Polling the debates

Posted: Oct 12, 2004 12:00 AM

 Last Friday night, George W. Bush bounced back with a more warm, humorous and passionate performance in the second of three presidential debates than he displayed in the first one. But a "scientific" post-debate poll, conducted immediately afterward by ABC News, gave the nation a different -- and arguably flawed -- view of the public's reaction. That causes me to wonder whether much polling serves to create self-fulfilling outcomes.
As a former debater, I freely admit that on an argument-by-argument basis, the second Bush-Kerry square-off was probably a draw. But this time, it wasn't just the president who found himself occasionally tongue-tied. Kerry tripped up a bit on abortion, for example.

 I have little doubt that Kerry trounced Bush in the first debate. I am equally confident that Bush won the second one. The main reason was the president's rediscovered warmth. Yes, that admittedly un-cerebral notion that people want a leader who comes across as natural, from-the-gut, friendly -- even funny. Bush balanced passion and aggression to depict himself as a forceful leader who can still be lighthearted and spontaneous, even in the heaviest of moments.

 When Kerry lambasted Bush for -- apparently according to the president's tax returns -- being a minor investor in a timber company, the president said, "I own a timber company? That's news to me." While the audience was still laughing, Bush made another quip asking if anyone in the audience wanted to buy a little wood. On that night, in that setting -- and doubtless helped by low expectations -- Bush bettered the more polished Kerry on style alone.

 Then came ABC's instant post-debate survey. The poll was released before the network's TV coverage of the debate even ended. It showed Kerry having won by three percentage points. To ABC's credit, they quickly noted that the poll had more Democratic respondents than Republican. And anchor Peter Jennings and commentator George Stephanopoulos both basically called it a tie. Stephanopoulos even went so far as to point out that the three-point differential essentially mirrored the same difference in Democrats and Republicans who were queried in the quick survey. Taking it a step further, ABC's political director Mark Halperin made overtures about a Bush bounce-back from the first debate.

 Even so, the poll in itself seemed inherently flawed. In most scientific, public political surveys, the raw numbers first are collected. If there are more Republicans or Democrats in the survey response -- as is usually the case -- the numbers are then adjusted, or "weighted," to accurately reflect the actual split between the two parties in previous surveys. In most major polls, the percentage of Americans who identify themselves as Republican or Democrats is virtually even -- hence the cliffhanger election of 2000. Other voters, a usually quite large segment dubbed "independents," have no stated allegiance to either party.

 Had the ABC poll been adjusted for just this one variable -- leave aside many other common ones like age, gender, etc. -- the results likely would have shown Bush either even with or ahead of Kerry for that one debate.

 Here's the key: Because this ABC poll was the first to be publicized, it quickly made its way through the media and political spin rooms and onto TV screens and newspaper headlines that night and the next morning.

 That's where the flaw in this process lies. Polls can become self-fulfilling prophecies for the actual results of elections. Sure enough, several more national polls on the debate were released over this last weekend. Most showed Kerry not only winning the second debate, but now either tied or winning the race for the White House.

 One must wonder what impact the distilled 15 seconds of the "early polls on the debate" had on the millions of potential voters who didn't watch it, but instead got their impressions from filtered sources known as newscasts.

 This is similar to the effect pollster John Zogby had last winter in Iowa, when -- out of the blue -- his survey showed the largely unknown John Kerry suddenly, almost magically, separating himself from Howard Dean and the pack of other, sometimes better-known Democrats.

 How and why did this happen? The answers are almost as hard to fathom as Zogby's flat-out declaration in May that Kerry would be the next president. Given that, can anyone be surprised that Mr. Zogby's poll earlier this week has Kerry surging to a three-point lead in the race for president?

 Our own polls are not perfect, and I constantly face criticism of them, just as I am doling out criticism myself in this column. Our latest InsiderAdvantage survey showed Bush with a smaller lead in Florida than other polls released at the same time. But it's one thing to release results that vary to match varying public sentiments as days and weeks follow each other. It's something else altogether to release polls that just so happen to vindicate polls and outright predictions made months earlier. Or worse, to release quick, "scientific" ratings immediately following a monumental presidential debate.