15th Constitutional Amendment includes the word “race” but does not identify or specify what constitutes any particular race.) Therefore, black African identity in the United States originated from servitude, not biology. Skin color emphasis waned as the 20th century approached (many censuses used “M” for mulatto but stopped as the 20th Century arrived), although racial categories would remain vague, at times arbitrary, biologically imprecise, and of questionable scientific validity throughout the 20th century.

Today, many use these categories to study the changing social, demographic, health, and economic characteristics of various groups and provide a historic record of the nation's population diversity, its shifting social attitudes and policies, and its growth. In addition, since the Civil Rights Act of 1965, race and ethnicity data (race and ethnicity are not used synonymously here; see Office of Management and Budget [OMB] Statistical Directive No. 157 and Last, 19888) have been used extensively to monitor and enforce laws for employment, voting rights, housing and mortgage lending, health care services, and educational opportunities. Among Federal agencies, this created a need for better population specific data regarding groups that historically suffered discrimination and differential treatment based on race. In 1977, the OMB responded by issuing the Standards for the Classification of Federal Data on Race and Ethnicity contained in Statistical Policy Directive No. 15.7 This policy was designed to assure uniform reporting standards of racially grouped data by Federal agencies in the United States. Concerned that the 1977 policies did not reflect the nation's increasing diversity, the OMB initiated a review in 1993.7, 9 The revised OMB policy was published in the 1997 Federal Register7 and is referred to henceforth as the Revised Standards. Below, some key events and concerns are considered regarding that review.

First, the OMB asked the National Academy of Science Committee on National Statistics to organize a workshop to discuss review issues (Table 1 shows important review dates). The United States Census Bureau and other agencies supported the review process, including work related to the 2000 Census program. In all, 23 bureaus and/or divisions from 18 Federal agencies as well as public and private institutions took part in the revision.7

Table 1. Important Dates in the Office of Management and Budget Race and Ethnic Standards Review
DateMilestone
  • OMB: Office of Management and Budget; CPS: Current Population Survey; NCT: National Content Test; RAETT: Race and Ethnic Targeted Test.

  • a

    From the Executive Office of the President, Office of Information and Regulatory Affairs Statistical Policy Directive no. 15, “Recommendations from the Interagency Committee for the Review of the Racial and Ethnic Standards to the Office of Management and Budget concerning changes to the standards for the classification of federal data on race and ethnicity.” Washington, DC: Office of Management and Budget, Federal Register, part II, pages 36873–36946. July 9, 1997.

Fall 1995OMB analyzes Federal Register notice comments; receives results of May 1995 CPS Supplement; continues to consult on options with affected groups
March 1996Census Bureau conducts NCT in preparation for 2000 Census
June 1996Census Bureau conducts RAETT in preparation for 2000 Census
November 1996Census Bureau provides test through January 1997 results from NCT and RAETT
Spring 1997OMB publishes Federal Register notice on research results and proposed decisions on changes, if any, to Directive no. 15a
Mid-1997OMB publishes final decision regarding any changes to Directive no. 15 in a Federal Register notice

Reviewers balanced their concern about historic racial data continuity with concerns about intragroup variations of the nation's population-based data sets and how these variations may influence data use. For example, Asian Indians were counted as “Hindus” in censuses from 1920 to 1940, as “whites” from 1950 to 1970, and as “Asians or Pacific Islanders” in 1980 and 1990 censuses.7 In addition, reviewers weighed the Federal government's earlier continuity goal of fluid racial identifiers with changing racial self-perceptions. Reviewers ultimately decided that keeping pace with historic changes in racial data inconsistencies was important, because they reflected social reality and public policy as well as technical decisions by data developers.

Reviewers also considered which agencies used health data, what and how health data resources were used, and how frequently these data were used. For example, frequent vital records data users, such as the National Center for Health Statistics (NCHS), the Office of the Assistant Secretary of Health, and the Centers for Disease Control and Prevention, evaluated the potential impact that the suggested racial category changes would have on medical and health research.7 Those data users concluded that changes may affect health researchers in several ways. For example, birth certificates initially relied on the mother or family members to determine the race of each biologic parent. The baby's race also was based on an algorithm of the parents' races. Currently, the NCHS tabulates birth data according to the mother's race. In censuses and surveys until 1970, racial data were based primarily on observations of government enumerators who administered surveys or questionnaires. Currently, the usual practice is self-administered forms and questionnaires, especially when collecting demographic information.

