To support its claims, the group, known as USCountVotes.org, shows that exit polls taken on Nov. 2 cannot be reconciled with announced vote tallies in some states.
The discrepancies, the organization points out, are worse in the U.S. than in one of two exit polls conducted during Ukraine's recent national election, which resulted in the results being thrown out by that country's Supreme Court last December. The two Ukrainian exit polls showed a discrepancy between the expected vote count and the official vote count of 4.7 to 10.7 percentage points. In the U.S., both USCountVotes.org and the company that conducted the presidential exit polling agree that the official vote count is at least 5.5 percentage points different from the expected vote count.
Americans, to be sure, are still exhausted from the 36-day legal ordeal that followed the 2000 presidental election. They understandably want the 2004 election to be wrapped up and consigned to history. For this reason, I've avoided writing about various problems with the 2004 election that computer scientists have speculated on for the past several months.
When USCountVotes.org released its comprehensive statistical analysis on March 31, however, the situation dramatically changed. There can be no doubt that questionable practices affected votes for president in some states.
Because the legitimacy of the U.S. government rests on the expectation that it is fairly elected, it's in the interests of Republicans, Democrats and Independents alike to examine the new evidence and use it to correct whatever problems are found. In addition, I believe this situation can show how computer science can be used when business groups with different interests want to assure themselves that any contested process will be handled faithfully.
Why Exit Polls Are More Accurate Than Others
Exit polling is a well-understood science. It's precise enough to be used in countries around the world by election monitors to detect tampering with official vote counts. There are various reasons for this:
• Sample size. Most pre-election state and national polls involve only 400 to 1,200 participants. If the sample is truly random, these polls are said to have a 95% chance of being within 3 to 5 percentage points of the results you'd get by asking every actual voter. By contrast, exit polls on election day have many more participants.
• Results on the ground. The 2004 U.S. exit poll involved 1,480 polling locations alone, plus surveying absentee voters. Of those voters who were approached when leaving the polls, about 53% agreed to answer the pollster's questions. Due to the much larger sample, exit polls often differ less than 1 percentage point from the official vote count.
• Fresh information. Exit polls also benefit from the fact that voters are being asked how they just voted inside the polling place only a few minutes earlier. Pre-election polls must ask people how they might vote "if the election were being held today."
The Official Explanation
Exit polling in 2004 was conducted for six major television networks and news organizations by the National Election Pool (NEP), a system designed by two private firms, Edison Media Research and Mitofsky International.
Because the exit polling produced figures that differred widely from some states' official vote counts, Edison/Mitofsky released a 77-page report on Jan. 19. The authors stated, "The weighted national survey numbers showed Kerry with 51% and Bush with 48% [a 3-point spread for Kerry]. The final national popular vote margin ended up being 2.5% for Bush. Thus, the national exit poll had an error of 5.5 points on the difference in the Democratic direction."
The Edison/Mitofsky report shows that the difference was not caused by selecting the wrong polling places. These locations proved to mirror the makeup of each state.
The report concluded that "Kerry voters were more likely to participate in the exit polls that Bush voters." This explanation seemed to satisfy the U.S. news media, and that seemed to be the end of that.
The Challengers' Explanation
The March 31 analysis by USCountVotes.org, however, demolishes the Edison/Mitofsky conclusion. The Ph.D.s who conducted the analysis found, using NEP's own data, that Bush voters actually had a slightly higher rate of participation in the exit polls than Kerry voters.
The signers of the critical analysis include such experienced hands as Frank Stenger, professor of numerical analysis at the University of Utah; Richard Sheehan, professor of finance at the University of Notre Dame; and Campbell Read, professor emeritus of statistical science at Southern Methodist University. These experts say in their paper, "No matter how one calculates it, the discrepancy cannot be attributed to chance."
Ron Baiman, Ph.D., is associated with the Institute of Government and Public Affairs of the University of Illinois at Chicago and is a signer of the report. He found that "statistically significant discrepancies of exit poll results from reported election outcomes were concentrated in five states, four of which were key battleground states," such as Ohio, Pennsylvania, and Florida. As most Americans know, the presidential vote in these states was close, and a shift in any one state would have tipped the election.
How The Votes Could Have Been Changed
Since ballot-box stuffing has a long and rich history in the U.S., the Ph.D.s behind the new report say they're disappointed that "Edison/Mitofsky did not even consider this hypothesis." Vote tampering has been attributed to both Democrats and Republicans in elections past. In the 1960 race, for example, John F. Kennedy won the state of Illinois by only 9,000 votes, and allegations that ballots were manufactured by the Daley political machine have never been settled with finality.
Asked how votes in one or more states could have been changed to affect the outcome in 2004, Bruce O'Dell, the vice president of USCountVotes.org, pointed out that about 30% of U.S. votes are now cast on equipment that cannot be audited. Once a count is issued by these electronic ballot boxes, the number cannot be double-checked against paper ballots.
"Many security professionals can identify a dozen different methods to alter these results," O'Dell said.
O'Dell pointed to Ohio, where a change of only about 60,000 voters would have given that state's electors, and thereby the election, to Kerry. A recount in that state was meaningless, he said, because (among other things) only a few, pre-selected polling places were recounted and equipment vendors had been inexplicably allowed to re-program election machinery after Nov. 2.
What We Know And What We May Never Know
For its part, the Edison/Mitofski report says, "We are in the process of an in-depth evaluation of the exit poll process in Ohio and Pennsylvania."
In a telephone interview, Joe Lenski -- vice president of Edison Media Research -- declined to comment on the USCountVotes.org analysis without prior approval of NEP's public relations chair. The immediate past NEP p.r. chair, Cathie Levine of ABC News, declined to answer questions via telephone and failed by press time to respond to an e-mail requesting comment.
USCountVotes.org is assembling a computerized database of precinct-by-precinct results from around the nation in hopes of finding more statistical patterns that can answer the questions that have been raised. The organization hopes to be ready by the 2006 and 2008 elections to use exit-poll data to catch vote tampering, if it occurs, almost immediately, O'Dell says.
This is a sensitive subject. I can hardly imagine a more divisive topic, on which my readers will certainly not be of one mind. For this reason, I plan to write a follow-up column next week, in which I'll include comments from at least one Republican, one Democrat and one Independent.
My only requirement is that you read the USCountVotes.org analysis and the original Edison/Mitofsky report. Send your comments to me by April 7 using the address on my contact page, as described below.
I know computer technology can be used for good or ill. I hope this is a case where technology can be used to reduce vote fraud, not to accelerate it.