Election Squad!

The Exit Polls Exit poll numbers from this year's election have been all over the board, and a hotly debated topic. The numbers depend on what method, what source, and even what time they were taken. We got these numbers from an MIT study that was done to investigate the statistical significance of the swing from exit poll results to the final vote counts. The study concluded that it was possible for these swings to happen, however, they noted that there was a 1 in 50,000 chance that all of these states could have made the swing. The significance test for us, NONE of the swings in exit poll results went in Kerry's favor across the country. It's also important to note, that CNN altered exit polls to compensate for the disparity between exit polls and reported vote (check it out). Interesting to note: Germany's election results are announced early in the evening based on exit polls. Their margin of error has been less than 0.1% for 2 generations.	..		So, what’s the problem with our exit polls? Probably nothing. Dick Morris, the first political consultant to Bill Clinton, who later became a consultant to the Republican party, and now a regular on The Fox News Network has some interesting thoughts on exit polls. “Exit polls are almost never wrong…according to ABC-TVs exit polls, for example, Kerry was slated to carry Florida, Ohio, New Mexico, Colorado, Nevada, and Iowa, all of which Bush carried. The only swing state the network had going to Bush was West Virginia, which the president won by 10 points.”
Mathematicians are interested in investigating the November 2004 election because the likelihood of election results being significantly different than exit polls in half a dozen battleground states is very very low. Check out this excellent report on the subject: The Unexplained Exit Poll Discrepancy(PDF) by Steven F. Freeman, PhD. (Cap Times article about Freeman) Regardless of why there were such swings with the exit polls, I think it's safe to say that they are misleading and can have an effect on the way people vote. For instance, if your sitting on your recliner after a hard days work in Ohio and and the news is showing your man is up by 4%, you might have less motivation to go fight for your team - they seem to be doing just fine without you. Especially when you're hearing stories about people waiting 5 hours to vote. At the same token, Gallup polls, released weeks before the election, had Bush with a ridiculous 14-point lead. The highly publicized early polling can have a large effect on the public's perception of the popularity of a candidate. MoveOn.org printed an ad in the NY Times, questioning the right-leaning pollster.

Electronic Voting (aka DREs)

This map shows the states that used electronic voting machines (DREs). There are two types of machines used in tallying votes: one that scans paper ballots, and DREs the ones that record the vote in the machine (with no paper trail). DREs have a number of security flaws that invite hackers and cheaters alike. DREs tallied about 30% of the votes during this election, up from 13% in the 2000 election.

In the key 20 swing states, it’s important to note the “success rate” of the DREs for the Republican Party. Bush won 79% of the swing states using DREs, while Kerry won 83% of the swing states not using DREs. The problem runs deeper when we look at Georgia (also using DREs since 2002), where they elected their first Republican Governor (Saxby Chambliss) in 134 years in 2002. This after the incumbent Democrat Roy Barnes had an 11 point lead in the polls going into the election. Numerous reports of missing memory cards and DRE malfunctions have been reported around Chambliss's unlikely 16 point swing. Read full story.

Florida Irregularities

Who was the biggest beneficiary of the focus on Ohio in this election? Florida. The state known for infamously deciding the 2000 election, avoided some of the public spotlight in 2004 thanks to Ohio. However, this year Florida seems to have even more strange circumstances that are causing researchers and statisticians to investigate. In many counties, there seems to be a baffling amount of excess votes for Bush, when compared with voter's party registration. We've highlighted some of the counties in question in the chart below. Some explanations for these voters swinging Republican are based on the "Dixie-crat" phenomenon. But theories like these attempt to debunk studies of major statistical anomalies without actual facts to back them up. In the graph below we are looking at real numbers that no one is disputing. (more on this debate)

• George W. Bush racked up more votes than there were registered Republicans in 47 out of the 67 Florida counties. It's certainly extraordinary this happened in 70% of counties, as well as the degree to which it occurred in many counties. (illustrated by the graph above).(Check out the full data for both E-voting and opti-scan counties at USCountVotes.org)
• In Florida, the registered Democrat/Republican ratios versus the reported Kerry/Bush vote had substantial statistical abnormalities occurred in mid-size counties that used the optical scanned ballots that were tallied on a PC, which are vulnerable to hacking. (Check out more charts and graphs illustrating this relationship)
• Another report from researchers University of California, Berkeley states 'Something went awry with electronic voting in Florida,' "For the sake of all future elections involving e-voting systems, someone must investigate and explain the statistical anomaly we found in Florida," the lead researcher said at a news conference. (full story)
• In 2004, Bush received 20,000 more votes than the number of registered Republicans (100.3%). In 2000, Bush received only 85% of the registered Republican vote.
• 72% of Floridians voted for an increase in the minimum wage, a Democratic measure, yet only 48% voted for Kerry. While this is not a direct correlation, we’d expect the gap to be smaller than 24%.

Alarming Statistics!
Wikipedia: the Free Encyclopedia has more stats		VotersUnite.org: log of machine malfunctions and other problems		• 268,159 more votes than voters in Florida

Adding up numbers is apparently very hard for the technology we've entrusted our democracy to.
Here's some examples that have surfaced:

• Franklin Co, OH

DRE

4,258 total votes recorded, and only 638 voters were registered. Only 260 voted for Kerry, so 3,893 extra votes went to Bush (USA Today Article)(CNN)

incident caught & corrected

• Warren Co, OH

Press was kept from witnessing tallying of votes because of a" terrorist alert" targeted at controversial swing state OH ...here's the story

terror alert!

• Mercer, OH

Unofficial returns showed 51,818 ballots cast but only 47,768 recorded for president story

(-4,050)

• Meckelburg Co, NC

4,000 extra votes story

+4,000

• Miami Co, OH

19,000 ballots were added after all precincts reported for county story

+19,000

• Guiltford Co, NC

optical scanners totaling the votes [ES&S]

22,000 vote discrepancystory

(-22,000)

• Palm Beach Co, FL

DRE

542,835 ballots tallied, 454,427 voters turned out (including absentee)
- this county had trouble in 2000 too.

+ 88,408
corrected?

• Cuyahoga Co, OH

What's up with Cuyahoga County?
Confusion surrounds impossibly inflated Republican turnout totals in some 30 suburban precincts. Discrepancy is attributed to computer flaw story

• Craven Co, NC

optical scanners totaling the votes [ES&S]