Part III
Data Collection and Research
Part I of our report provided definitions of race and racial discrimination from a social science research perspective and an explication of various types of race-based discrimination and the mechanisms by which overt and subtle forms of discrimination may occur. Part II reviewed the strengths and weaknesses of several broad methods for conducting research on racial discrimination, including laboratory and field experiments, analysis of observational data and natural experiments, and direct measures of racial attitudes and experiences with discrimination from reports in surveys and administrative records.
The discussion of each method in Part II emphasized the difference between descriptive analysis and causal inference. For example, it is one thing to find differences in educational or income levels between minorities and whites and quite another thing to draw a causal inference by which some part of those differences can be validly and reliably attributed to racial discrimination. It is also not straightforward to relate discriminatory attitudes to discriminatory behaviors that have adverse consequences for racial groups. Some of the problems that impair the ability to draw valid causal inferences include that experiments cannot vary the race of any one individual, observational data lack key variables that contribute to differential outcomes among race and ethnic groups, and direct reports of discriminatory behavior and experiences can be biased in one or more respects. In short, there are no ready answers for researchers and policy analysts who are looking to provide definitive information on which to base public and private organization policies about ways to ameliorate discrimination and its effects.
In Part III, we identify priority areas for research and data collection that can help build a stronger base of knowledge about the incidence, causes, and consequences of racial discrimination in a variety of domains. Our discussion emphasizes the need for research that draws on the strengths of different kinds of measurement methods and data sources. Such research requires concerted cooperative efforts among funding agencies that have traditionally funded certain kinds of studies and certain disciplines and among researchers themselves.
Part III comprises Chapters 10–12. Chapter 10 provides a more detailed description than was initially provided in Chapter 2 of federal government standards for collecting data on race and ethnicity and how federal racial categories have changed over time with changing societal conceptions of race. Although not always consistent with scholarly concepts of race, the federal standards are important because they shape much of the data that are available for analysis of racial discrimination, disparities among racial groups, and related topics. The chapter considers measurement issues that affect reporting of race and ethnicity and makes recommendations for continued governmental collection of race data and methodological research to understand reporting effects.
Chapter 11 considers the concept of cumulative discrimination and how racial discrimination may have effects over time and across different domains. Cumulative effects may be missed using some of the methods described earlier in this report. Because so little empirical research has been conducted on cumulative phenomena, either over time or across domains, we treat this topic as a matter of priority for future research. Our discussion in this chapter begins to consider theories and possible approaches that may help researchers interested in studying mechanisms of cumulative discrimination and their effects.
Finally, Chapter 12 provides suggestions to program and research agencies of next steps for building an agenda for research and associated data collection. The aim of this chapter is not to develop a detailed agenda per se; rather, it brings together the recommendations that are in earlier chapters and puts them in a framework of the need for and power of multidisciplinary studies that draw on multiple methods and data sources. Because of the difficulties of measuring racial discrimination, the best analyses will make use of findings from a variety of studies that, ideally, are implemented within a common conceptual and measurement framework.