On the Sunday before the vote in the Massachusetts Senate race, my polling firm, InsiderAdvantage, conducted a survey for a prominent Washington-based news organization. The poll was released the next day, Monday, the day before the election. The results made waves by showing Republican Scott Brown with 52 percent of the vote, which is exactly what he ended up with. The poll ran in various other news organizations around the nation. And that's when I started taking a beating.Unlike many pollsters, we release all of the internal numbers that show how various demographic groups plan to vote -- men and women, different races, various age groups and people of different political persuasions. Our poll, heaven forbid, showed now Sen.-Elect Brown's opponent, Martha Coakley, with 43 percent of the vote. (She got 47 percent.) The poll also had 2 percent going to the third candidate, one Joseph Kennedy, who is not related to the famous Kennedy clan. The remainder of poll respondents said they were undecided.
Understand that Coakley had a terrible time on the campaign trail the weekend that culminated with our Sunday poll. She campaigned with Ted Kennedy's son, who reportedly kept calling her "Marsha" rather than Martha. And she really stepped in it when she inadvertently insulted, of all people, Boston Red Sox fans. She blurted out that Sox legend Curt Shilling was a New York Yankees fan!
Polls are accurate most of the time, even though some will never accept that. Even so, polling is like watching sausage being made. The internal components that yield the final polling results are sometimes hard to stomach.
As soon as our poll was released, two things happened. First, I unexpectedly fell very ill. I mean, can't-lift-your-head-off-the-pillow ill. It lasted a week, and I lost 10 pounds. But it was the second happening that really made me sick.
The good people at the publication that carried our poll were attacked viciously by the Democratic National Committee and by rival media. The attacks included some valid points, although the same attacks also proved the biases that many have in making assumptions about certain demographic groups.