Strange vote: The reason my numbers differed from Slate's

A man fighting both the bug and a deadline should avoid blogging, but I can't stop thinking about the oddities of this election. Among the most gnawing of the problems: The allegation that exit-poll discrepancies were greater in states that allowed e-vote verification.

Here's what Slate had to say about what I had to say:

Cannon's analysis doesn't jibe with Slate's exit-poll numbers. A comparison to the latest vote tallies shows Slate's final exit numbers in the paper-trail states of New Hampshire (undervalued Bush by 5 percent) and Nevada (undervalued Bush by 3 percent) were less accurate than those in Ohio (2 percent off), Florida (3 percent), and New Mexico (2 percent). The other state Cannon lists, Illinois, won't require a paper trail until 2006.

The numbers I used were provided by another blog-person, who won't be named here. If those numbers were off-base, the egg facial should stay squarely on my own mug.

But were they off?

The key words: "latest vote tallies." Concerns about the exit poll discrepancies sent me scurrying onto the net around 6 a.m. on November 3. The numbers I found were probably posted a couple of hours before then, and presumably originated from televised reports. Those numbers -- which showed congruence between exit poll data and the count in Nevada and New Hampshire -- were not final.

After the final vote came in, the congruence disappeared. Once again -- as was the case in New Mexico, Ohio and Florida -- the exit polls undervalued the Bush vote.

Which means that a last-minute flurry of votes went to Bush. Not impossible. (Things to do: Check which county's votes were counted last in Nevada.)

Bottom line: Both Slate's numbers and mine were probably accurate; the difference is explained by timing.

The fact remains that the discrepancies are suspicious, because they always go in one direction. Nevada -- let us concentrate on that state for the nonce -- only helps prove the point.

Since international election overseers view exit poll disparities as evidence of vote tampering, we have good reason to suspect foul play. And if we allow ourselves such suspicions, what can we further deduce from a last-minute Bush surge in Nevada?

"Hold on!" I hear Mr. Scoffee saying. "Are you implying that there was 11th-hour cheating in Nevada as well? How could they hope to get away with it? Yes, Nevada has electronic voting, but they also mandate a paper trail."

True. But as I understand the situation (and I'll be happy for any correction), nobody looks at that paper trail unless someone demands a recount. And nobody is going to demand a recount if the election has already been decided elsewhere. If Nevada's five electoral college votes aren't crucial, then no-one will see if those slips of paper contain Kerry votes, Bush votes, or chicken soup recipes.

So let us add this item to our list of suspicious circumstances: In Nevada, Bush received a sudden surge of votes -- a "bulge" if you will -- only after Kerry's chance of victory in Ohio went from slim to grim. The moment Nevada lost all chance to be a "player" in this election, the pro-Bush electoral floodgates opened.

But why pad the vote in Nevada? Earlier today, I noted one good reason to hack the system in non-crucial states: To give W the widest plausible popular vote victory and thereby justify talk of a mandate.

I plan to offer further musings and research as soon as work and health permit. In the meantime, here are some links for further investigation.

The latest offering from Raw Story offers numbers, plus an unsuccessful attempt to get feedback from the firm that did the polling. Some comments from the readers are infuriating: "Exit polling is unscientific," and the respondents are only those who choose to cooperate. All such commentary ignores some basic points: Why did the errors on November 2, 2004, all go in one direction? Why do we not see such patterns in other countries which use exit polling? Why was exit polling in this country so much more reliable in the days before e-voting?

Donkey Rising offered this explanation:

What happened? Some combination of bad precinct samples, response bias, or failing to accurately account for early and absentee votes must have been at work. Whatever it was, it was a major problem. In 2000, the national exit poll also overestimated Gore's vote, but not by nearly as big a margin.

Lame, lame, lame.

Bad precinct samples? Response bias? Poor weighting of absentee ballots? C'mon, think about it: Exit polls have been a staple of the American political scene for decades. In the past, they did not present problems of this sort. Are we arguing that the pollsters get worse at their jobs as the years pass? Don't you think it an odd coincidence that their skill level should diminish just as e-voting grows in popularity?

Think, too, on this: This is the second presidential election (as Donkey Rising helpfully reminds us) in which the exit polls indicated that the Democrat would receive more votes than the "actual" count gave him. The 2002 elections also saw disparities between poll predictions and final numbers. The errors always go in one direction -- state after state, year after year.

A conservative poster to the above-mentioned site opined that the exit pollsters were evil librul plotters who deliberately skewed their results to discourage Bush voters -- even though everyone in the "liberal" media refused to report the poll numbers. This same fellow says that guys like me are just "conspiracy theorists."

More pleasing is the latest by Greg Palast, who reminds us of the problem of "spoilage" -- egads, more chads! -- in non-evote districts. Spoiled ballots (which can usually be un-spoiled by a glance at the actual card) are a problem primarily in minority areas. His comments on the scene in New Mexico deserve quotation here:

Already, the election-bending effects of spoilage are popping up in the election stats, exactly where we'd expect them: in heavily Hispanic areas controlled by Republican elections officials. Chaves County, in the "Little Texas" area of New Mexico, has a 44 percent Hispanic population, plus African Americans and Native Americans, yet George Bush "won" there 68 percent to 31 percent.

I spoke with Chaves' Republican county clerk before the election, and he told me that this huge spoilage rate among Hispanics simply indicated that such people simply can't make up their minds on the choice of candidate for president. Oddly, these brown people drive across the desert to register their indecision in a voting booth.

On Slate, Martin Plissner, who calls himself a friend to one of the two men in charge of the National election Pool, offers a disingenuous "defense" of exit polling which comes down to this argument:

But the real problem is not that the exit polls were wrong. They were about as accurate as they usually are. The problem was that in the age of the Internet the exit polls were being seen by thousands of people who didn't know how to read them. Like any sophisticated weapon, they are dangerous in the hands of the untrained.

He's talking about you and me, folks.

And -- quite predictably -- he refuses to address the point I've made repeatedly. I don't demand that exit polls be completely accurate, but I do want to know why the errors always went in one direction. And I want to know why they err in the same way election after election. Read Plissner's piece carefully. Read it three times. Tell me if you get even a hint of an explanation.

There's much more to be said, but I've already outlined the way the argument will go and go and go. We will continue to ask: "Why do exit polls always undervalue Republicans?" And guys like Plissner will always respond: "Exit polls are not always accurate." Which avoids the damned question.

When a coin keeps landing head's up, you gotta wonder if there's something funny about that coin.