« AAPOR Conference: Correspondents Wanted | Main | AAPOR: Day One »

May 11, 2005

On "Smackdowns" and Fraud

The big news at the newly minted Huffington Post is that two of its "celebrity" bloggers have taken up the exit-polls-as-evidence-of-vote-fraud debate.  What's remarkable about yesterday's exchange between ABC sportscaster Jim Lampley and National Review Correspondent Byron York (here, here and here, to say nothing of the pat-self-on-back praise by standup comic Robert George) is how devoid it is of fact.   The Huffington Post has a long blog roll (that kindly includes MP), yet Lampley and York demonstrate little awareness of what others have said about this issue since November. Their "smackdown" includes a lot of impressive name calling but nary a link to the ongoing debate.   

Consider this paragraph from Lampley's initial post:  He sees evidence of fraud in the fact that the "extremely scientific" Las Vegas oddsmakers, having presumably consulted the leaked exit polls, set Kerry as the favorite.  He continues: 

It is damned near impossible to go to graduate school in any but the most artistic disciplines without having to learn about the basics of social research and its uncanny accuracy and validity. We know that professionally conceived samples simply do not yield results which vary six, eight, ten points from eventual data returns, thaty's why there are identifiable margins for error. We know that margins for error are valid, and that results have fallen within the error range for every Presidential election for the past fifty years prior to last fall. NEVER have exit polls varied by beyond-error margins in a single state, not since 1948 when this kind of polling began. 

Wow.  I'm guessing there are a few grad school instructors who may want to assign that paragraph to their students.  How many problems can you identify?

Jim, some advice:  You might want to put that degree to good use and browse a bit here (read these two closely) or perhaps just jump to the end (here and here) and consider where the debate is now.   Note the use of links throughout to source material like this report from the exit pollsters themselves or the one criticizing it here.  Bloggers do a lot of this.

What is ironic about the Lampley-York "smackdown" is that it comes on the same day an official investigation actually uncovered true evidence of fraud, though perhaps not the variety that Lampley imagines.  Yesterday, an appointed task force in Milwaukee that included a Republican appointed U.S. attorney, a Democratic county district attorney, the FBI and the Milwaukee police found what everyone seems to agree is "clear evidence" of fraud.  The AP published a widely distributed story, but Greg Borowski of the Milwaukee Journal Sentinel, the newspaper whose own reporting led to the official investigation, provides the must-read coverage. Read it all. 

According to the JS story, the true evidence of fraud comes from "200 cases of felons voting illegally and more than 100 people who voted twice, used fake names or false addresses or voted in the name of a dead person."  All of these were found in the city of Milwaukee, where John Kerry received 71.8% of the vote (up from 67.6% in 2000). 

It is, of course, theoretically possible that such fraud may have benefited either side, but the Democratic stronghold of Milwaukee is not exactly the place one expects to find padding of the Republican vote.  Presumably that is why Republican officials used the news to push a photo ID requirement for voting.

It is worth noting that the officials saw no evidence of a widespread conspiracy.  Quoting the AP story:

Investigators did not uncover any proof of a plot to alter the outcome of the hotly contested presidential race in Wisconsin's largest city and have filed no criminal charges.

"There is not the evidence of an overriding conspiracy in all of this," U.S. Attorney Steven Biskupic said.

Instead, the task force reported "widespread record keeping failures and separate areas of voter fraud." Biskupic said the faulty records will make it tough to prosecute many of the crimes, although he will file charges if he thinks he can prove wrongdoing in any cases.

Thus, the Milwaukee investigation appears to offer even more evidence of the sort of sloppiness associated with the random "reporting error" MP described recently.  Consider these examples from Borowski's article: 

The city's record-keeping problems meant investigators from the FBI and Milwaukee Police Department have logged more than 1,000 hours reviewing the 70,000 same-day registration cards, including 1,300 that could not be processed because of missing names, addresses and other information.

Indeed, about 100 cards described as "of interest to investigators" cannot be located, officials said. And within the past few weeks, police found a previously lost box of the cards at the Election Commission offices...

The newspaper also identified numerous cases in Milwaukee where the same person appears to have voted twice, though that analysis was hampered by major computer problems at the city.

Those problems, which city officials labeled a "glitch," meant hundreds upon hundreds of cases where people are incorrectly listed as voting twice. These are in addition to cases of double voting identified by investigators.

Apparently, mistakes happen.  A lot. 

Until Republicans and Democrats find a way to agree to clean  up the process, cynicism about the fairness of the count will continue.

