White Lies, Bad Polls

Before last week's unexpectedly close Virginia contest, Pollster Harrison Hickman got revealing results by making an offbeat correlation. When white voters were questioned by white pollsters, Hickman found, they favored Republican Marshall Coleman by 16 points. But when whites were telephoned by interviewers with recognizably black intonation, they leaned to Douglas Wilder by 10 points.

The fact that Americans are notoriously unreliable when answering questions related to race was dramatically evident in the Virginia and New York City elections. Although several surveys in the final fortnight gave Wilder and David Dinkins comfortable leads (as high as 15 points for Wilder and 18 points for Dinkins), both contests turned out to be squeakers.

The phenomenon is not new: seven years ago, Los Angeles Mayor Tom Bradley seemed to be leading in California's gubernatorial election -- until the ballots were counted and he lost by less than a point. Some whites were reluctant to admit to pollsters that they planned to vote against a black.

Racism in the crude sense does not necessarily motivate people to misinform pollsters, Hickman says. Rather, some respondents succumb to a misguided urge to give answers they think will please the questioner. Whatever the reason, pollsters in black-white contests should learn to take the discrepancy into account -- at least until such racial match-ups cease to be novelties.