Another concern was the pending 2000 Census. Therefore, the Bureau of the Census, the NCHS, and several divisions of the Centers for Disease Control and Prevention evaluated racial data quality. However, this evaluation was limited to measurement reliability (i.e., measurement repeatability or consistency under identical or similar circumstances), not measurement validity (the degree to which the actual [true] value is measured) and accuracy. These reliability studies found that Federal data bases displayed 95% consistency in individual responses for white and black populations and 90% consistency for the Asian and Pacific Islander population in the National Content Reinterview Survey of the Census Bureau's 1990 census.7 For American Indians and Alaska Natives, reporting was less consistent at 63% in the same reinterview survey.7 Reporting race also was less consistent for persons of multiple race, Hispanic persons, the foreign born, and persons who could not read or speak English well. The NCHS reported that Asians and American Indians sometimes are misreported as “white” on death certificates, and death rates would be underestimated for these groups if it is found that this misclassification is significant.7 Reviewers ultimately concluded that “black” was a reasonably reliable data category. However, validity and accuracy questions remained unresolved.

Appropriate terminology also was a concern. For example, reviewers considered whether “black” or “African-American” were appropriate terms for Americans of black African descent. The Revised Standards state that the guiding principle is terminology “familiar and acceptable” to respondents, and studies are cited regarding respondent preference for both terms.7 Overall, this meant that either term was appropriate but that there was some analytic concern regarding Caribbean immigrants who may be misclassified if “African-American” was substituted for “black” and whether the suffix “American” was appropriate, because the term “Asian” was not given a hyphenated suffix.

Despite the OMB review, racial categories in the United States remain vague and imprecise. More important, they were never developed for cancer research, they provide limited biologic information, and their usefulness as valid racial categories in cancer research is uncertain.8, 10, 11 Table 2 summarizes the recommended Revised Standards7 minimum racial categories. (The Revised Standards indicate that ethnicity should be ascertained first: “Hispanic” is the only ethnic category, regardless of race. Concurrently, the Revised Standards indicate that it is preferable to collect data on race and ethnicity separately to maintain flexibility and to assure data quality.7 This implies that race and ethnicity are separate population parameters, although how they differ is unclear.)

Table 2. Definitions of the Racial and Ethnic Categories for Federal Statistics and Program Administrative Reportinga
CategoryDefinition
  • a

    From “Recommendations from the Interagency Committee for the Review of the Racial and Ethnic Standards to the Office of Management and Budget concerning changes to the standards for the classification of federal data on race and ethnicity.” Washington, DC: Office of Management and Budget. Federal Register, part II, pages 36873–36946. July 9, 1997.

Race 
 American Indian or Alaskan NativeA person with origins in any of the original peoples of North America and who maintains cultural identification through tribal affiliations or community recognition
 Asian or Pacific IslanderA person with origins in any of the original peoples of the Far East, Southeast Asia, the Indian subcontinent, or the Pacific Islands; this area includes, for example, China, India, Japan, Korea, the Philippine Islands, and Samoa
 BlackA person with origins in any of the black racial groups of Africa
 WhiteA person with origins in any of the original peoples of Europe, North Africa, or the Middle East
Ethnicity 
 HispanicA person of Mexican, Puerto Rican, Cuban, Central American, South American, or other Spanish culture or origin, regardless of race

Accurate Racial Categories?

  1. Top of page
  2. Abstract
  3. A Brief Historical Perspective
  4. Accurate Racial Categories?
  5. Priorities
  6. Opportunities
  7. An Alternative
  8. Recommendations
  9. Acknowledgements
  10. REFERENCES

“The chief danger to our philosophy, apart from laziness and wooliness,…is treating what is vague as if it were precise…”

Ramsey FP. The foundations of mathematics. London: Routledge & Kegan Paul LTD, 1931:269.

