Friday, March 25, 2005

041115 Posting--Recount, Anyone?

I apologize for sending so much ‘after the fact’. However I asked Matt Kerbel (Villanova PolySci professor) about the issue of recounts, and I strongly suspect that you will find his comments very interesting. I am also attaching a PDF of the study he alludes to that was done by the UPenn professor on the discrepancies between the exit polls and ‘official’ tallies.

As with most of you, I have no interest in getting my hopes up again just to be dashed. However, I think that this is too important to disregard.

Paul Baumbach


I try to avoid the temptation to adopt Oliver Stone's paranoia, however .

How serious is the recount hubbub in the blogosphere? I hear that Nader is challenging NH, and the Green/Libertarians are about to challenge the Ohio results? Given that Ohio had (reportedly) 70% of votes cast with paper ballots, what did Florida 2000 and the US Supreme Court decision mean for paper ballot recounting in Ohio in 2004? Is this all 'not a chance in ****' dreaming? I feel like I need to pinch myself.

Have you reviewed the pro & con academic views of the exit poll discrepancy and have a (as much as possible) non-partisan, academic/scientific take?

If you have time, I'd love to hear your thoughts on this.


Hi, Paul. She has not sung -- yet. And it's because -- of all people -- Ralph Nader is going to get a recount in New Hampshire.

Without getting all Oliver Stone about this, there is a lot of circumstantial evidence that points to as yet unexplained tabulation irregularities -- a neutral, empirical concept. I am not talking about fraud, although it may be fraud. I am taking about data that do not fit with other data.

As you know, Paul, around 6 p.m. on Election Day we were looking at a comfortable Kerry victory in line with what the Incumbent Rule would predict. Bush's margin in the national exit poll was 48% -- exactly where you would expect it to be based on his final average in the last set of polls (where he was at 48.5%). Then without warning the big states (Pennsylvania, Ohio, Flordia) began to skew heavily toward Bush as the actual results came in. Is this possible? Yes -- the exit polls could have undersampled Bush's support. But it is highly unlikely. A piece published last week by a Penn statistician put the conditional probability that these three polls would be off by as much as they were at 250 million to one. You don't have to be a statistician to wonder whether it was the actual vote count that was off, not the exits.

In North Carolina, where roughly one-third of the electorate voted in advance, there is evidence (and, I have to caution, without having seen the actual raw data I can't comment on its accuracy) that for every race except for the presidential and senate contests the early vote margins were within one percentage point of the election day margins. Yet, Bush picked up seven points in the election day vote over the early vote. If this can be verified, it suggests that something was different about the election day vote for president in that state that did not apply to other contests -- something that led to a spike in Bush's tally.

There are data in Florida and Ohio suggesting that several heavily Democratic counties surprisingly gave Bush an unexpectedly high proportion of their vote. The Florida counties have been explained as being "Dixiecrat" counties, and some of them were. But, the interesting thing about these counties is they share a voting method -- optical scanning. And it is in the optical scanning tabulation that I think we may have the common link among these and other irregularities. But to comment on this I have to move from reporting data to analyzing it, which means I am about to speculate.

It is possible that the computers that compiled and tabulated votes from these machines -- the central computers -- overcounted Bush votes, perhaps by a large margin and perhaps on a national scale. Perhaps. My understanding is that the tabulations are compiled on

PC's running Microsoft Access spreadsheets. I further understand that the spreadsheets were not password protected, that the computers were connected to the web, and that the data were not encripted. In other words, they were there for the taking should someone want to hack them. Be aware that I am drawing a conclusion that fits the data. I could be wrong. But it is within reason that I am right.

This is where Ralph Nader comes in. He has requested and will get a recount of the optical scan votes in New Hampshire. Nader has asserted that he believes these machines boosted Bush's tally by upwards of 45,000 additional votes -- a significant number in a small state. Because there is a paper record of the vote -- those optical scan strips will be hand counted -- it will be possible to determine if he is right. If there is any hope to challenge the vote, it will most likely come from here. Should Nader be proved correct it will call into question all the optical scan ballots, and you will see requests for hand recounts nationwide. It won't matter WHY the count was off at this point. It will simply matter that it was.

Should this not come to pass, either because Nader's recount is ultimately blocked or because it fails to demonstrate a measurable problem, there is still the possibility that the election can turn on an Ohio recount being called by the Green and Libertarian candidates (which will go ahead if they raise enough money to pay for it), or by Kerry closing the gap in Ohio on the strength of provisional and absentee ballots. There are other challenges in the works as well -- see: I suspect all these challenges are a much longer shot, though.

Also, keep in mind that while many of us cling to the remote hope that Kerry could still become president, equally important issues of transparent elections and the winner's mandate are also in play. One reason why so much blog energy is being expended on this issue is that it is easy for Kerry partisans to believe that Bush stole the election. All the dirty tricks and partisan dealings leading up to election day in states like Ohio and Florida, against the backdrop of what happened in 2000, against the backdrop of a Republican Congress' unwillingness to mandate a paper trail for electronic machines -- all this leads people to believe that it could have been stolen. Whether or not it was, the issue of holding transparent elections in the future is a critical one, and that cause can only be advanced by delving deeply into what happened two weeks ago, regardless of who wins. And in the likely event that all this is insufficient to overturn Bush's victory, it will and should cast a shadow over his second term -- if it becomes a mainstream story.

The New Hampshire recount, if and when it goes forward, should push this story into the mainstream. For whatever it's worth, a reporter friend told me Friday that he has two friends who are political writers for Newsweek, and they think that something funny is going on. An event like a Nader recount of New Hampshire would give them the hook they need to start writing about it.

So, I'd say don't expect Kerry to become president, but know that it is still a longshot possibility. Feel free to share this with the others who came to the talk at the church. I'm sure they're concerned as well.



No comments: