2006 general election
In 2006, riding a wave of backlash against the GOP, Democrats recaptured the House and Senate. To many, this was proof that election rigging for the GOP was a myth: otherwise, how could they have lost? However, multiple analyses suggest that GOP fraud was in place, but got swamped by eleventh-hour political sea changes in the Democrats' favor. 2006 was, in all likelihood, a denied Democratic landslide.
In 2006, Edison conducted a national exit poll of US House elections. The unadjusted exit poll showed Democrats doing nearly 4% (3 million votes) better than the national aggregate of official House results. While Edison and the media were quick to claim this red shift was caused by oversampling Democrats, the Election Defense Alliance (EDA) analysis debunked that theory. In fact, the exit poll might have actually been skewed towards Republicans.
Edison exit polls ask many questions in their survey: who you voted for, demographics (age, race, gender, income), party ID, political questions (approval ratings, voting history, policy issues), and more. The responses can be compared to past elections and surveys of the public at large to assess whether the unadjusted exit polls were skewed toward a particular party.
EDA first looked at three responses well-correlated with candidate choice:
- Bush approval: The exit poll had it at 42%, while the average of normalized pre-election tracking polls had it at 41%. If Democrats were oversampled in the exit poll, Bush's approval would be lower than 41%, not slightly higher.
- Congress approval: The exit poll had it at 36%, while the average of normalized pre-election tracking polls had it at 31%. Congress was Republican, so if Democrats were oversampled, Congress's approval would be lower than 31%, not significantly higher.
- 2004 presidential vote: The exit poll had the Bush-Kerry margin at 2%, compared to the 2.8% popular vote margin. Given the imprecision in exit poll crosstabs, they're quite close, and a Democratic oversample that caused a 4% discrepancy should have lowered the margin much more. This isn't even accounting for the irregularities in the 2004 election, which indicate Kerry may have been the true winner. If ones compares to the 2.5% Kerry win the 2004 exit polls, the 2006 exit poll almost certainly oversampled Republicans.
None of these three metrics make a case for Democratic oversampling. Many of them, if anything, imply the opposite.
They also compared voting demographics to UMichigan's 1994-2004 ANES assessment of the electorate:
- Race: The exit poll had 80% white voters (split between parties) and 10% black voters (overwhelmingly Democratic). ANES's white voter average was 74% (6% below the exit poll), and its black voter average was 13% (3% above the exit poll). In fact, the ANES white turnout never reached as high as 80%, and the black turnout never reached as low as 10%.
- Party ID: The exit poll had 39% Democrats, 35% Republicans, and 26% independents. ANES had continually shown a higher Dem than Rep party ID, with a 4% advantage at the low end of their historical margins. Adding independent leaners, the Democratic advantage had never dropped as low as 4%.
Once again, these two demographic measures disprove Dem oversampling in the exit polls, and quite possibly indicate GOP oversampling.
Having established the 2006 red shift's legitimacy from the unadjusted exit poll, EDA also looked at the adjusted exit poll. Adjusted exit polls are the result of forcing an unadjusted exit poll to match the official vote count. If this causes incongruities in the adjusted exit poll crosstabs, it's a good sign that the vote counts are wrong. The adjusted 2006 exit poll had Bush winning the 2004 election by 6% among its respondents, way over his official 2.8% margin (which may itself be fraudulent in Bush's favor). Other metrics also increased implausibly; for instance, Bush approval went up to 43% (2% above the tracking polls).
The unadjusted exit poll either matched the electorate or underrepresented Democratic voters. And the adjusted exit poll had illogical crosstabs, suggesting the official vote count was erroneously too Republican. Unless there was a massive Republican turnout surge that tracking polls failed to pick up on, which was clearly not the case, this necessarily indicates corrupt vote counts favoring the GOP.
In other words, the true Democratic victory was much larger (by 4%, or 3 million votes) than officially reported. 2006 was a Democratic landslide of massive proportions, reduced by an attempt to rig the vote that fell short.
The 2006 exit poll was confirmed by pre-election polls. Both the exit poll and the average of pre-election polls had Democrats winning by 11.5%.
Looking at the pre-election polls also hints at why the GOP rig failed. The Cook generic congressional ballot had a massive Democratic surge in October. Throughout the month, the Democratic advantage increased from 9% (50% to 41%) to 26% (61% to 35%). This was probably due to the Mark Foley sex scandal and the related GOP sex scandals that were uncovered. Last-minute political developments gave the Democrats a significant and unexpected advantage. It's likely that a GOP rig (which requires time to prepare and deploy) was put in place before this October surprise, and couldn't be recalibrated in time for the Democratic groundswell. The vote flip ultimately turned out to be too low, denying the Democrats a landslide but falling short of a GOP victory.
Competitive race targeting
EDA also found that competitive races were more likely to be red shifted. In a telephone exit poll, they asked voters who they picked in both competitive and noncompetitive elections (among governor, Senate, and House). The poll discrepancy for competitive elections was significantly higher than the discrepancy for noncompetitive elections, with the shift almost always favoring the GOP. Since the same set of voters answered for all races, the effect of sampling bias was controlled for. A sampling bias would skew both competitive and noncompetitive races, meaning it can't account for the higher incidence of red shifts in competitive races. The most viable conclusion is that competitive elections were better targets for election fraud, since they're easier to flip without attracting suspicion, and were thus rigged.
Not only does EDA's study of competitive races confirm their exit poll analysis, it puts it in greater context. The 3 million House votes shifted away from Democrats were likely concentrated in the close House races. This allows a rough calculation of how many seats the Democrats lost due to fraud.
Virginia Senate race
The 2006 House races were mainly analyzed, but it stands to reason that other races were rigged as well. In particular, the Senate would be a compelling target. Democrats won four Senate seats by 6% or less, and without the October surprise, vote rigging probably would have given them to Republicans.
Particularly noteworthy is the Virginia Senate race between Jim Webb and George Allen. With the Senate ultimately going 51-49 for Democrats, the balance of power in the Senate hung on this extremely tight race. Webb, the Democrat, eked out a close 0.4% win. It was widely expected that Allen would call for a recount, given the closeness of the race and its importance to the GOP. But instead, he conceded and forewent a recount.
This was shocking, giving the race's political importance to the GOP. According to Jonathan Simon, Allen conceded after reading a mysterious piece of paper someone handed onstage. The implication is that Virginia was an attempted GOP rig which failed, and Allen was forced to prevent its exposure. Virginia had many horribly insecure voting machines (such as the WINVote) which would have been easy to commit fraud on. But Webb overcame the rigging, so the GOP needed to ensure the election didn't face any scrutiny. Doing so would have uncovered the attempted fraud and likely undone the whole GOP rigging enterprise.
Unlike 2000, 2002, and 2004, the Democrats didn't have a victory stolen from them. But this was only due to a lucky October surprise. Had the GOP sex scandals not unfolded, the prepared GOP rig likely would have done its job and left Congress in Republican hands. Unfortunately, the Democrats' 2006 win was used to "debunk" the notion of rigged elections. This dangerous false sense of security ensured that election integrity wasn't taken seriously, paving the way for countless more stolen Democratic victories in the following years. 2008 probably would have been thrown to the GOP were it not for another Democratic surge.
- 2006 GA-4 congressional primary, in which Cynthia McKinney discovered many problems
- 2006 FL-24 congressional race
- Jonathan Simon and Bruce O'Dell, "Landslide Denied", 2006
- Election Defense Alliance, "Fingerprints of Election Theft: Were Competitive Contests Targeted?", 2011
- Voting Rights Task Force conference with Jonathan Simon on 2015/06/08 (part 3) @ 03:56 - describes how George Allen mysteriously conceded
- David Swanson, "Karl Rove Announces Plans to Steal Election", 2006/10/26
- Uncounted documentary footage of Cliff Arnebeck fighting against voter suppression efforts in the 2006 Ohio election
- FL-13 congressional race: Christine Jennings (D) vs. Vern Buchanan (R)
- FL-24 congressional race: Clint Curtis (D) vs. Tom Feeney (R)
- Brad Friedman, "Clint Curtis Files Congressional Challenge to Election of U.S. House Opponent Tom Feeney", 2006/12/26
- Election contest submitted to the US House
- "Walking for Democracy" video, 2007/02/23
- Jim Cirile, "EXCLUSIVE INTERVIEW: Clint Curtis Documents Evidence of Widespread 2006 Florida Election Fraud", 2007/03/04