Addicted to Polls
Parsing the political-numbers watch.
By Gwen Brown
I have a confession to make: I've become an addict. Every morning, I fly to the computer to check the latest news and opinion about the presidential campaign. What concerns me is that my first stop — after NRO, of course — is always RealClearPolitics.com, to which I return throughout the day. I neeeeeeeed to find out what the latest polls are indicating. Who is up or down in the latest ABC/Washington Post poll? CNN/USA Today/Gallup? Zogby? Fox News? Rasmussen? What's the average of all the latest polls? How is the race going in the battleground states? Florida? Ohio? Pennsylvania? Wisconsin? I'm such an addict that "Quinnipiac" now rolls easily off my tongue. In fact, my addiction has become so great that I fear I will need a twelve-step program starting on November 3.
Maybe I've already taken the first of those twelve steps by slowing my speed-browsing — and my breathless hunt for the day's numbers — and by thinking about what I'll teach my students in future about the role that polling played or had the potential to play in this year's presidential election. There will certainly be the usual topics: how polls are constructed, how questions are worded, the size of a sample, what a margin of error means, how a tracking poll differs from others, etc. But this election cycle has also produced new topics that I'll raise with my students.
First, I wonder how many of those called by pollsters are actually responding to the queries. I'm not thinking of the hang-up rate by those whose dinner is disturbed, but rather by those who harbor suspicions that pollsters are biased and thus decide to opt out of participating. A Washington Times story last Friday assured me that major polling organizations have "impressive track records of accurately predicting the outcomes of presidential elections," but I wonder how they'll fare this year. This election season has seen its share of widely publicized accusations of media bias — the CBS/forged documents story, the ABC/Mark Halperin memo, and most recently, the Sinclair Group's decision to air portions of the anti-Kerry documentary Stolen Honor. When a caller identifies him- or herself as representing the ABC/Washington Post poll, or the CBS/New York Times poll, are there some people already primed to consider the caller tied to a media outlet with demonstrated ideological preferences? In the eyes of some voters, has the Gallup organization tarnished its longstanding reputation for quality polling by linking its efforts with CNN and USA Today? Does the Fox News Dynamic Opinion poll suffer the stereotype of representing a news organization with a "conservative bias"?
A second question concerns the definition of a likely voter. I've always told my students that a poll focusing on likely voters is probably more reliable than one focusing on registered voters. Some polling organizations simply ask how likely a person is to vote and then rely on the response as meaningful or trustworthy. But might some respondents identify themselves as highly likely to vote simply out of a desire to be seen as doing their civic duty? Other polling organizations use a series of screening questions to determine for themselves whether the respondents are likely to vote, including whether they voted in the last election. This may be a better method, but it still depends on the pollster's blind trust in the truthfulness of the answer, since it cannot be independently verified.
Polls have proliferated in recent years, and they do not all use the same procedure to distinguish likely from registered voters. Moreover, as pollster John McLaughlin told Brit Hume of Fox News last week, the 2000 campaign may have provided a reason to change how the distinction is made. The number of those voting in important states such as Florida was up in 2000, and if the prognosticators are correct about the competitiveness of this election, the same may occur this year. Focusing heavily on likely voters may not result in the most accurate account of opinion, particularly if the "screen" is whether an individual voted in the last election. If more voters than ever are being registered this year — as we have been told by both parties — will it be harder to sort out the likely from the registered voters? No single screening mechanism is likely to be perfect.
A third question is about the influence of the media on polling. The constant refrain is that this is a "very close" race, that we are a "divided nation," and that the election at this point is "highly emotional," "volatile," and "polarizing." What effect, if any, does this have on those called upon to share their views with a pollster? At a time when we are told that we are a "divided nation," is it easier — and seemingly less confrontational — to refuse a pollster's request for an opinion?
With respect to the media's behavior, we will surely be treated to the quadrennial retrospective of media organizations deconstructing their coverage of the campaign and bemoaning the fact that their focus was so heavily on the "horserace." Yet true to form, the "horserace" is practically all we hear about at this point. The same media stars who dissected this shortcoming after the last elections on sober think-tank roundtables broadcast on C-SPAN act this year as though they never had those conversations. If respondents are reluctant to part with an opinion this year, it might be because they're exhausted by the media's typical coverage: "It's Bush by a nose!" "No, it's Kerry by a whisker!" And if the next presidential election is as close as this one seems to be, should we experience even more reticence from respondents?
A final question I will want to ask my students concerns what communication scholar Elisabeth Noelle-Neumann referred to as the "spiral of silence." Noelle-Neumann argued that we tend to conceal our preferences when we believe we are in the minority. The question that will reveal the most is therefore, "Regardless of your personal opinion, do you think most people [fill in the blank]?" Instead of asking about a personal preference, one should ask about the person's prediction of an outcome. Noelle-Neumann's research indicates a strong relationship between people's predictions and the actual outcome. In fact, some pollsters are currently asking a version of this question regarding the presidential campaign: "Who do you think will win the presidential election?" In response to just such a question, the Fox News Opinion Dynamics poll, for example, found that 48 percent predict a Bush win while 31 percent predict a Kerry victory.
Thus we might be experiencing a variation of the "spiral of silence" in this race. There is a widespread belief that this is a very nasty political campaign. News stories abound of local campaign headquarters experiencing theft and vandalism, of participants in the early-voting Florida precincts reporting voter intimidation, of the theft or destruction of yard signs, and of damage to cars with campaign bumper stickers. Apprised by the media that they live not only in a "divided nation" where the race is "too close to call," but also in one where election anger erupts in incidents of vandalism and intimidation, might some voters just hunker down into silence and refuse to express their own preference, even to a pollster they don't know and are likely never to meet? Perhaps some would be willing to predict an outcome, but are censoring themselves when it comes to stating their own wishes. If this is the case, just how accurate are the polls?
There are times when I remember that my poll-consumption habit is an unhealthy one, and needs to be tempered by a reality check. But it's a habit that is awfully hard to break, and Election Day will not mark an end to my addiction. After all, the next day someone starts polling on who the likely contenders will be in 2008, and I'll be looking for those early numbers. I'm a hopeless politics junkie. For now, though, I should take a tip from my students who have a different preferred poison. I can just think of Election Day as the close of the first season of Survivor — and look forward to next.
— Gwen Brown is associate professor of communication at Radford University in Virginia.
nationalreview.com |