Accuracy is derived from precision. To be precise, we must have parameters with unique measurement boundaries. If measurement implies existence, then precise measurement implies that investigators who measure race believe that races truly exist. Unfortunately, we do not always accept the truth, and our beliefs occasionally lead us to rejecting it. Yet, when a valid parameter exists, truth-perception errors occur. To correct these perception errors, concepts such as precision, reliability, and bias are used to describe measurement validity, consistency, and variation.12 Our research epistemology encourages the use of methods that minimize or eliminate instrument, measurement, and criterion errors. We are encouraged to repeat measurements, examine correlations between factors, and/or recalibrate instruments to a new standard when the previous standard is proven invalid. Although many texts describe population measurement hazards,12, 13 physicist Roger Penrose succinctly describes the deception that awaits those who assume the existence of a valid, measurable parameter.14 His general narrative leads to a specific yet unsettling conclusion for breast cancer investigators who are interested in racially disparate outcomes: To measure races assumes that races exist.

Clearly, the assumption that valid racial categories exist is fundamental to any investigation of racially disparate data. Empirically derived, valid research categories allow accurate terminology and precise measurements. Using accurate terms to formulate precise hypotheses is important, because epidemiologic hypotheses are most effective when they are translated easily into accurate exposure and outcome terminology and variables. If exposures cannot be specified or measured, then research questions can be modified or new methods can be developed. Precise measurements are necessary, because they aid hypothesis testing by providing distinct boundaries. Therefore, we must first establish that a parameter exists (is valid) and that our tools measure precisely its unique properties before we address data accuracy. Based on valid assumptions, precise measurements ensure accuracy by minimizing misclassification. Hence, if races exist, then precise racial measures yield accurate racial data.

Are Revised Standards categories valid? A look at health research publications suggests that scientists and institutions in the United States have long assumed that human races exist and that our instruments accurately measure races. Further evidence of this assumption is the proliferation of publicly and privately funded programs that support investigations of racially disparate health statistics. Although such programs may be effective in addressing group specific health disparities, the role of race is uncertain and is ripe for misinterpretation. If we assume that “true” human races are measurable, then we must closely examine our assumption that races exist. (The assumption that races exist among Homo sapiens is a subject of debate and is not addressed here.)

Do Revised Standards categories confound relations between population subgroups, environment, social class, and breast cancer? It is interesting to note that the Revised Standards provide clues:

1) The racial and ethnic categories set forth in the standards should not be interpreted as primarily biologic or genetic in reference. Race and ethnicity may be thought of in terms of social and cultural characteristics as well as ancestry. 2) Respect for individual dignity should guide the processes and methods for collecting data on race and ethnicity; ideally, respondent self-identification should be facilitated to the greatest extent possible, recognizing that, in some data-collection systems, observer identification is more practical.7

In short, the categories are designed to convey no primary biologic or genetic information: They are vague and facilitate self-identification whenever possible. Therefore, Revised Standards categories are neither nature nor nurture, but both. Consequently, their use dilutes findings in at least three ways. First, the categories infer minimal information about biology or genetics—information that is fundamentally important to contemporary cancer models. Second, although they are convenient, these categories do not represent distinct boundaries, and their meanings overlap with other terms.7, 8 Finally, allowing respondents to use “fluid” identity criteria decreases statistical resolution by increasing the likelihood of misclassification. Unfortunately, increases in “multiracial” families in the United States and the need to select several Revised Standards categories have compounded the problem further.15

Therefore, breast cancer researchers need precise, innately biologic categories that promote accurate identification. Their development should be guided, at a minimum, by whether there are several human races or one human race.

Priorities

  1. Top of page
  2. Abstract
  3. A Brief Historical Perspective
  4. Accurate Racial Categories?
  5. Priorities
  6. Opportunities
  7. An Alternative
  8. Recommendations
  9. Acknowledgements
  10. REFERENCES

“…in science, all terms that are really needed must be undefined terms.”

Popper K. In: Miller D, editor. Popper selections. Princeton, NJ: Princeton University Press, 1985:97.

