Who Is Eligible? Should Affirmative Action Be Group- or Class-Based?
Darity, William, Jr., Deshpande, Ashwini, Weisskopf, Thomas, The American Journal of Economics and Sociology
Affirmative action (AA) consists of a set of anti-discrimination measures intended to provide access to preferred positions in a society for members of groups that would otherwise be excluded or underrepresented. It provides a mechanism to address contemporary exclusion, (1) particularly a mechanism to desegregate elites. AA can be utilized to change the demography of elite position holders, making those positions more representative of the ethnic/racial/caste/gender composition of the society as a whole. (2)
Policy-makers who contemplate positive discrimination policies on behalf of marginalized ethnic/racial groups grapple with the issue of whether such policies should be "class-based" or "group-based." By "group" we mean any ethnic or racial identity group, including caste groups. Group-based programs benefit only members of the targeted groups. Class-based programs require a means test of some type, usually an income cut-off above a poverty line. When members of marginalized groups are vastly over-represented among the poor, they will be more likely than others to benefit from a class-based program, but such a program obviously also will benefit members of nonmarginalized groups.
The purpose of this article is to explore the impact of group- versus class-based AA on increasing access to under-represented groups. The article demonstrates that the access of the under-represented groups to preferred positions gets diluted, as group-based AA is substituted by class-based AA. Intuition suggests that color-blind criteria for preferences (such as class-based affirmative action) would not do as well as group-based criteria in making opportunities available to members of an under-represented group, for as long as there are any people in the class-preferred category other than members of the preferred group, the access of the latter is diluted. Our article demonstrates systematically that the intuitive belief is indeed valid, and further explores how the amount of the dilution is greater (a) the smaller is the group as a proportion of the overall population and (b) the less group status is correlated with class status.
Group- Versus Class-Based AA
We will not summarize here the entire body of literature on the "class versus race" debate in the context of AA (see Weisskopf 2004: Chapter 16, for a discussion of some key arguments). Nevertheless, it would be useful to restate a few aspects of the debate. The advocates of class as the basis of AA believe that socioeconomic status captures the variety of kinds of disadvantage most comprehensively (see, for example, Kahlenberg 1996) and that, hence, socio-economic integration is a better means of promoting social and economic justice than racial integration. Magnus and Mick (2000) advocate class-based AA in admissions to medical schools. They believe that lower-class status is a present handicap, not an intergenerational one, and it can be assessed at the individual level, not just in the aggregate. However, as Darity, Dietrich, and Guilkey (2001) demonstrate, inter-generational handicaps can be assessed at the individual level, quite apart from the fact that an inter-generational handicap also is a present handicap. Magnus and Mick further argue that doctors who are originally from poorer backgrounds are more likely to be sensitive and responsive to the needs of poorer patients. However, the latter is at least as true for ethnic or racial groups as it is for class; additionally, a medical student who comes from poverty may eventually become a rich doctor, which may reduce his or her sensitivity to those who continue to be poor.
It has also been argued that, given the large overlap between race and class, these might be two alternate ways of capturing essentially the same set of individuals. Fryer and Loury (2005) contend that "for a fixed distribution of traits, any color-blind affirmative action policy is less efficient than the optimal color-sighted policy calibrated to achieve the same degree of racial diversity," thus providing support to our hypothesis. …