UPDATE:  I should probably take back "devoid of fact" with respect to the Byron York half of the "smackdown."  His latest reply to Lampley  in the thread catches up to the state of the debate as of about mid-January, and includes a quote from a certain blogger/pollster you may recognize. 

[Hat tip to reader Rick Brady for emailing links to the original Lampley post and the AP story]

Related Entries - Exit Polls

Posted by Mark Blumenthal on May 11, 2005 at 08:44 AM in Exit Polls | Permalink

Comments

Jim Lampley can't tell squat from Shinola about boxing, his "area of expertise." Why would he know squat about anything else.

Posted by: David | May 11, 2005 7:43:57 PM

Jim Lampley can't tell squat from Shinola about boxing, his "area of expertise." Why would he know squat about anything else.

Posted by: David | May 11, 2005 7:53:23 PM

Great post. It validates how I thought Lampley got his job. He's pretty, not smart.

'Social Research'? Does he mean statistics or sociology?

'professionally conceived samples' What? WTF is that?

Hilarious.

Posted by: Fides | May 11, 2005 9:14:14 PM

Until Republicans and Democrats find a way to agree to clean up the process, cynicism about the fairness of the count will continue.

I hear zip, zero, nada from Dems about cleaning up the process. Instead, the Dems are pulling out all the stops to block the most elementary and unobjectionable reform - the requirement to show a photo ID.

This isn't about finding a way to clean up the process that the Dems agree to. This is about jamming reforms down their throats, if that is what it takes.

Posted by: R C Dean | May 11, 2005 10:02:51 PM

But RC Dean, requiring ID could suppress the vote. We want to encourage voting, not supress it. So what if it leads to a little fraud... ;-)

Posted by: Rick Brady | May 11, 2005 11:13:14 PM

So how do we improve our voting systems? Four areas of the process need to be addressed:

1) Registration-we need to ensure that only legal voters are registered. Do we do a national database? Do voters get a "citizen number?" How do we handle snowbirds and other dual state citizens-ask them to declare what state they are voting in for national elections?

A uniform national date for closing registrations would be enforced so that the national voter database can be reconciled.

Registration can take place after this date (including same day registration) but these votes would be provisional until the voter is confirmed as eligible.

2) Voting process-a paper trail is a requirement. I actually like the fact that different states use different technology-it makes the system less prone to corruption-you have to hack lever machines, and punch cards, and DREs, and optiscan.

I have no objections to requiring government issued picture ID.

I'm not sure what can be done about long lines, which was a reported problem, especially in Ohio. More backup equipment that can be deployed to areas with long lines? I know in my polling station (optiscan) there weren't enough booths, so I just sat filled out my ballot while standing in line. (I'm in CA-the ballots get long with all the initiaves, and if you wait until you're int eh booth to learn about the issues, it can take a while to vote.)

3) Tallying process-this should be open at least to designated representatives of all parties. A reconcilation to ensure that the number of votes counted match the number of votes cast should be quickly done. (It took WI six months to do it for Milwaukee-way too long.)

4) enforcement-individuals convicted of fraud should lose their voting privileges for life. A conspiracy to commit fraud should result in hefty fines and up to life in prison.

Just my 2 cents.

Marty

Posted by: Marty H | May 12, 2005 1:27:33 AM

MysteryPollster,

Your puny efforts at reason and discourse are no match for links to other people's work coupled with boxing analogies and comparisons to Scott Peterson and O.J.

http://www.huffingtonpost.com/theblog/archive/2005/05/to-byron-york-and-other-o.html

But none of that is necessary, because the entire Edison/Mitofsky report is irrelevant to the argument, given that it is based on the assumption the final official vote tally is accurate. Make no mistake: my argument is that the final official vote tally is anything but accurate, that it is the product of massive vote fraud carried out through the programing of Diebold voting machines and various other machinations aimed at suppressing, destroying or losing Kerry votes.

Down Goes Frazier, errrrr ... Byron York!

My argument is that what were accurate were the exit polls. As one Ivy League research methodologist has noted, "Apparently the pollsters at Mitofsky and Edison have found it more expedient to provide an explanation unsupported by theory, data or precedent than to impugn the machinery of American democracy."

Various statisticians have reported that the odds on the occurrence of variances from exit polls to actual results such as were produced in this election range up to 959 000 to 1. Sounds like DNA. As US Count Votes notes in a statistical abstract, "No matter how one calculates it, the discrepancy cannot be attributed to chance."

My goodness - to use another boxing analogy:

Jim Lampley - Tomato Can.

