When Does Taking Account of Discrimination by Others Amount to Discrimination Itself? A Question Posed By the Obama and Clinton Candidacies

Caucus-goers and primary voters in both major parties surely want to select a Presidential candidate they trust and admire, whose values they share, and perhaps most of all, who stands a good chance of winning the general election. For ideological outliers in each party, these criteria often conflict. A socially conservative Republican might be happiest with Mike Huckabee as the nominee, but worry that he will alienate moderates in the general election, thus helping elect a Democrat. Likewise, a left-leaning Democrat might think that Dennis Kucinich would be the best President, but fear that nominating Kucinich would greatly increase the likelihood of the Republican nominee winning. Such voters must therefore decide whether and how to trade off their own political preferences against the candidates' "electability."

The potential for this sort of tradeoff exists in most Presidential primary election years, but in the current cycle it has an additional twist, at least for Democrats. This time around, calculations of electability may take into account the likelihood that some number of general election voters will shun the eventual Democratic nominee on either racist or sexist grounds.

Should Democratic primary voters who would like to see Hillary Clinton as President nonetheless vote against her on the ground that her persistent high negatives in opinion polls may reflect unfair double standards that some substantial fraction of the public applies to powerful women? Should primary voters who think Barack Obama would make the best President nonetheless vote against him because they worry that, regardless of what people tell opinion pollsters, some core of white voters will vote against Obama because he is African-American? Would voters who are themselves not sexist or racist be acting as conduits for sexism or racism were they to make their choice on the basis of such calculations?

To gain some traction on these questions, in this column I will examine how the law treats such instances of what we might call indirect discrimination--decisions that defer to, or take into account, expected discrimination by others. I will then ask whether the law's general prohibition on such considerations should, as an ethical matter, carry over to the decisions of individual voters as well.

The Palmore Case and Title VII

The leading Supreme Court case on indirect discrimination is the 1981 ruling in Palmore v. Sidoti. After the Sidotis' marriage ended in divorce, the couple's daughter was placed in the custody of her mother, who, shortly thereafter, changed her name to Palmore, upon remarrying. All of the original members of the Sidoti family were white, but Ms. Palmore's new husband was an African-American man. Citing, among other things, the fact that children of inter-racial couples face prejudice and discrimination, Mr. Sidoti petitioned for a change in custody, which the Florida family court judge granted on that basis. Finding that both parents would provide a loving home, the judge ruled that it was nonetheless in the child's best interest to be raised in a family in which she would not face the added stress of racial prejudice.

When the case reached the U.S. Supreme Court, the Justices unanimously rejected the Florida judge's reasoning. In a terse, almost dismissive opinion for the Court, Chief Justice Burger acknowledged "a risk that a child living with a stepparent of a different race may be subject to a variety of pressures and stresses not present if the child were living with parents of the same racial or ethnic origin." Nonetheless, he said that the state could not make such private bias the basis for an official decision: "The Constitution cannot control such prejudices but neither can it tolerate them."

The Palmore principle is also reflected in Title VII of the 1964 Civil Rights Act, the core federal workplace anti-discrimination law. The federal courts have uniformly held that customer preferences for a certain kind of employee--waitresses instead of waiters, say, or vice-versa--do not excuse an employer from the duty not to discriminate on the statutorily-proscribed bases, unless the criterion is a so-called "bona fide occupational qualification" or "BFOQ." Even then, courts typically reject claims that customer preference can turn what would otherwise be mere bias--in favor of female flight attendants, say--into BFOQs.

Cases like Palmore and the Title VII customer preference decisions accept the reality of discrimination. Nonetheless, they construe the anti-discrimination mandate to forbid indirect as well as direct discrimination.

Why Palmore Was Actually a Hard Case

That resolution, however, is easier to justify in the Title VII cases than in a case like Palmore. When an airline is told by a judge that it cannot discriminate against male prospective flight attendants, then the airline risks alienating some of its customers, but suffers no substantial business harm, because the same rule applies to its competitors too. (To be sure, at the margin, some people who previously flew because of the attractive stewardesses might switch to automobile transportation, but it is difficult to imagine that this effect is substantial.) Thus, the cost of enforcing the anti-discrimination norm in the face of biased customer preferences is borne almost entirely by the biased customers whose preferences go unfulfilled, which is altogether appropriate.