What is uniquely Afro-American? Who is uniquely Afro-American? How does one become uniquely African-American? Perceptions of these questions can shape priorities. Therefore, nongovernment researchers and government researchers may have different priorities, because each may have a different mission. Questions to consider now are the following: 1) Do races exist among Homo sapiens, is “African-American” (black, Negro, colored) a racial category and can the category be measured, and is it uniquely related to breast cancer? 2) When we decide that current categories are inadequate, how do we make the transition to other categories and reeducate ourselves about contemporary, racially ambiguous categories? If the above questions are more important than the Revised Standards priorities,7 then investigator priorities should emphasize breast cancer research goals and not Federal data reporting standards.

First, the advances and progress made in developing breast cancer diagnostic categories must occur similarly for population-based ancestral categories. Our advances in breast cancer diagnostic categories are based on measurement tools that rely heavily on empiric, biologically inherent clinical parameters and emerging high-resolution technologies that are rooted in the natural history of breast cancer. These advances have created more precise breast cancer disease categories and less disease misclassification, and they suggest that our ability to develop precise disease categories exceeds our ability to develop accurate racial categories.

Second, we should consider the impact of a transition from current racial categories to other categories and avoid future use of Revised Standards categories. Past and current research using Revised Standards categories should be viewed with skepticism when used to connote race. For example, researchers may consider eliminating Revised Standards categories if they do not explain exposure and disease correlations. This may lead to reports that accurately emphasize social class, toxic environmental exposure, immune status, or other factors.

Last, our priorities should emphasize reeducating everyone (researchers especially) about the accuracy and use of past and current racial categories as well as stating the purpose of any new research standard. The Revised Standards is an important transition but should not remain the “gold standard” for breast cancer research. As Popper suggests above, are current terms necessary if we really understand the meaning of race, or do the categories represent convenient, strategic, racial constructs and beliefs?

Opportunities

  1. Top of page
  2. Abstract
  3. A Brief Historical Perspective
  4. Accurate Racial Categories?
  5. Priorities
  6. Opportunities
  7. An Alternative
  8. Recommendations
  9. Acknowledgements
  10. REFERENCES

“We have sequenced the genome of three females and two males who have identified themselves as Hispanic, Asian, Caucasian, or African-American. We did this … to help illustrate that the concept of race has no genetic or scientific basis.”

Venter JC. Remarks at the Human Genome Announcement. The White House, June 26, 2000.

If race has no genetic or scientific basis, then what may explain the racial focus of some breast cancer research? Historic racial definitions, their widespread use, and our beliefs are responsible in part. Another, more subtle source of racial focus among researchers in the United States is the government's need to categorize the population racially. For example, as Federal agencies pursue their missions, they create large data sets containing data elements from the Revised Standards. Unfortunately, these data make secondary data analysis attractive, because previously collected data often are cheaper to obtain, although they were collected for purposes other than breast cancer research. Regrettably, many researchers either fail to recognize or caution readers about the inherent weaknesses of Revised Standards categories and do not acknowledge that their tacit acceptance encourages beliefs that races exist, our ability to accurately measure race is resolved, and that current categories withstand rigorous construct, content, and criterion validity challenges (Revised Standards categories must correspond to actual races [construct], incorporate actual racial information [content], and temporally correlate with actual external racial criteria [criterion]).

Recognizing and developing valid, population-based breast cancer research categories will be determined by our commitment to empiric methods and understanding of breast cancer biology. We must describe breast cancer in all of its manifestations, develop hypotheses that provide opportunities to test our observations, and determine the relation and meaning of our observations in the context of all we know about breast cancer. If our categories do not explain our observations after extensive testing, then they should be replaced with categories that better explain both past and current observations. If our categories do not withstand all relevant critical review, then we should develop categories that endure all critical analyses. Finally, if Revised Standards categories do not predict health outcomes with greater likelihood compared with new categories, then we should continue our search for new categories. Opportunities to improve population-based research categories will occur if we remain committed to empiric methods and frequently reexamine our beliefs. However, until we decide that human races exist, less emphasis on race and more emphasis on valid categories would be an improvement.