Posted by: BumperStickerist | May 12, 2005 9:56:24 AM

The problem with any poll is simple, as my insurance claims adjuster wife observed this morning about changes and clear discrepancies about eyewitness testimony to an accident: People lie.

Posted by: Kevin Murphy | May 12, 2005 11:07:16 AM

Lampley has a new response (tally 5 now)

http://www.huffingtonpost.com/theblog/archive/jim-lampley/to-byron-york-and-other-ostriches.html

In which I guess York and mystery pollster is an ostrich?

here is the text...


To Byron York and Other Ostriches

Byron York has treated me fairly and without rancor, and I am grateful for that. Certainly I am more in his wheelhouse than mine, and I'm honored that he saw fit to engage me in this little set-to we've conducted since Monday. I fired a lead right, Rep. John Conyers shouted encouragement from my corner, then York delivered a hook to the body. I shot back an uppercut, then he loaded up a right hand and attempted to bring an end to the discussion.

Byron York's most recent refutation of my charge that irregularities in the 2004 Presidential election demand criminal investigation cites quotes from the report of Edison/Mitofsky, the two-company partnership which provided exit polls to the major television networks, on the vast discrepancies between those polls and the official results of the election. The report, which Mr. York has helpfully highlighted in his second post and which runs to about eighty pages, essentially offered the conclusion that an five-and-a-half point gap between final poll numbers and the national popular vote tabulation-- a variance more than four times the statistical margin for error of 1.3%-- can be attributed to shy Republicans. The Washington Post summarized the conclusion: "procedural problems compounded by the refusal of large numbers of Republican voters to be surveyed led to inflated estimates of support for John Kerry." With this, in effect, York dismisses the exit poll variance argument.

I could go on at length here about the curious disconnect between the actual data in the report and its guesswork conclusion, how Edison/Mitofsky systematically validate all their sampling choices and their methodology, in effect eliminating any logical underpinnings for their ultimate summation, all the while selectively ignoring the lopsided skewing of pro-Bush discrepancies in the most critical swing states. I could spend some time dissecting what I believe is an obvious whitewash, a delicate sidestep away from the potential public relations disaster of being tied forever to the most notorious election theft in history.

But none of that is necessary, because the entire Edison/Mitofsky report is irrelevant to the argument, given that it is based on the assumption the final official vote tally is accurate. Make no mistake: my argument is that the final official vote tally is anything but accurate, that it is the product of massive vote fraud carried out through the programing of Diebold voting machines and various other machinations aimed at suppressing, destroying or losing Kerry votes. My argument is that what were accurate were the exit polls. As one Ivy League research methodologist has noted, "Apparently the pollsters at Mitofsky and Edison have found it more expedient to provide an explanation unsupported by theory, data or precedent than to impugn the machinery of American democracy."

Various statisticians have reported that the odds on the occurrence of variances from exit polls to actual results such as were produced in this election range up to 959 000 to 1. Sounds like DNA. As US Count Votes notes in a statistical abstract, "No matter how one calculates it, the discrepancy cannot be attributed to chance."

So let me put it in Foxspeak. If all the circumstantial evidence related to potential vote fraud in this election were gathered up into one big file for the Scott Peterson jury, they'd convict. The jury that might look at all this and acquit? O.J. Simpson. Politics make strange bedfellows.

Posted by: topsecretk9 | May 12, 2005 11:56:59 AM

Lampley has a new response (tally 3 now)

http://www.huffingtonpost.com/theblog/archive/jim-lampley/to-byron-york-and-other-ostriches.html

In which I guess York and mystery pollster is an ostrich?

here is the text...


To Byron York and Other Ostriches

Byron York has treated me fairly and without rancor, and I am grateful for that. Certainly I am more in his wheelhouse than mine, and I'm honored that he saw fit to engage me in this little set-to we've conducted since Monday. I fired a lead right, Rep. John Conyers shouted encouragement from my corner, then York delivered a hook to the body. I shot back an uppercut, then he loaded up a right hand and attempted to bring an end to the discussion.

Byron York's most recent refutation of my charge that irregularities in the 2004 Presidential election demand criminal investigation cites quotes from the report of Edison/Mitofsky, the two-company partnership which provided exit polls to the major television networks, on the vast discrepancies between those polls and the official results of the election. The report, which Mr. York has helpfully highlighted in his second post and which runs to about eighty pages, essentially offered the conclusion that an five-and-a-half point gap between final poll numbers and the national popular vote tabulation-- a variance more than four times the statistical margin for error of 1.3%-- can be attributed to shy Republicans. The Washington Post summarized the conclusion: "procedural problems compounded by the refusal of large numbers of Republican voters to be surveyed led to inflated estimates of support for John Kerry." With this, in effect, York dismisses the exit poll variance argument.

