Most people would agree that equal opportunity to participate as a full and functioning member of society is important. Nonetheless, existing social and economic disparities among racial and ethnic groups suggest that our society has yet to achieve this goal. For instance, Hispanics have higher school dropout rates than other racial and ethnic groups (Hauser et al., 2002). The black–white wealth gap remains large (Conley, 1999; Oliver and Shapiro, 1995). Young Native Americans are incarcerated in federal prisons at higher rates than any other minority racial group (Smelser and Baltes, 2001; Weich and Angulo, 2002). And some Asian Americans, among other minority groups, have poorer access to health care services and treatments than whites (Institute of Medicine, 2003). Such racial disparities are pervasive and may be the result of racial prejudice and discrimination, as well as differences in socioeconomic status, differential access to opportunities, and institutional policies and practices.
Such racial disparities persist despite the many legal and social changes that have improved opportunities for minority racial and ethnic groups in the United States. Several factors may contribute to racial differences in outcomes, including differences in socioeconomic status, differential access to opportunities, and others. One factor that should be considered is the role of racial discrimination. Overt discrimination against African Americans and other minority groups characterized much of U.S. history; a question is whether and what types of discrimination continue to exist and their effects on differential outcomes.
Although researchers in specific disciplines have investigated discrimination in particular domains, there has been little effort to coordinate and
expand such research in ways that could help to better understand and measure various kinds of racial and ethnic discrimination across domains and groups and over time. To address this problem, the Committee on National Statistics convened a panel of scholars in 2001 to consider the definition of racial discrimination, assess current methodologies for measuring it, identify new approaches, and make recommendations about the best broad methodological approaches. In particular, this panel was asked to conduct the following tasks:
Give the policy and scholarly communities new tools for assessing the extent to which discrimination continues to undermine the achievement of equal opportunity by suggesting additional means for measuring discrimination that can be applied not only to the racial question but in other important social arenas as well.
Conduct a thorough evaluation of current methodologies for measuring discrimination in a wide range of circumstances where it may occur.
Consider how analyses of data from other sources could contribute to findings from research experimentation, such as the U.S. Department of Housing and Urban Development paired tests.
Recommend further research as well as the development of data to complement research studies.
Although there is substantial direct empirical evidence for the prevalence of large disparities among racial and ethnic groups in various domains, it is often difficult to obtain direct evidence of whether and to what extent discrimination may be a contributing factor. Differential outcomes by race and ethnicity may or may not indicate discrimination. Examples of studies using methods that persuasively measure the presence or absence of discrimination are rare, and appropriate data for measurement are often unobtainable. As a result, there is little scholarly consensus about the extent and frequency of discrimination and how it relates to continuing disadvantages along racial and ethnic lines (Fix and Turner, 1998).
One reason it is difficult to assess discrimination is that changes have occurred in the nature of prejudiced attitudes and discriminatory behaviors. With the passage of the Civil Rights Act of 1964 and other laws that prohibit discrimination because of race in a variety of domains, overt discrimination is less often apparent. However, discrimination may persist in more subtle forms. Indeed, social psychological research suggests that relatively automatic and unexamined cognitive processes, of which the holder (and sometimes the target) may not be fully aware, can lead to discrimination (Devine, 1989; Fiske, 1998). These subtleties make defining and measuring discrimination more difficult.
STUDY APPROACH AND SCOPE
The panel’s goal in this report is to review and comment on the methods used in various social scientific disciplines to identify types of racial discrimination and measure their effects. The report is designed to help social science researchers, policy analysts, federal agencies, and concerned observers better understand how to assess racial discrimination in different domains, drawing on different social science methods and data sources as appropriate. To approach this important but difficult task, the panel focused on defining relevant concepts, examining various methodological approaches and data sources, and considering directions for future research.
