We should have trusted the numbers.
On Monday we predicted, based on Lycos user searches, that the Republicans would retain control of the House and the Senate -- and that George W. Bush would win the popular vote for the presidency.
We were two-thirds right. The GOP did keep their majority in both houses of Congress. (Probably.)
But Al Gore won the popular vote by about 200,000 votes out of the roughly 98 million votes cast for him and Bush. (The electoral vote, as we pointed out Monday, is a whole different matter.) In other words, Gore beat Bush 50.002% to 49.998%.
Our Web search numbers actually had this right, correctly predicting a close race with Gore slightly in the lead. Gore held a thin search edge on Bush for eight of the past 10 weeks. Last week Gore had a 51.9 to 48.1% lead. Two weeks ago Gore's margin was 50.3% to 49.7% -- very close to the final vote.
But then we got fancy, trying to figure in the effect of Ralph Nader's candidacy and the heavy edge of GOP searches in general. That led us to predict Bush would edge Gore in the popular vote. We now see we should have butted out and let the numbers do the talking.
In general the Web seemed to exaggerate the power of Ralph Nader and the Green Party. As we noted on Tuesday, Nader pulled over 16% of searches in last week's three-way comparison with Bush and Gore. And the Green Party had nearly as many searches as the Democratic Party.
In actuality Nader and the Greens pulled only 2-3 % of actual votes. No doubt Nader's presence did influence the election, but not in the vast proportions indicated by searches.
Still, add our 2-for-3 election performance here to our 4-for-5 performance in this year's Academy Awards, and Lycos users are batting 75% on the big predictions. That's not bad.
And next time we'll trust the numbers.
For more coverage of this fascinating election, please visit our special Lycos Vote 2000 site or the political mavens from Slate.