We must ask what the Revised Standards racial categories mean and whether they matter in breast cancer research. If we are encouraged continually to investigate racially disparate health statistics based on current categories, then electronic and print media organizations will continue to report racially disparate findings. Consequently, people who rely on these communications media for health information will continue to attribute racial membership to disease risk and will interpret and embody that information in numerous, unanticipated ways.16 The complex epidemiologic relations between breast cancer and race17 require precise, appropriate risk communication. Without an understanding of race within our species, we may forever be unable to resolve any important distinctions that may improve lives or that may provide accurate information to lower the risk of breast cancer or other diseases.

An Alternative

  1. Top of page
  2. Abstract
  3. A Brief Historical Perspective
  4. Accurate Racial Categories?
  5. Priorities
  6. Opportunities
  7. An Alternative
  8. Recommendations
  9. Acknowledgements
  10. REFERENCES

“For reasons that seem to transcend cultural peculiarities … we construct our descriptive taxonomies and tell our explanatory stories as dichotomies, or contrast between inherently distinct and logically opposite alternatives.”

Gould SJ. Deconstructing the “science wars” by reconstructing the old mold. Science. 2000;287:253–261.

Alternative breast cancer research categories must clarify and advance our understanding of fundamentally important human biologic differences. For example, gender is a reliable, convenient definition boundary with intrinsic biologic meaning to cancer. However, often, more than a biologic dichotomy is required to determine important human differences. What breast cancer researchers need is diversely stratified, population-based research categories that convey primarily biologic and genetic information and that are grounded in biologically inherent ways to ancestry and breast cancer. The major histocompatibility complex (MHC) polymorphisms and associated proteins are potential candidates.

MHC Class I and II genes and their respective antigens are the most numerous, divergent, and evenly distributed within Homo sapiens18 and allow convenient, reliable definition boundaries with essential biologic meaning to ancestry and cancer. MHC Class I polymorphisms provide researchers with tools to investigate evolutionary selection, recombination, and genetic drift models as well as providing intrinsically biologic information about immunity and its role in breast cancer oncology. MHC-based population subdivisions have intrinsically biologic correlations with models of human speciation as well as with the diseases that have coevolved with humans. In addition, this approach does not exclude our use of cultural behaviors, beliefs, and practices as risk factors but stratifies humans into immunologic groups that are related innately to oncogenesis and that also represent distinct population categories.

Finally, MHC polymorphisms are not presented here as the only alternative. Several problems are apparent. For example, there is overall insensitivity to non-white population antigens, because current serologic typing was developed with alloantisera and cells that were obtained from whites.18 Epidemiologic barriers would require today's participants to know their MHC genotype and/or antigen profile, information that generally is unknown by most people and is not found commonly in their medical records. Current MHC-based breast cancer investigations would require tissue from study participants and would oblige investigators to consider additional legal and ethical issues associated with tissue requests.19 In addition, a change in United States policy requiring MHC antigen profiles and genotyping at birth would be necessary to make this a standard part of health records (similar to ABO erythrocyte antigen typing at birth). Despite these problems, MHC alleles and their proteins may provide a better explanation of past and current biologic and genetic population subgroups, may withstand critical scientific reviews better, and may provide better models that predict breast cancer outcomes compared with the Revised Standards categories. Such models await empiric discovery and criticism but will be necessary steps in advancing current scientific knowledge and understanding of breast cancer and race. Finally, new categories would not prevent the use of Revised Standards categories as potential social class identifiers if it can be shown, for example, that “African-American” is a valid social class in the United States. It is time to consider whether racial categories in the United States are economic social strata derived from 17th century colonial capitalism5, 6 or are empirically derived boundaries with unique biologic meaning.