The purpose of this report is not to promote a single “right” way to measure discrimination. In some situations, one approach may be more easily implemented and more credible; in other situations, another approach may be more appropriate. Often, multiple approaches will be needed to provide credible evidence about the prevalence of discrimination in a domain. Thus, the panel attempts to identify the broad range of approaches for measuring discrimination and to provide a critical review of their relative credibility when applied in different situations. The panel develops a cross-disciplinary research and data collection agenda for action by public and private funding agencies and the research community.
The report makes no attempt to actually measure current or past levels of discrimination in any domain. Our purpose is not to report numbers or impacts but to provide guidance and encouragement to researchers and policy analysts as they work across domains to identify where discrimination may be present and what its effects may be.
In the first part of this report, the panel defines the concepts of race and racial discrimination from a social science perspective, which we believe is the appropriate perspective for research and policy analysis on discrimination. When referring to race in the report, the panel uses the categories established by the federal classification standards (U.S. Office of Management and Budget, 1997) to identify whites, blacks or African Americans, American Indians and Alaska Natives or Native Americans, Asians, and Native Hawaiians and other Pacific Islanders. According to these standards, Hispanics or Latinos are referred to as an ethnic group. Yet, although the panel was asked to consider racial discrimination, Hispanics (a rapidly growing ethnic population) also face discrimination. In addition, concepts of race and ethnicity are not clearly defined for many Hispanics, so for these two reasons our discussion often refers to Hispanics as well as to specific racial groups. Throughout the report, the term disadvantaged racial group is used to refer to groups in the United States (e.g., blacks) whose disadvantage can be linked historically to discriminatory practices and policies and who are, consequently, part of a legally protected class.
The panel is concerned with broad types of discriminatory behaviors and processes that have negative consequences for disadvantaged racial groups in various social and economic arenas. We draw on sociological, social psychological, and other literature to develop our definition of racial discrimination. We also discuss the conceptual possibility that discrimination may operate not just at one point in time and within one particular domain but at various points within and across multiple domains throughout the course of an individual’s life. The panel acknowledges that the effect of such cumulative discrimination may not be easily identified or measured.
In interpreting that part of its charge to review measurement methods, the panel chose to address broad approaches that could be applied across domains, rather than making recommendations about specific approaches for particular domains. Therefore, although examples are used throughout the report to illustrate efforts to measure discrimination in particular circumstances, our main focus is on methods (e.g., experiments, observational studies, survey research) that can be used to study discrimination under many different circumstances.
The examples of disparities and discrimination measurement that we provide come from research in five domains: labor markets and employment, education, housing and mortgage lending, health care, and criminal justice. Although not the only domains of concern, these are key areas of social interaction for which discrimination can seriously limit life opportunities; these are also among the areas for which the federal government regularly collects administrative and survey data long used by researchers to study discrimination and discriminatory effects. We do not provide an exhaustive set of examples for each of these areas. Rather, a selected bibliography of important literature reviews, major reports, and other work on data collection and analytical methods used in each of these domains is provided at the end of this report.
Much of the discussion in this report on such topics as statistical inference, experimental design, and data quality is relatively technical in nature. Although sometimes dry, the import of this discussion should not be misunderstood by readers who are deeply concerned about the possible extent and continued effects of racial discrimination in American life. It was our shared concern about racial discrimination that drew each member of the panel into the in-depth discussions of measurement reflected in this report. Because we view racial discrimination as a crucial social issue, we believe it is essential to use the most credible and accurate measurement approaches.
In carrying out this study, the panel met and deliberated over a period of almost 2 years. We held meetings, invited speakers, and commissioned several papers (see Box 1-1); we requested input from prominent scholars on key issues; reviewed a large body of literature on salient aspects of the law and criminal justice, labor markets, housing markets, education, and
Three papers were commissioned to inform the panel’s work on this report. Smith (2002) reviews methods for measuring racial discrimination, focusing primarily on survey-based approaches. Ross and Yinger (2002) examine the use and quality of data on race collected for administrative purposes, as well as issues of comparability and interpretation that arise for both enforcement officials and scholars attempting to study discrimination. Finally, Nelson and Bennett (2003) review the courts’ use of statistics to make decisions in cases alleging racial discrimination in employment. These papers are available directly from the authors.
