So how was the crooked polling company so accurate?

[TexasLawyer]

Inquiring minds want to know!!!

I am sure you all saw this article about the fraudulent pollsters. I did not think that much about it, since it sounded like a plain-vanilla rip-off. But then I saw that this company claims to have extremely high accuracy at predicting election results.

It could be that the company is lying about their track record, but if the claim is true, it seems bizarre that pollsters who routinely fake their results could nevertheless be so accurate in predicting actual election outcomes.

Any theories?


Pollster Pleads Guilty to Fraud
The Associated Press
Thursday, September 7, 2006; 12:22 PM


BRIDGEPORT, Conn. -- The owner of DataUSA Inc., a company that conducted political polls for the campaigns of President Bush, Sen. Joe Lieberman and other candidates, pleaded guilty to fraud for making up survey and poll results.

Tracy Costin pleaded guilty Wednesday to one count of conspiracy to commit wire fraud. Costin, 46, faces a maximum of five years in prison and a fine of up to $250,000 when she is sentenced Nov. 30.

As part of her plea agreement, Costin agreed to repay $82,732 to the unidentified clients for 11 jobs between June 2002 and May 2004. DataUSA is now known as Viewpoint USA.

<snip>

http://www.washingtonpost.com/wp-dyn/content/article/2006/09/07/AR2006090700437.html

From a legal web site, posted last year (3/11/05) shortly after the pollsters were indicted.


http://lawprofessors.typepad.com/whitecollarcrime_blog/2005/03/if_you_cant_tru.html

The company's website (www.datausainc.com) has the following statement: "DataUSA upheld the highest standards in data collection as demonstrated in the excellence of our data. DataUSA's predictions of voting behavior and consumer behavior routinely reflected to be within 1% to 2% of actual behavior. We are proud of the quality data collected and want to thank you for teaming with DataUSA and for your support over the years."

The page "http://datausainc.com" is no longer there, so I guess I'll have to take this guy's word for what it used to say

[BleedingHeartPatriot]

1. Bizarre question. I didn't click on the links,because of the odd OP.

Your OP is predicated on the assumption that the "source" is credible and you then explain away the credible source. :shrug: MKJ

[DoctorMyEyes]

2. I see what you're asking

I see what you're asking - that perhaps they were accurate in predicting election results because they had inside info about rigged elections.

Except we don't know if they really were accurate or if they just claimed to be on their website in order to attract more clients. You know? Lied.

It would be an interesting theory if someone could come up with evidence of their actually predicting accurate election results BEFORE an election.

[Land Shark]

3. We are missing facts here and Febble has more ideas i'm sure

but one Febble pointed out in Election Forum was that if the data is made up in the same proportion as legitimate data (basically to rush to a deadline) it wouldn't distort the overall percentages at all, and if the slightly smaller sample was still on the mark, they could still have accurate polls.

Just missing quite a few facts here, though, but the above is one possibility.

[Igel]

4. Things on the Internet frequently have ghosts.

Here's where the ghost www.datausainc.com lives:

http://web.archive.org/web/*/http://www.datausainc.com

The most recent archived copy of their page has the statement cited:
http://web.archive.org/web/20050303173601/http://www.datausainc.com/

I'd add that it's a damned strange commercial webpage, because that's about *all* it says. In fact, in early '05 the commercial webpage was replaced by a "so long, and thanks for all the fish" web page":

To all DataUSA clients,
Thank you for your business and support over the past years.
Best,
Tracy

Sounds like they were shutting down; and the page cited by your source was overly defensive, from a moribund or defunct company.

It's also possible that they weren't lying on the 'cover-your-ass' webpage, but were right by accident. Or 'routinely' doesn't deny the possibly of some wrong, fudged, numbers.

Presumably there was evidence they faked numbers.

[BleedingHeartPatriot]

5. Nuance escaped me. The OP was clear, but not. Glad others could

see.

OK, now I get it. Do I get points for the haiku? :-) MKJ