Faulty Media Polls Fuel Iowa 'Expectations' Game

Change font size:  A |  A

As you’ve already been told a thousand times, with only a day to go before the Iowa caucuses, the polls are showing a statistical three-way tie between Barack Obama, Hillary Clinton and John Edwards atop the Democratic field, and a similarly close two-way race between Republicans Mike Huckabee and Mitt Romney.

But the polls are probably wrong. Or maybe they’re right -- we won’t really know until Thursday night when the actual results are announced. And that’s the problem -- the media have given the polls so much emphasis that the actual results will only matter to the extent that they differ from the media’s pre-election expectations, i.e., only to the extent that this week’s polls are inaccurate.

In just the last month, RealClearPolitics has posted the results of 55 pre-Iowa caucus polls (27 for the Republicans, 28 for the Democrats). These are mostly media-generated polls, with a few conducted by universities. It’s because of these polls that reporters think they know who is and is not a frontrunner, who is and is not rising and/or falling, and who is and is not hopelessly behind.

But four years ago, the media read the polls as predicting a tight-four way race in Iowa’s Democratic caucuses. On NBC’s Today on January 18, 2004, the morning before the Iowa caucuses, newsreader Rehema Ellis summarized the “too close to call” conventional wisdom: “In Iowa, Democrats running for president head into the home stretch before tomorrow’s caucus meetings. It is a close race. The latest MSNBC/Reuters-Zogby has Senator John Kerry and former Governor Howard Dean picking up one point. Kerry now has 24 percent, and Dean 23 percent support among those polled. Congressman Richard Gephardt and Senator John Edwards have 19 and 18 percent, respectively.”

On the same program, Washington bureau chief Tim Russert declared, “It’s impossible to predict this race. It is so fluid. It is extraordinary to watch as these candidates are now campaigning 18 hours a day. It is a wide-open race. Any of the four could win.” The other networks offered similar pre-caucus predictions.

The actual result was not even close to a four-way tie. John Kerry took 38 percent of the vote, 58% more than the MSNBC poll predicted, while John Edwards took 32 percent, 78% more than he was supposed to get. On the other hand, Howard Dean received 18 percent of the vote, 22% less than predicted, while Dick Gephardt received 11 percent, 42% less than the survey said.

And you know the history of how the results were interpreted. Kerry had a “smashing and surprising win” (ABC’s Good Morning America) and a “huge win” (CBS’s The Early Show), while NBC’s Today christened Kerry “the big winner in Iowa.” The networks were also impressed with John Edwards’ showing, while suggesting Howard Dean had a “miserable” night (ABC).

But suppose the media’s polls had shown Kerry with somewhere between 35 and 40 percent of the vote for several days before the caucuses? Indeed, suppose that every poll in the month before the caucuses were held were entirely predictive of the results, right down to the last percentage point. The networks would probably have been pretty unimpressed with the results, and declared themselves ready to move on to New Hampshire.

It’s not the case that pollsters are ridiculously awful at polling; it’s obvious that many voters either decided late in the game or changed their minds. But that fact only underscores the uselessness of polling a group of people who are still in the process of deciding -- the results are bound to be wrong and shouldn’t be touted as “news.” Yet the media persist in using polls to establish “expectations” for each candidate, rewarding those candidates who “beat expectations” and punishing those who fall short.

If the history of actual caucus results is a reliable guide, Thursday night’s results will differ significantly from the pre-caucus polls. Within minutes of hearing the outcome, journalists will impute great meaning and significance to a candidate’s “better-than-expected” or “worse-than-expected” performance.

Manipulating the media’s expectations -- it’s how losers such as Gary Hart in the 1984 Iowa caucuses, Bill Clinton in the 1992 New Hampshire primary and John Edwards in the 2004 Iowa caucuses manage to re-cast themselves as an election’s true winners. And the degree to which these candidates benefit in future primaries and caucuses shows the inordinate power that the media have in today’s primary system.

It’s entirely unrealistic, of course, to hope journalists will ever refrain from making their own “expectations” the key to a candidate’s success or failure. But it would be nice to see the networks organize their campaign coverage around the substantive issue debates that distinguish the candidates from each other, not a cornucopia of polls that are as worthless as rotten fruit on election day.