Florida presidential math still puzzles Times

'Where the spirit of the Lord is, there is liberty'

ConantNews and Znewz1 are site names used in the past by Paul Conant. This article was posted elsewhere in November 2001.

• ConantNews upholds freedom of:
• speech and press
• academic and theological inquiry

Nov. 13, 2001--Fuzzy math. That's how I grade the New York Times report, published Nov. 11, of a recount of 175,000 Florida presidential ballots.

From a mathematical standpoint, the Times did not fully justify its conclusion that the Supreme Court had not cast the deciding ballot in the presidential race.

The most obvious problems:

*The error rate for the recount, conducted on behalf of a group of media organizations, was not reported in my edition, though the Times devoted two full interior pages to the story.

*The estimated error rate for the statewide count was not given explicitly, though it was referred to implicitly.

I have no doubt that the error rate for the media recount was low, because, from what I can gather, essentially three counts were done, using different talliers, and the result was an average of the three counts. Though such a technique is likely to greatly reduce the error rate, it also implies that some error is inevitable in such a count.

Nevertheless, we cannot tell whether the difference between the official count and the recount falls inside the error rate. If so, then the recount would be a statistical draw. In other words, we still wouldn't know who actually won. It is hard for the nonmathematical to accept that there may be limits to the accuracy of human knowledge.

But let us, for the sake of argument, grant that that vote difference was above the error rate so that confidence is high that Bush prevailed for those disputed votes.

Though the Times asserts that, depending on tallying techniques, statewide figures showed Gore might have won, the paper did not clearly state what its experts calculated the statewide official count error rate to be. Previous reporting showed that the official statewide difference between Bush and Gore was within that error rate -- an average of the error rates of different balloting methods -- which was put at about 1 percent. It seems quite unlikely that the figures from the media recount favoring Bush would have changed the statewide difference sufficiently to overcome the statewide margin of error.

That is, if the media recount proves that the Florida outcome is no longer a statistical dead heat in which it is impossible to know who won, the Times did not make a clear-cut case for that theory. Hence, in the broad sense, it appears that the Supreme Court did cast the deciding ballot, though I suppose that, in a vacuous sense, one can defend the Times' claim.