Recommendations

  1. Top of page
  2. Abstract
  3. A Brief Historical Perspective
  4. Accurate Racial Categories?
  5. Priorities
  6. Opportunities
  7. An Alternative
  8. Recommendations
  9. Acknowledgements
  10. REFERENCES

Given this country's long and difficult racial history, breast cancer researchers must recognize and report important human health differences. The following recommendations are directed at that objective: 1) Establish priorities to develop population-based identifiers for breast cancer research. 2) Identify opportunities to develop improved population-based categories in breast cancer research by i) using scientifically valid population subgroup categories for breast cancer research; ii) discontinuing the use of Revised Standards categories as racially valid identifiers; and iii) reeducating Americans about the use of racial identifiers in breast cancer research. 3) Use valid assumptions to determine whether races exist among Homo sapiens, e.g., i) if races exist, then determine what instruments measure the unique properties of racial subpopulations; and ii) if races do not exist, then develop alternative categories and determine what instruments measure their unique subpopulation properties. 4) Increase funding that develops scientifically valid population group identifiers for breast cancer research. A comprehensive consideration of the specific research and broad societal implications of these recommendations is needed.

Acknowledgements

  1. Top of page
  2. Abstract
  3. A Brief Historical Perspective
  4. Accurate Racial Categories?
  5. Priorities
  6. Opportunities
  7. An Alternative
  8. Recommendations
  9. Acknowledgements
  10. REFERENCES

The author thanks Larre R. Figgs for editorial assistance and Isabel Estudillo for her efforts in facilitating article reviews.

REFERENCES

  1. Top of page
  2. Abstract
  3. A Brief Historical Perspective
  4. Accurate Racial Categories?
  5. Priorities
  6. Opportunities
  7. An Alternative
  8. Recommendations
  9. Acknowledgements
  10. REFERENCES
  • 1
    Jones LA, Brawley O, Johnson-Thompson M, Lythcott N, Newman L. Overview of the Summit Meeting evaluating research in African American women. Cancer. 2003; 97 (1 Suppl ): 207210.
  • 2
    Clarke CA, West DW, Edwards BK, Figgs LW, Kerner J, Schwartz AG. Existing data on breast cancer in African-American women: what we know and what we need to know. Cancer. 2003; 97 (1 Suppl ): 211221.
  • 3
    Horton JO. Free people of color: inside the African American community. Washington, DC: Smithsonian Institution Press, 1993.
  • 4
    Quarles B. The Negro in the making of America. New York: Simon and Schuster, 1987.
  • 5
    Williams E. Capitalism and slavery. Chapel Hill: University of North Carolina Press, 1994.
  • 6
    Morgan ES. American slavery-American freedom. New York: WW Norton & Company, 1975.
  • 7
    Executive Office of the President, Office of Information and Regulatory Affairs. Recommendations from the Interagency Committee for the Review of the Racial and Ethnic Standards to the Office of Management and Budget concerning changes to the standards for the classification of federal data on race and ethnicity. Fed Reg. 1997; II: 3687336946.
  • 8
    Last JM. A dictionary of epidemiology, 2nd ed. New York: Oxford University Press, 1988.
  • 9
    Wallman KK, Evinger S, Schechter S. Symposium: race/ethnicity and the 2000 Census. Am J Public Health. 2000; 90: 17041708.
  • 10
    Marshall E. DNA studies challenge the meaning of race. Science. 1998; 282: 654655.
  • 11
    Nelson WJ. Racial definition: background for divergence. Phylon. 1986; 47: 318326.
  • 12
    Clayton DHM. Statistical models in epidemiology. New York: Oxford University Press, 1996.
  • 13
    Armitage PBG. Statistical methods in medical research. Oxford: Blackwell Scientific Publications, 1990.
  • 14
    Penrose R. The emperor's new mind: concerning computers, minds, and the laws of physics. New York: Penguin Putnam Inc., 1991.
  • 15
    Williams DR, Jackson JS. Race/ethnicity and the 2000 Census: recommendations for African Americans and other black populations in the United States. Am J Public Health. 2000; 90: 17281730.
  • 16
    Lupton D. Risk. New York: Routledge, 1999.
  • 17
    Schottenfeld D, Fraumeni JF Jr. Cancer epidemiology and prevention. New York: Oxford University Press, 1997.
  • 18
    Parham POT. Population biology of antigen presentation by MHC class I molecules. Science. 1996; 272: 6774.
  • 19
    Dunn CM CG. Protecting study volunteers in research: a manual for investigative sites. Boston: Center Watch Publishers, 1999.