Electoral Vote Fraud
Matthew Yglesias and
James Joyner both have pretty good posts today on Pres. Bush's hopes of winning the all-important Electoral College vote.
Matthew says:
If you look at the latest state by state numbers, Kerry's actually doing well enough that he could drop Florida and Pennsylvania (he's already losing in Ohio) and still win the election based on the other states where he's leading.
James, is a little more optimistic than that, but:
Still, I agree with Matt: the math favors Kerry
But, where exactly are they getting their state-by-state poll numbers and why should that matter? They both site
electoral-vote.com as their source of information. But is that site a reliable source of information?
A scan of the site reveals that it is a highly partisan front in favor of the Kerry campaign. For instance, on one page President Bush's record is examined with links to a site accusing the Bush family of ties to the Soviet Union and Nazi Germany. It also links to an article on how Bush stole the election. Electoral-vote.com is more than a partisan front--it is run by nutjobs!
A better analysis of the state-by-state electoral numbers can be found at election-projection.com. This site is run by Scott Elliot who is up front about his partisan leanings. But what makes Scott's site so good is that his electoral college vote count predictions are made using a publicly available formula. To add credence to his site, his prediction now is that Kerry wins.
These differences are very important. For instance, Scott's numbers show that Bush is ahead in several states that election-projecton show in Kerry's camp. These include TN, AZ, OH, WV, & NV. Update: Matthew Yglesias just e-mailed and, as a result I had to go back and change this--see why we use peer review in academia?). In fact, it seems that Scott has both OH and NV in Kerry's camp--the exact opposite of electoral-votee.com. It turns out, the differences aren't as important as I earlier suggested--both polls yield almost identical results! (hat tip to Matthew Yglesias for pointing that out)
The moral of this story is not to trust just any website. Yes, Kerry is slightly ahead in some state-by-state polling numbers but the race is still early. My own prediction is that Bush wins with either the exact same electoral margin as before, or slightly ahead, this time carrying MN and NM. Just a guess, but it is at least as well informed as those of election-projection.com---and probably a whole lot closer to the target.
UPDATED MORAL OF THE STORY: Blogs can be and often are wrong. This one included. Especially when writing in the heat of the moment. Ok, so both sites have similar outcomes, but I am still sticking with my original charge the electoral-vote.com is a DU fringe front and should not be the source of any credible information.
Why would I stick to this, even after my glaring mistakes were pointed out? Because, I just don't trust someone who is so far out in left field that he would actually site material linking Bush to the Nazis!!! That is beyond partisanship, that is wacko land. Our biases come out in many ways, especially when we select pieces of information used in analysis. I can't but help to think that someone who thinks Bush stole the election (literally) won't have some major data selection biases.
Posted by: Rusty at
10:19 PM
| Comments (3)
| Add Comment
Post contains 568 words, total size 4 kb.
1
Rusty, the electoral-vote.com people may be nutjobs, I don't know, but it's not effecting their polling numbers, because they're not doing their own polls. They draw from a stable of mainstream polls, and use whichever one is most recent for each state. Now, I don't think this is the best methodology, because some polls are flukish, even if done properly (I don't really buy TN as a toss-up or that AZ is trending Kerry, and I don't think Kerry is in much trouble in Maine, for example. Those polls are probably outliers.) If I were doing it, I'd average a few of the most recent polls from several reputable polling outfits, perhaps weighting slightly toward the most recent one. So I don't think they made the soundest methodological choice, but I don't see how it has anything to do with their ideology/nuttiness. Their choices are upfront, I haven't gone back and checked the numbers they attribute to Zogby, ARG, CBS and others, but I figure if they were misquoting them, someone would have pointed it out.
So how, precisely, does their method demonstrate bias?
Posted by: djw at August 04, 2004 03:49 AM (8RUbb)
2
Very thoughtful rejoinder. The problem with the methodology seems to be that the site administrator simply chooses which polls to include and which polls to ignore. It may be that there is some reason for including some polls and not others, but as someone who kind of has a little crush on methodology, this bothers me. The first rule of scientific methodology is to set your standards first, and then let the chips fall. That site does a good job of defining terms--such as what it means to be 'weak' or 'strong' in favor of a candidate--but does a poor job in letting us know which polls will be used.
Now, I'm not saying that he is misreporting the numbers from the polls, only that there may be a possible selection bias error. The problem with selection bias is that the person doing the selecting usually doesn't realize his own bias! That is, we all tend to think of ourselves as 'in the mainstream', even when we're really way out in left or right-field.
My problem with the site is that whoever runs it is so far out in left-field that he may be ignoring certain polls (not necessarily on purpose) at the expense of others.
Posted by: RS at August 04, 2004 02:21 PM (JQjhA)
3
You may have a point, I don't know. At first glance I thought he primarily drew from the same list of "proper" polling outfits most people seem to pay attention to, but looking again, I see a lot more polls I haven't heard of--many connected to univerities, which generally do it right, but not mainstream. Still, the bulk of swing state polls seem to come from generally recognized respectable outfits (SUSA, ARG, Rasmussen, Zogby, etc although Zogby tends Democratic a little I think). In the end, it doesn't matter much if the Rhode Island, Idaho and New York polls come from dubious sources.
But I also kinda want to challenge you on an implication of your post--that the further you get from the political mainstream, the less fair and objective you are. This is a common assumption, and I'm simply not convinced it's warranted. I my time in academia, I saw some people with some pretty "out there" views (left and right) who were exceedingly concerned with careful and objective methodology in their research, and some centrists who were sloppy and myopic. I think centrism can be mistaken for objectivity, which is a big mistake. I also think those with non-mainstream views might be more inclined to be extra-careful because those who know there views will be expecting bias, so they must be more careful to be fair and avoid potential bias in order to be taken seriously. Centrists don't have this concern to the same degree. And left, right, and center, the academics I've known are united through their obsession with being taken seriously by their peers.
Posted by: djw at August 04, 2004 06:44 PM (8RUbb)
Hide Comments
| Add Comment
23kb generated in CPU 0.2296, elapsed 0.5141 seconds.
118 queries taking 0.4863 seconds, 247 records returned.
Powered by Minx 1.1.6c-pink.
