What do we support when we support affirmative action?

Alternate post title: When all you have is access to the GSS dataset, everything starts to look like a GSS variable.

To expand on my reply to Jeremy under this post, it is important to distinguish between support for (or opposition to) affirmative action in principle and support or opposition concerning concrete affirmative action programs or policies. I believe that support for affirmative action is funnel-shaped: many Americans are, to steal from George Orwell,

… the sort of person who is in sympathy with the fundamental aims of [affirmative action], who has the brains to see that [affirmative action] would ‘work’, but who in practice always takes to flight when [affirmative action] is mentioned.

Fewer, but still many Americans – “a plurality, if not a slim majority” – will confess their support to affirmative action in some general fashion, the vaguer the better. Then the funnel continues narrowing. Fewer support affirmative action policies, such as quotas or preferential hiring, and by the time you get down to cases, such as the misadventures of Ricci and Co. in Connecticut, the opposition to affirmative action seems insurmountable.

I would argue that this isn’t just a matter of tricky wording on survey questions. Affirmative action policies often reflect a general desire to even out the playing field, rather than any precise or reliable data about what works or will be popular. As a result, I believe, there is a group out there of people who would support some sort of affirmative action program(s), but may oppose specific manifestations of affirmative action, particularly in high-profile cases such as Ricci, or, earlier, the University of Michigan’s infamous affirmative action system for undergrads.

Somewhere out there is a data set better suited to addressing the empirical questions of whether this group of affirmative action proto-supporters exists, how big it is, and how much influence it might wield. Until then, all I’ve got is this half-baked analysis of data from the 2006 General Social Survey. I looked at the results for two items: one asking respondents to indicate their support for preferential hiring and promotion of blacks, and another asking respondents to indicate their support for government aid to blacks. Throughout, I make the assumption that the latter item represents a (more) general principle of affirmative action, while the former represents a (more) concrete policy or set of policies. The wording of the two questions is as follows:

(Preferential hiring)
Some people say that because of past discrimination, blacks should be given preference in hiring and promotion. Others say that such preference in hiring and promotion of blacks is wrong because it discriminates against whites. What about your opinion — are you for or against preferential hiring and promotion of blacks?

(Government aid)
Some people think that [blacks] have been discriminated against for so long that the government has a special obligation to help improve their living standards. Others believe that the government should not be giving special treatment to [blacks] . Where would you place yourself on this scale, or haven’t you made up your mind on this?

Filtering the data for cases that have valid responses to both items, I obtained the two following tables, breaking down the frequencies of each response to the two questions. (Click for readable size.)

Preferential hiring

Government aid

Forfeiting originality, here is how I interpreted this data in my earlier comment:

So what’s going on here? By the most conservative interpretation, nine more people in the sample (1%) support generic “government aid for blacks” than the more specific (but still very vague) “preferential hiring and promotion.” More likely, this difference is greater, since the aid question included a “middle-ground” option (AGREE WITH BOTH), but the preferential hiring question did not.

What this says to me is A.) that our best chance at broadening support for affirmative action is to reach out to the 1% (but probably more) in the middle who are sympathetic to the principle but hostile to a specific program … Sometimes, the solution will lie in modifying the affirmative action program under consideration rather than demonizing its opponents. Also, B.) that arguments in favor of affirmative action must be made differently for the principle than for specific programs: general arguments about chronic racial inequality may sway people’s minds on the principle, but they will likely win few new allies for the programs.

Mere hours after I wrote this, I realized that there were probably more than 9 people in the sample who supported the principle but opposed the policy. The hypothesis I should have been testing was whether more people favored aid and opposed preferences than favored preferences and opposed aid. Given my limited toolkit, I reached straight for a crosstabulation of the two variables:

Crosstab of “preferential hiring” and “government aid,” column percentages

First things first – there is a clear (and statistically significant) positive correlation between support for preferential hiring and support for government aid. However, in the cells which buck this correlation (along the counterdiagonal) there does appear some evidence of the phenomenon I have been suspecting. To wit: 99 respondents (10.54%) support government aid but oppose preferential hiring, while only 23 respondents (2.45%) oppose government aid while supporting preferential hiring. This result is slightly more exciting than the “1%” comparison of the separate frequencies tables above.

Again, running with the assumption that “special [government] treatment” is more of a principle and less of a policy than “preferential hiring and promotion,” this analysis seems to lend some credibility to the existence of a layer of “proto-support” for affirmative action: individuals who favor the spirit and aims of affirmative action programs, but may disagree with the implementation. To wring some implications out of this mess:

As above, we must be aware of the difference between support for a principle and support for a policy. As far as affirmative action goes, recognizing that there are people who favor the former and object to the latter is crucial. This means that not every defense of an affirmative action policy needs to go down as if the overarching principles are under attack. Particularly, we should reach out to anyone who might be a proto-supporter – perhaps even working with them to reform affirmative action policies so that they might become effective and palatable.

Taking a broader view, however, any stratum of proto-support for affirmative action seems wedged between much heftier blocks of true believers. Those who reject the idea of affirmative action may or may not be swayed by arguments about racial inequality, but they will almost certainly not be persuaded to like affirmative action policies so long as they believe their backing principles are false.

I suspect that tailoring the affirmative action message to these different constituencies is harder than it might seem. Tweaking and rebranding affirmative action programs might appeal to the proto-supporters, but would further convince the diehard opponents that affirmative action is being “shoved down their throats.” On the other hand, speaking directly to the roots of the hardcore opposition – “racism, ignorance, selective observation” – is likely to leave the proto-supporters feeling spurned and insulted.

And that’s all the original and unoriginal opinions on the matter that this no-statistical-package-having blogger cares to share at present. I should add, as I had originally replied to Jeremy,

Of course, I could be entirely wrong and misdirected about this, and twisting the data beyond recognition, but that’s for you to decide.