By contrast, the Palmore case makes an innocent party--the former couple's minor child--bear the cost of the anti-discrimination rule. It does nothing to prevent the Palmores' intolerant neighbors from subjecting the child to discrimination.

Yet the alternative in Palmore was also problematic. Ms. Palmore was likewise an innocent party, and the Florida family court judge's decision to strip the girl's mother of custody because some of her neighbors were prejudiced against interracial couples would have converted an ugly private prejudice into an official act of the state.

Thus, despite the Supreme Court's unanimity, Palmore was, or should have been, a difficult case. The legal principle that the law should not place its imprimatur on discrimination may be fine in the abstract, but in some sense, that principle was going to be violated no matter who was awarded custody. To rule as the Florida family court judge did was indeed to bow to prejudice. But to rule as the U.S. Supreme Court did was--if we accept the factual premise that discrimination against interracial couples in 1981 Florida was real and substantial--to knowingly subject a minor to discrimination.

Indirect Discrimination in the Voting Booth

What lessons do these cases teach for the dilemma faced by this year's Democratic primary voters? In a strict sense, the answer of course is none. Neither Title VII nor the Equal Protection Clause--the legal basis for the holding in Palmore--applies to the decision about whom to vote for.

Nonetheless, these cases give effect to the same sorts of moral principles that one might properly think ought to govern our daily lives. Anybody who strives not to act in a racist or sexist manner has reason to worry over whether a decision to take account of others' racism or sexism violates a personal moral duty.

So, is a decision whether or not to vote for a particular candidate, given the likely impact of discriminatory votes by others, more like the Title VII case or more like Palmore? It appears to be more like the latter, in that there are no cost-free tradeoffs.

Suppose you think that Candidate O would make the best President but you also think that he would be a weaker general election candidate than C or E because of likely race discrimination against O. If you withhold your vote from O, then you become the anticipatory vehicle of race discrimination against O, and you deny your party the chance to nominate the candidate that you believe would be best for the job (to the extent that your vote, in combination with those of others thinking the same way, makes a difference).

Yet if you are right in your assessment of the likely conduct of general election voters, then a vote for O is a vote to cut off your nose to spite your face. If O becomes the party's standard-bearer, he will still face racially-discriminatory voting, but now it will result in the competing party's nominee taking office. As in Palmore, but unlike in the Title VII customer preference cases, there is no way to make the true culprits--here, those who vote based on racist motives--bear the cost of their actions.

Thus, there is no easy answer to the problem of other people's racism or (if you substitute C for O, and O for C, above) sexism--at least if we assume that other people will in fact be influenced by racist or sexist motives.

Of course, no one knows for sure how Senator Obama's race or Senator Clinton's sex will play out in the general election. One can easily imagine that excitement over either candidacy would so increase voter turnout in the general election as to overcome any negative effects based on race or sex. There is also genuine debate about whether the so-called "Bradley effect," in which some number of white voters tell pollsters they will vote for a black candidate but then do not, is even a real phenomenon. And thus far, Clinton's sex appears to be working to her advantage: Her teary-eyed moment in New Hampshire seems to have helped her candidacy in a way that it is hard to imagine a similar moment helping a male candidate. Previously, the most famous (supposed) electoral tears were those of Edmund Muskie, and they arguably cost him the nomination in 1972.

Moreover, even if one thinks that Obama's race and Clinton's sex would each be a small net negative in the general election, a Democratic primary voter choosing between them could conclude that these effects roughly cancel each other out. No one really knows whether the price Obama would pay with racist voters is greater than, equal to, or less than the price that Clinton would pay with sexist voters. And with the Edwards candidacy fading fast, there are no other realistic choices.

Accordingly, Democratic primary voters may find themselves excused from having to make the difficult choice between setting themselves up for the possibility of failure in the general election due to the prejudice of some voters, on one hand, and acting as an advance conduit for such prejudice, on the other.

In such circumstances, perhaps voters can overlook questions of electability and actually vote for the candidate they would most like to see become President. Imagine that.