The Women's Freedom Network Newsletter November/December, 2002;  Volume 9, No. 6 Pundits, Polls & Projections by Daniel Dreisbach

Why did media pundits and pollsters misread the polls leading up to November 5th? Or, did they read the polls accurately and simply ignored, disbelieved, or misreported their findings? The media polls were either accurate and they misinterpreted, mistrusted, or misreported the data, or the polls simply failed to detect the Republican trend. All these factors, no doubt, were partially responsible for the mainstream media's disastrous coverage of the mid-term elections. The public should find little comfort in any of these explanations.

Both independent and partisan pollsters reported that, by the Friday before Election Day, their polls were showing that the undecided voters were trending Republican and that Republicans could expect a 2 to 4 seat gain in the Senate. This was not reported in the national media. Why not? Was it because the media misinterpreted or mistrusted the data, they did not want to undermine the political drama (suspense boosts election night ratings), or a Republican sweep was not what they wanted to hear and report?

A number of Republican pollsters complained loudly that their internal numbers throughout the campaign bore little resemblance to media generated data. For example, the internal polling by Wayne Allard, Colorado's incumbent Republican Senator, consistently showed daylight between the Senator and his Democratic challenger, numbers confirmed by the 6-point margin of victory on November 5th. Yet, the media routinely described the contest as "too close to call," a "toss up," or a "possible Democratic pick-up." Republican Senate campaigns in North Carolina, South Carolina, Texas, and, in the state that turned the biggest surprise of the night, Georgia, similarly complained that the media reported that these races were much closer than they actually were. The consistent disparity between internal and media generated numbers prompted a number of campaigns to check and recheck their methodologies, practices, and computations only to conclude, as the official ballot tallies confirmed, that their internal figures were more reliable than the media's.

Why did the media pollsters miss or fail to report the attitudes and shifts evident to internal campaign polls?

The networks raised additional suspicions about their objectivity this year, as on election night 2000, in their pattern of projecting winners. In 2000, the networks routinely projected narrow Democratic victories hours before projecting comfortable Republican wins. The most egregious example, of course, was Florida, where major networks erroneously projected Al Gore the victor before polls closed in the western panhandle. By contrast, it took the networks 2 and 3 hours after poll closings to project George Bush the winner in states like Colorado, Arizona, Arkansas, and West Virginia, which he won handily by 6 to 10 point margins.

The networks were unusually cautious in projecting winners in 2002, largely because of the failure of VNA; nevertheless, similar patterns of "projection bias" emerged in this election as in the last. In Arkansas, for example, where the incumbent Republican governor won his race by a similar margin as a Democratic senate challenger defeated the incumbent Republican, a couple major networks projected a Democratic pick-up in the Senate hours before projecting a Republican gubernatorial hold. Why? Is this empirically verifiable of that elusive beast known as media bias, or is there some benign explanation? I do not know the answer, but the apparent pattern of projecting Democratic wins long before Republican victories deserves an explanation.

Once gain, an election night leaves the media and the public with unanswered questions about the mainstream media's coverage of partisan politics. Should the American public trust the media to report political news fairly and accurately? The question should be asked and deserves to be answered by those who deliver the news.