I could go on at length here about the curious disconnect between the actual data in the report and its guesswork conclusion, how Edison/Mitofsky systematically validate all their sampling choices and their methodology, in effect eliminating any logical underpinnings for their ultimate summation, all the while selectively ignoring the lopsided skewing of pro-Bush discrepancies in the most critical swing states. I could spend some time dissecting what I believe is an obvious whitewash, a delicate sidestep away from the potential public relations disaster of being tied forever to the most notorious election theft in history.

But none of that is necessary, because the entire Edison/Mitofsky report is irrelevant to the argument, given that it is based on the assumption the final official vote tally is accurate. Make no mistake: my argument is that the final official vote tally is anything but accurate, that it is the product of massive vote fraud carried out through the programing of Diebold voting machines and various other machinations aimed at suppressing, destroying or losing Kerry votes. My argument is that what were accurate were the exit polls. As one Ivy League research methodologist has noted, "Apparently the pollsters at Mitofsky and Edison have found it more expedient to provide an explanation unsupported by theory, data or precedent than to impugn the machinery of American democracy."

Various statisticians have reported that the odds on the occurrence of variances from exit polls to actual results such as were produced in this election range up to 959 000 to 1. Sounds like DNA. As US Count Votes notes in a statistical abstract, "No matter how one calculates it, the discrepancy cannot be attributed to chance."

So let me put it in Foxspeak. If all the circumstantial evidence related to potential vote fraud in this election were gathered up into one big file for the Scott Peterson jury, they'd convict. The jury that might look at all this and acquit? O.J. Simpson. Politics make strange bedfellows.

Posted by: topsecretk9 | May 12, 2005 11:57:36 AM

Rick wrote:

"But RC Dean, requiring ID could suppress the vote. We want to encourage voting, not supress it. So what if it leads to a little fraud... ;-)"

Rick, RC Dean made a very ill informed comment, and it would have been better to point out the burning desire on the part of Democratic voters for election reform.

It should be very instructive that Mark is a Democratic pollster.

I would really want the MP blog to maintain its incredibly high level of discourse.

If RC Dean wants to join the discourse maybe he could explain why he thinks Democrats say zip about cleaning up the process. I don't think any serious minded readers of this blog could accept such a frivolous statement.

Posted by: Alex in Los Angeles | May 12, 2005 4:15:19 PM

Lamply has entered the land of the non-disprovable hypothosis. Like Creationists who see proof of intelligent design in there own doubts about evolution, or believers in anthropegenic global warming, for whom a bitter blizzard provides proof, not only that the Earth is warming, which it is, but that Man is causing it. I know it is tempting to argue with him, on the grounds that he might see the obvious, but he never will.

Posted by: moptop | May 12, 2005 9:39:06 PM

Looks more to me like a post hoc ergo propter hoc approach to thinking on Lamps' part.

Who was that lady who, being flabbergasted that Nixon won in 1972, said "But nobody I know voted for him..."?

I hope Lamps comes down from this. I actually once used to have respect for the man, but he is getting destroyed by York.

Not an auspicious beginning for Lamps the Blogger...

Posted by: JD | May 12, 2005 11:16:24 PM

What Lampley is saying is: Everybody knows that between a heavily scrutinized election with 120 million respondents, and some exit poll with 13,000 respondents conducted by inexperienced college kids, the exit poll is bound to be more reliable. And since the actual vote tracked the consensus of the top 10 pre-election polls, it is obvious that those pre-election polls were also fixed. And not only did Karl Rove fix the 2004 election, he also fixed the 2000 and 1996 elections (which showed a similar exit poll bias). He also fixed the 1997, 2001 and 2005 British elections (which the Labor Party, not the Conservatives, won).

I ran a statistical analysis and there's a 950,000 to 1 chance that Lampley is looneytunes.

Posted by: CivilWarGuy | May 15, 2005 12:41:11 AM

Civil War Guy, your comment made me laugh. One correction though: "nd not only did Karl Rove fix the 2004 election, he also fixed the 2000 and 1996 elections." The polls were actually much closer in 2000 and 1996, but almost as biased in 1992. The point is that all the presidential election polls since 1988 showed Democratic bias. The bias was largest in 2004, but not much larger than 1992.

Posted by: Rick Brady | May 15, 2005 1:03:09 AM

The comments to this entry are closed.