Article excerpt


Fort Lauderdale, Fla.

* Anna Greenberg was directly on point with her thoughtful piece, "Why the Polls Were Wrong" [Jan. 1]. As someone who lives in the heart of the south Florida battleground, in Democratic-vote-rich Broward County, I would also provide an addendum. Just as the major polls were wrong for the reasons cited by Greenberg, let's also credit the election night Voter News Service with being accurate. While Republicans like James Baker decry the networks' "erroneous" initial call for Gore, all one has to do--besides debate recounts, chads and hanging chads--is to factor in the 22,400 intended Gore votes in Palm Beach County, apportioned between overvotes for President and erroneously cast votes for Pat Buchanan. These people, whose votes did not ultimately count, were certainly factored into projected totals favoring Gore on election night. As Greenberg asserted, it was the second call, which went to Bush early Wednesday morning, that was in error. Of course, Baker and the Republican high command touted that projection as accurate.

William Hare

New York City

* It was not the polls that got it wrong this election. It was Anna Greenberg. The national pre-election polls reported by the media had their best polling performance since 1976. All ten traditional polls were within the margin of error. These polls had an average error on the winner of 1.1 percent. In the election Bush and Gore were almost dead even. Two polls had Gore ahead slightly, six had Bush ahead slightly, one was off by an average of 2.5 points each on Bush and on Gore and the Harris poll had the race even. There was one other poll that was significantly off. It used experimental interviewing methods and was not reported widely by the media. Rasmussen conducted it and issued an apology after the election.

Contrary to Greenberg's assertion, only two polls "'weighted up' the GOP share" to match the parties' usual share of the vote. Greenberg complimented one of those polls for its fine performance. None of the polls got the voter turnout wrong. All used the same methods they have used for years to assess likely voters.

Linking the pre-election polls to the networks' mistaken call of Bush as the winner in Florida on election night is just plain silly. None of the pollsters involved in the pre-election polls had any connection to the election night projections. The Bush projection was made from the 97 percent of the vote that had been counted by Florida voting officials.

One assertion Greenberg makes is disingenuous. The "internal Gore polling" she uses as a benchmark for what the public polls should have shown (a) was not available to the public, and therefore not subject to peer scrutiny; (b) was not constant over the last two weeks as she asserts; and (c) was partially the work of her father. The Nation should not be citing private partisan political polling as a standard against which to measure the public polls. The public polls disclose their methods, data and results. The partisan polls do none of these things.

Polling Review Board
National Council on Public Polls


Cambridge, Mass.

* I thank William Hare for his letter. Mitofsky, Taylor and O'Neill correctly note, as I did in my editorial, that many of the national poll results released on or close to Election Day fell within the margin of error. First of all, using the margin of error as the measure of accuracy is rather narrow, given the great volatility of the tracking polls and the wide variation among the polls during this election season. Second, the margin of error is somewhat tangential to the larger point that most of the national public polls favored Bush and that this perception of Bush's ascendancy influenced press coverage of the race prior to and after the election. Finally, while there were no official ties between these national public polls and the exit polls conducted by Voter News Service (a link I did not make in my editorial), to argue that the collective weight of these surveys did not influence press coverage of the election is also a bit silly. …