Media’s Misuse of Polls Distorts Presidential Coverage
Journalists and analysts say the presidential campaign doesn’t begin until Labor Day; however, if the last two elections are guides, the election will already be decided by then or shortly thereafter. In 1988 and 1992, news analyses were more intent on sustaining the perception that the outcome was still in doubt. In particular, voters were misled by national polls, which suggested the races were tightening to a far greater extent than was true. Moreover, because of wildly oscillating results, and a general misunderstanding of the nature of the surveys, voters were more likely to be confused than informed by campaign coverage. Unfortunately, the media is repeating the mistakes of those campaigns.
Journalists consider it irresponsible to declare the race over even if the evidence shows the candidate doesn’t have a chance; therefore, they prefer to maintain the fiction that the race is in doubt to stimulate ongoing interest in their reports. Their tool is the “horse race” poll, a national survey that purports to show which candidate is leading and by what margin. The state by state polls in 1988 and 1992 allowed the media to report that, based on electoral vote projections, the race for the Presidency was over weeks before the election, but the press relied instead on the national polls that suggested otherwise.
In 1988, the outcome was clear by mid-September, but it was not until October 12, 1988, that a network reported that Republicans had a lock on enough electoral college votes to win. About this time, Bush held double-digit leads in states where 70 percent of the votes would be cast and 330 electoral votes decided. Despite the obvious landslide that was developing, ABC was vilified for suggesting the race was over when nearly a month of the campaign remained. This reaction had a chilling effect on all the media’s coverage from that point on, as journalists qualified every report to avoid the appearance of writing off Michael Dukakis.
In 1992, Bill Clinton never had fewer than 360 electoral votes in state polls (with at least 40 reporting). He had a “lock” (a double-digit lead) in states worth no less than 214 electoral votes until the day before the election. A month before, Clinton led in 38 of the 48 state polls, worth 447 electoral votes, and had a lock on 265.
Complications began to arise in late October when national polls showed Clinton’s lead shrinking to as little as 1 point. Reporting dramatic shifts in the margin erroneously implied the race was significantly narrowing at a time when state polls indicated Clinton still had a lock on well over 200 electoral votes.
As in 1988, the head-to-head “horse race” polls dominated coverage in 1992. These updates gave the public a false sense that the gap between the candidates fluctuated. In fact, the lack of volatility was apparent in the electoral vote projections.
Except election day and two early summaries that contained surveys from less than 35 states, Clinton’s expected electoral vote total never dropped below 350. Rather than oscillating, the electoral vote totals illustrated how Clinton started with a huge lead in early September and built it through the middle of October. He did not lose ground until the last four days of the campaign, and, even then, the decline was not serious enough to raise doubts about the coming landslide.
Perhaps the best example of how national polls presented a misleading picture of the election was the attention given to Ross Perot. Fluctuations in Perot’s standing produced suggestions that he could be elected or significantly influence the race. In fact, Perot was not a threat to win a single state. He never received more than 28 percent in a state poll. Even in that survey, conducted in Idaho, he trailed by 7 points. Perot’s impact was limited to a handful of states where he forced one or both candidates to focus more attention than they otherwise would have to insure that voters they needed did not defect to the independent candidate.
It is certainly a different matter for the press to predict a winner of the election in September than on election night, but it is also hypocritical to claim that exit polls have no impact on election day but could influence voters earlier. Accurately reporting that a candidate has the race locked up might energize the electorate and the campaigns rather than discourage turnout. The reason for not doing this seems to have less to do with feelings of civic or journalistic responsibility than fear of generating resentment.
The concern of appearing too sure of the outcome leads reporters to use more qualifiers than would normally be acceptable in news stories. When the press reports on how candidates are doing in the states, journalists rarely explain how they determine who is ahead or what constitutes having a state “locked up.” At most, reports say a candidate is leading in that state’s polls. But there are leads and there are leads.
The use of state polls has drawbacks; for example, they tend to be conducted less frequently and therefore do not necessarily reflect current opinion. Nevertheless, the tabulation of state results is a far more accurate indication of the closeness of the race than the national surveys.
If all the data show a candidate is far ahead, and is not likely to be defeated, campaign reporters should say so. It is no more responsible to suggest a race is close when state surveys indicate the opposite than to say a candidate is expected to win.
Instead of telling us about bumps, blips and bounces, the press would do a greater service by simply reporting how the candidates stand where it counts — in the states.