Monday, May 16, 2005

Fisking Jim Lampley on the 'Stolen' 2004 election

Jim Lampley posits that the oddsmakers in Vegas give us a big clue as to how the 2004 presidential election was stolen by Rove et al.:

"Oddsmakers consulted exit polling and knew what it meant and acknowledged in their oddsmaking at that moment that John Kerry was winning the election.

And he most certainly was, at least if the votes had been fairly and legally counted. What happened instead was the biggest crime in the history of the nation, and the collective media silence which has followed is the greatest fourth-estate failure ever on our soil."

When you get outlier results you look to a problem with your exit polling ...not to some idiotic (and I mean idiotic) conspiracy theory to attempt to discount the real vote tallies. Mitofsky, who *did* the exit polling, argues that it was a self-selection problem in the sample (see below). This is a reasonable and rational explanation. A vast conspiracy to alter the election tallies is not. Exit polling is a survey of a sample of voters. The vote tallies are the *population* of voting results. Exit polling is an attempt to get an estimate on that population. You don't favor an estimate from a sample over a census of the population! You don't learn that in upper-level survey sampling classes (which I have had, by the way), you learn it in STAT 101. This is akin to having a sensor go off on your engine suggesting you are low on oil. It is quite reasonable to conclude that you are, in fact low on oil. Indeed, you take your car by a reputable mechanic, and he draws the same conclusion. However, you personally check the engine oil the next morning and see that it is, in fact, full. But instead of concluding that you have a problem with the sensor, you decide that someone must have snuck into your garage last night and filled the oil in your car. Why? Because the mechanic thought you were low on oil, too. This stolen election nonsense is *just* that ridiculous.

Lampley's point is essentially that oddsmakers know what they are doing, hence their placing the odds for Kerry at 2:1 is supposed to tell us something.

"People who have lived in the sports world as I have, bettors in particular, have a feel for what I am about to say about this: these people are extremely scientific in their assessments. These people understand which information to trust and which indicators to consult in determining where to place a dividing line to influence bets, and they are not in the business of being completely wrong."

Of course, Lampley ignores the fact that Kerry wasn't at 2:1 the night before, as can be seen in this Washington Post article.

"Maybe all the academics and pollsters should spend less time fine-tuning their election algorithms and just pay attention to the Vegas odds. "Online traders wagering on the outcome of next week's presidential election are giving President Bush slightly better odds of victory than Sen. John Kerry. But trading-exchange operators caution against reading their data too strongly in Bush's favor, saying the race remains remarkably close," Wired News reported on Tuesday. "On two real-money wagering sites, TradeSports and the Iowa Electronic Markets, traders give Bush a 59 percent and 61 percent chance, respectively, of being re-elected, as of Monday. A third site, NewsFutures, where traders wager for prizes and bragging rights, has the race virtually deadlocked, giving Bush a 51 percent chance of victory." "

The oddsmaking results that Lampley cite are *entirely* the product of the exit polling results that were circulating the morning of the election. Just as our mechanic's conclusion was entirely based on the sensor. Hence, talking about those results is no different than talking about the exit polls. That oddsmakers would respond to this exit polling information doesn't provide *any* more information than the exit polling itself provides. As the computer geeks say, 'garbage in, garbage out.' Lampley in no way addresses the argument that the exit polls are flawed. He assumes they can't be flawed because the oddsmakers relied upon them. This is obviously silly.

Again, all we have here is the exit polls...the same thing that we started with. Lampley doesn't add any new information by noting that oddsmakers relied upon this information. And the exit pollers themselves offer a more reasonable conculsion:

"Our investigation of the differences between the exit poll estimates and the actual vote count point to one primary reason: in a number of precincts a higher than average Within Precinct Error most likely due to Kerry voters participating in the exit polls at a higher rate than Bush voters. There have been partisan overstatements in previous elections, more often overstating the Democrat, but occasionally overstating the Republican. While the size of the average exit poll error has varied, it was higher in 2004 than in previous years for which we have data. This report measures the errors in the exit poll estimates and attempts to identify the factors that contributed to these errors."

Now, I know several political scientists who think Mitosfky screwed up (and will say harsher things than that in private). But even Mitosfky himself doesn't cling to the fantastical notion that his exit polls were on the money and what really happened was some vast, multi-state conspiracy to increase the President's vote tallies. Indeed, he specifically *denies* that the exit polls support the stolen election thesis:

"Exit polls do not support the allegations of fraud due to rigging of voting equipment. Our analysis of the difference between the vote count and of the exit poll at each polling location in our sample has found no systematic differences for precincts using touch screen and optical scan voting equipment."

You can read their report here

Oddsmakers are no more 'scientific' than stock brokers...and we well know that they have responded to erroneous information in the past. Indeed, the sensitivity of the markets to jobs reports (reports that are often revised substantially after the fact) illustrates this very point. And the exit polling data *was* erroneous. As admitted by the exit pollers themselves. The simple fact is this, *all* of the indicators prior to the election indicated that this would be a close election and most average estimates had Bush ahead by 1-3 points. Indeed, the rolling average I used to make my own prediction had Bush ahead 2.5 points on election eve. The exit polling numbers were well out of line with all of our previous indicators. As an outlier, we should be even more inclined to look towards problems with their methodology or data collection. We certainly shouldn't jump to idiotic conspiracy theories. Conspiracy theories that make the JFK conspiracy look simple, practical, and on a small scale! Sell crazy somewhere else, Jim Lampley. We're all full up, here. D.GOOCH

3 Comments:

Blogger Pat said...

Great post, loved the oil analogy. Lampley's comment that the oddsmakers are very scientific is hilarious; I suspect that what places like Tradesports do is use a computer program to adjust values based on bid and ask prices. I followed the Tradesports and Iowa Electronics Markets virtually daily during the election season, and hourly on election day, and the only time Kerry took the lead after Labor Day were those few hours when the exit polls were the buzz. And a lot of people noticed some suspicious trading in the accounts as well--the volume in Kerry contracts was low despite the apparent surge.

1:37 PM  
Blogger Judith said...

what Pat said. The electronic markets showed Bush winning by about what he did.

12:53 AM  
Blogger Donald said...

Exactly, Pat. Whether or not there was any Soros-inspired shannigans going on, the fact is Lampley's argument simply doesn't hold water. Thanks for your notes, Pat & Judith. D.GOOCH

10:21 AM  

Post a Comment

<< Home