The panel also commissioned several papers for a workshop on measuring racial disparities and discrimination in elementary and secondary education (see Appendix A). The purpose of the workshop was to expand and improve the statistical capability of the U.S. Department of Education and other federal agencies to measure and track discrimination. The four commissioned papers relating to measuring racial disparities and discrimination in education are published in Teacher’s College Record (Farkas, 2003; Holzer and Ludwig, 2003; Mickelson, 2003; Ryan, 2003).
health care; investigated the ways in which race is defined in various federally funded surveys; reviewed the literature on race, prejudice, and discrimination; and examined other literature on survey design, experimental evidence, and statistical analysis.
This report is divided into three parts. The chapters in Part I provide a conceptual framework for thinking about racial discrimination. Chapter 2 explores the meaning of race as a social construct and provides historical background on the complex issues surrounding race in the United States and how it is measured in the decennial census and other federal data collections. Chapter 3 defines discrimination from a social science perspective and explains why we focus on racial discrimination. Our definition of racial discrimination is informed by legal concepts of discrimination, but it also encompasses behaviors and processes that may not be unlawful or easily measured. Chapter 4 provides a framework for understanding how racial
discrimination may operate. As the discussion indicates, there are different ways in which discrimination can occur and various mechanisms that can result in discriminatory behavior. Identifying various sources of discrimination is a crucial first step in developing theories or models of discrimination and using them to guide data collection and research for measuring the presence and extent of different types of discrimination.
The chapters in Part II examine methodological approaches to measuring discrimination and the advantages, limitations, and best techniques associated with each. Chapter 5 provides a general framework for inferring causation and a brief introduction to some of the topics covered in detail in the chapters that follow. Chapter 6 focuses on experimental methods, including field and laboratory experiments. Chapter 7 describes the use of statistical analysis of observational data to measure discrimination, reviewing the necessary assumptions and potential credibility of various approaches. Chapter 8 focuses on approaches employing attitudinal and behavioral indicators of discrimination, including methods based on survey data and administrative records. Each of these chapters describes specific approaches and the situations in which they can be implemented and may be appropriate. Where possible, we also attempt to identify more and less credible approaches, providing guidance for future scholars seeking to use the most effective methods. Chapter 9 at the end of Part II addresses issues of racial profiling, as an illustration of an area in which measuring discrimination is difficult.
The chapters in Part III present the panel’s priorities for data collection and research for improved measures of race and racial discrimination. Chapter 10 describes the data collected by federal statistical and administrative agencies that may support analysis of racial discrimination and its effects. The discussion focuses on concepts and measures of race and ethnicity in federal data sources, how different measures may affect distributions and consequent analyses for racial and ethnic groups, and research that is needed to improve federal measures. Chapter 11 considers the nature of cumulative effects of discrimination within and across multiple domains, seeking to identify techniques that can be used to provide a fuller measure of the impact of discrimination when it occurs over time and in more than one social arena. Little empirical work has been done on cumulative discrimination, so research and data collection in this area are important to pursue.
Finally, Chapter 12 suggests next steps for program and research agencies to build a research agenda that is directed to priority needs for measuring racial discrimination. The aim of the chapter is not to develop a detailed agenda per se, which is beyond the panel’s scope and resources, but to suggest a series of steps whereby agencies may identify priority research topics; evaluate them for feasibility and cost-effectiveness; and bring to bear the necessary conceptual frameworks, research methods, and data. Whether
conducting research from a policy perspective or more basic research, it will be important to support multidisciplinary studies that draw on a range of methodologies and data sources.
The report ends with two appendixes: Appendix A presents the agenda for the Workshop on Measuring Racial Disparities and Discrimination in Elementary and Secondary Education held by the panel in July 2002; Appendix B provides biographical sketches of the panel members and staff.