Slashdot is powered by your submissions, so send in your scoop


Forgot your password?
User Journal

Submission + - Who won? How do you know?

doom writes: "I think they call them "exit polls" because people bolt for the exits when you mention them, but I'm still fascinated by the subject myself, and this book is one of the reasons why. In Was the 2004 Presidential Election Stolen? , the central focus is, of course, on the infamous exit-poll discrepancies of the 2004 US Presidential election; but the authors also put it into context: they discuss the 2000 election, the irregularities in Ohio in 2004, the electronic voting machines issues, and the media's strange reluctance to report on any of these problems. Further, in the chapter "How did America really vote?", they compare the indications of the raw exit-poll data to other available polling data. Throughout, Freeman and Bleifuss do an excellent job of presenting arguments based on statistical analysis in a clear, concise way.

The heart of the book in my opinion, is Chapter 5, "The inauguration eve exit-poll report": The Edison and Mitofsky firms that conducted the NEP exit polls later released a report trying to explain how they could have gotten it so far wrong. Freeman and Bleifuss, of course, take issue with the presumption that the discrepancies must be "errors", and argue in a different direction. This section makes an exciting read (in a nerdy sort of way) it's an impressive piece of statistical judo: Freeman and Bleifuss take on Edison/Mitofsky with their own data, and totally shred their conclusions. The authors show:
  • That the exit-poll discrepancies had a statistically significant correlation with the use of electronic voting machines, with races in battleground states, and in almost all cases favored the Republicans.
  • The "Reluctant Bush Respondant" theory looks extremely unlikely: response rates actually look slightly better in Bush strongholds than in Kerry strongholds; and while media skepticism remains strong among conservatives, it has been on the rise among Democrats, and yet the data shows no shift in relative avoidance of pollsters.
They also deal with the various other excuses that were floated shortly after the election:
  • The discrepancies can't be shrugged off with an "exit polls are not reliable" — theory shows that they should be better than any other survey data, and history shows that they always have been pretty reliable.
  • There was no upswing of support for Bush throughout election day — that impression was entirely an artifact of the media "correcting" the exit-poll figures to match the official results.
One of the book's authors, Steven Freeman, was one of the first to examine the exit-poll discrepancies, and as a professor at University of Pennsylvania with a background in survey design, he was well equipped to begin delving into the peculiarities he had noticed. Overall, this is an excellent book for people interested in evaluating the data; with lots of graphs that make it easy to do informal estimates of the strength of their conclusions (just eye-balling the scatter, the correlations they point to look real, albeit a little loose, as you might expect). There's also an appendix with a very clear exposition of the the concept of statistical significance, and how it applies to this polling data. There are of course, limits to what one can conclude just from the exit-poll discrepancies: "We reiterate that this does not prove the official vote count was fraudulent. What it does say is that the discrepancy between the official count and the exit polls can't be just a statistical fluke, but commands some kind of systematic explanation: Either the exit poll was deeply flawed or else the vote count was corrupted. "

This is a remarkably restrained book: unlike many authors addressing this controversial subject, Freeman and Bleifuss have resisted the temptation to rant or speculate or even to editorialize very much. Freeman claims that he is not a political person (and adds "I despise the Democrats"); possibly this has helped him to maintain his neutrality and focus on the facts of the case.

Personally, I found this book to be something of a revelation: in the confusion immediately after the 2004 election, I had the impression that the people who wanted to believe that it was legitimate at least had some wiggle room. There was some disagreement about the meaning of the exit polls: there was that study at Berkeley that found significant problems, but then the MIT study chimed in saying there wasn't, so who do you believe? The thing is, the MIT guys later admitted that they got it wrong: they used the "corrected" data, not the originally reported exit poll results. The media never covered that development, and I missed it myself...

On the subject of electronic voting machines, They include a chapter discussing electronic voting in general which covers ground that is by now familiar with most readers here: the strange case of Wally O'Dell and Diebold; and also the lesser known problems with ES&S. Have you heard this one? "In 1992, Hagel, then an investment banker and president of the holding company McCarthy & Co., became chairman of American Information Systems, which was to become ES&S in 1999. [...] In the 1996 elections, Hagel launched his political career with two stunning upsets. He won a primary victory in Nebraska [...] despite the fact that he was not well known. Then, in the general election, Hagel was elected to the Senate in what Business Week described as 'an unexpected 1996 landslide victory over Ben Nelson, Nebraska's popular Democratic governor.'"

Also, my experience is that a lot of people need to hear this point: "The voting machine company Datamark, which became American Information Systems and is now known as ES&S, was founded in 1980 by two brothers, Bob and Todd Urosevich. Today, Todd is a vice president at ES&S and Bob is CEO of Diebold Election Systems."

It's impossible to see how you can come away from this situation without seeing that we badly need reform of the electoral system: even if you don't believe the 2004 election was "stolen", how do you know the next one isn't going to be? A paper trail that can actually be recounted would be a nice start, eh? But only a start. As the author's point out: "We devoted a chapter to the ills of electronic voting, but a critical lesson of the 2004 election is that not only DREs, but all kinds of voting machine systems are suspect. Edison/Mitofsky data showed that while hand counted ballots accurately reflected exit-poll survey results, counts from all the major categories of voting machines did not."

In one short passage, the authors list a few "grounds for hope", but following up on these points is not encouraging: The Diebold-injunction law suit in California brought by VoterAction has since been denied and one attempt at a paper trail amendment, HR 550 has stalled out; but then if you look around you can still find other grounds for hope: HR 6200, the Paper Ballot Act of 2006.

Oh, but if you're looking for an answer to the question posed by the book's title, the authors conclude: "So how did America really vote? Every independent measure points to a Kerry victory of about 5 percentage points in the popular vote nationwide, a swing of 8 to 10 million votes from the official count."

So, of the many and various potentially depressing books out there about the state of the United States, I recommend this one highly: it addresses a critical set of issues that everything else depends on (when you hear from yet another Monday-morning-quarterback about what the Democrats need to do to win, remind yourself that maybe they haven't actually been losing).

A prediction for the upcoming election, in the light of this book: The Republicans may let the House go, but they will drop a heavy finger on the scale to retain the control of the Senate. Keep an eye on the three close states: Missouri, Tennessee, and Virginia (see Andrew Tanenbaum's site). Note that Tennessee and Virginia are both high risk states without paper-trails, so you can expect "surprise" upsets in favor of the Republicans there. Missouri is in better shape as far as election integrity goes, but even if there's an actual upset there in favor of the Democrats, control will still (just barely) remain with the Republicans."

Life in the state of nature is solitary, poor, nasty, brutish, and short. - Thomas Hobbes, Leviathan