Results in the Pennsylvania exit poll, not available from previous contests, underscore a potentially sensitive point should Barack Obama win the Democratic party’s presidential nomination: The role of race in voters’ decisions.
In Pennsylvania, as in previous primaries, white voters who rated the candidates’ race as an important factor in their vote were more apt to support Hillary Clinton, compared with those who said race wasn’t important. An open question has been whether this is best seen simply as affinity voting – whites more apt to support the white candidate, akin to women supporting the female candidate and blacks the African-American candidate – or as something less benign.
The new element is the matchup of Obama and John McCain in a general election trial heat in the Pennsylvania exit poll. The result: Among whites who called race an important factor, just 54 percent said they’d support Obama vs. McCain; the rest either went for McCain (27 percent) or said they wouldn’t vote. That compares with Obama’s much higher support – 72 percent – among whites who said race was not important in their vote.
We should be careful not to overanalyze this result. The number of whites who called race important was small – 13 percent of Pennsylvania voters – and the net effect of the differential is to cost Obama 2 percentage points of total vote, while giving McCain around a point and a half.
This may signal a problem for the Democrats if Obama's the nominee; after all, these results are among voters in a closed Democratic primary – people presumably loyal to the party. On the other hand, these Democrats were caught smack in the midst of their intramural food fight, before the general election campaign has begun. Looking ahead to November is fundamentally premature.
One reason the effect of race does bear watching – beyond the results themselves – is that they’re not replicated in similar measures of the impact of the candidates’ sex. Clinton has done better overall with women who say the sex of the candidates is important in their vote – 26 points better in Pennsylvania, about the average in all primaries to date. (Men who call the candidates' sex important also have been more apt to support Clinton, by 9 points.) But here's the crucial difference: whether Pennsylvania voters said sex was important or not important in their vote did not materially affect their preference in a Clinton-McCain matchup. That makes race, and its impact on an Obama-McCain contest, look like a different kettle of fish.
In a related point, the Pennsylvania data suggest that white voters who saw Clinton as not honest were about 20 points more apt to support her over Obama anyway, if they rated race as an important issue (compared with those who saw her as not honest, but said race was not important). But in this equation sex is a factor as well: Women who saw Clinton as not honest likewise were about 20 points more apt to support her if they rated the candidates' sex as important. In any case, these are very small groups – 4 percent of voters in each case – so the net effect again was very small, and not at all decisive.
As far as we can see, the impact of race as a self-described “important factor” in vote preference has not tipped the balance in any primary to date (with the possible exception of the extremely close contest in New Mexico). Nonetheless, the discussion of race, and sex alike, raises questions of how best these effects are measured and how best to understand them – in effect, how to parse out positively influenced voting behavior from negatively influenced voting.
Any way you cut them, the Pennsylvania results are a reminder of the complexities involved – and of the scrutiny the role of race likely will receive if it's Obama who leads his party into the November election.