November 9, 2004 · Lessig
The six news organizations at the left contracted with two polling organizations at the right to provide exit poll information one week ago today. Those data were inconsistent with the actual results — significantly so. Dick Morris says that “this was no mere mistake. Exit polls cannot be as wrong across the board as they were on election night. I suspect foul play.” The aim of the evil doing in Morris’ judgment was to suppress Bush votes — which it apparently didn’t. But the same data are used by skeptics on the other side, to suggest that those who had a hand in tallying the vote — in particular, one company whose President had promised to deliver Ohio for the President — changed the votes that the exit poll surveyed.
I think both claims are bunk — I don’t think there was a conspiracy to suppress the Bush vote, nor do I think Diebold stole the election for Bush — but there are obvious puzzles that need to be resolved. First, there is Morris’ point — exit polls are just not that wrong. Second, there are the insanely inverted county votes in the many heavily Democratic counties in Florida that had their votes counted by optical scan (and tallied by Diebold machines among others). Why were the polls so bad? Why did Democrats in those counties overwhelmingly defect to the President while remaining “liberal” in their other votes?
These are questions of fact that can be answered, or at least understood, if the facts were known. The Exit Poll Consortium should enable that knowledge. It would be a relatively simple regression to map exit poll data against counties or precincts with suspect machines. More importantly, it would be relatively easy to isolate where, if anywhere, suspicion should be directed.
The Exit Polls have done enough damage to this election. My bet is that it was incompetence at Edison/Mitofsky. But those firms owe it to this Nation to release their data totally, so that a wide range of competent statisticians can evaluate whether and where the problem was.
And more importantly for the blog space: If blogs are going to be something more than the CB radios of journalism, we need an ethic to treat this sort of question ethically. Anyone who is surprised that a voting machine didn’t work has been living on Mars for the last 100 years: Always, and in every election, voting machines fail. That fact should force us to a sensible architecture for voting machines — one which we don’t have just now for electronic voting machines. But it isn’t, itself, evidence this election was “stolen.”
No one can, or should, utter such words without the data to back it up. Instead, we should demand what, in this context, should be our right: to have access to the data. There is irresponsibility somewhere. Let us not add to it here.