5/31/00 6:45 p.m.
Lies, Damned Lies and the Washington Post
How not to predict the results of an election.

Robert A. George is an editorial page writer
for the New York Post------------------------------------RAGGEDmail@aol.com

t was the Friday of Memorial Day weekend, so the front page of that Fount of Beltway Wisdom — the Washington Post — might have escaped under the radar: "Is This Any Way to Pick A Winner?" was the title of Robert Kaiser's piece. The gist of it was that, as far as leading political scientists are concerned, this election is over. All those polls showing George W. Bush with various leads over the vice president? Fugeddaboutit!! From mathematical models of presidential elections going back to 1952 (caveat number one), the "experts" tell Kaiser that Gore will win between 53 and 60 percent of the two-party vote (another major caveat, but more on that in a bit). The pointy-heads look at such "objective" factors as popularity of the incumbent president and the performance of the economy in the quarter preceding election day.

This story is interesting on several levels. The immediate conservative reaction is that this is another classic case of liberal media bias. If the Post had the need to run a "political analysis" piece, why not do a story on the fact that Bush has had a lead in almost every single poll since Super Tuesday? Why run a story talking about mathematical models anyway? And people wonder why right-of-center individuals question the "objectivity" of the press.

Another question: Why put this non-news story on Page One? Why not stick it in the middle of the "A" section — or better yet, put it in the Style section?

Yes, the role of political scientists gives the study a veneer of academic "objectivity" — thus making the case that it should be up front.

But a strong case could be made that these numbers are actually LESS "scientific" than the average poll. At least today's voters have their own set of objective factors (e.g., the impact of a president behaving with in a sexually reckless manner and then lying to the country about it) to take into account — rather than using yesterday's factors and applying them to today's circumstances.

"It's not even going to be close," asserts Michael Lewis-Beck, predicting that Gore will get more than 56 percent of the two-party vote. Lewis-Beck apparently accurately forecast the Clinton-Dole election within a fraction of a percentage point. If you're beginning to think this sounds like an Enquirer story that introduces a psychic that "accurately predicted the crash of TWA 800," you're beginning to catch on. The "two-party" line is notable because the academics Kaiser interviews dismiss the impact of the "third" and "fourth" parties. Under this model, they would say that Clinton would have beaten Bush in 1992, whether Perot was in the race or not. An arguable point, but let's give it to them anyway.

Aside from it being a media sloppy wet kiss to perk up the Gore campaign, the major problem with this story is the lack of intellectual rigor these political scientists put into their analysis. Christopher Wleizen of the University of Houston (he nearly perfectly pegged Bush Sr.'s vote in 1988) admits that his formula would have underestimated the challenging candidate's vote in 1960, 1968, and 1976.

He attributes this to undercounting the "time for a change" factor. Now, those three contests are interesting: 1960 is widely seen as one of the most corrupt elections in history. In 1968, the sitting president dropped out of the race AND there was a major third-party challenge which ripped apart the Democratic base (far more so than Perot's effort did in 1992 to Republicans). In 1976, a presidential resignation and a non-elected president running had to be major factors.

But these are not mentioned by any of the political scientists in this article. For these people, numbers have to tell the whole story. They can't lie — which is a lie in and of itself. Numbers can be interpreted in any way you wish depending upon what other variables you introduce. The people Kaiser speaks to take their formulas back to 1952. Interestingly, of the ten presidential elections between that year and 1992, Republicans won seven. Of the three Democrats who won, the aforementioned 1960 and 1976 were by only the slimmest of margins. That's an interesting pattern. But fascinatingly enough, every election (presidential or otherwise) since 1992 has repudiated some aspect of conventional political wisdom or accepted truth.

Clinton's first victory ended the so-called "lock" that had put a Republican in the White House for 20 of the previous 24 years. The 1994 election ended the Democrats' four-decade control of Congress. 1996 was the first time a Democrat President and a Republican Congress were reelected simultaneously. Finally, 1998 was the first time in more than 60 years that the party controlling the White House won seats in Congress.

Now, given these objective factors from the last several years, why would political scientists not consider that the upcoming presidential election might not be a slam-dunk for Al Gore? Better yet, why would a politically-focused paper such as the Washington Post not ask such a question, or factor it into a critique of these mathematical models?

Well, that's the question that brings us full circle.

Couldn't be that ol' media-bias devil, now could it?