The Oregon Voting System: Statistical Evidence that it Works
Jan. 2, 2012
This statistical analysis of Oregon’s voting history provides evidence that the vote-by-mail system introduced in 1998 has been a success.
In 1988 Bush was the de-facto incumbent as Vice President. Dukakis led by 55.0-42.9% in the Oregon exit poll and won the recorded vote by 51.3-46.6%. He also won the unadjusted state exit poll national aggregate by 50.0-49.0%. But Bush won the national recorded vote by 53.4-45.6%.
In 1992 Bush was the incumbent. Clinton led Bush by 49.3-25.7% in the Oregon exit poll but won the state recorded vote by just 42.5-32.5%. He led the unadjusted state exit poll national aggregate by 45.7-34.8% and the national recorded vote by 43.0-37.4%.
In 1996, Clinton was the incumbent. He led Dole by 48.4-37.9% in the Oregon exit poll and won the state by nearly the same margin: 47.2-39.1%. He led the unadjusted state exit poll aggregate by 50.2-39.8% and the national recorded vote by nearly the same margin: 49.2-40.7%. His 54.7% two-party Oregon share exactly matched the National recorded share.
In 2000, Gore won Oregon by 47.0-46.5% and led the unadjusted state exit poll national aggregate by 50.8-44.5%. He won nationally by nearly the same margin (48.4-47.9%) and had a 48.3% Battleground share. The National True Vote Model indicates that he won by 50-47%. Nader had 6% in Oregon and 3% nationwide. Allocating the excess 3% Nader vote, Gore would have won Oregon by approximately 50-47.5%. He had 50.8% in the unadjusted state exit poll national aggregate.
The close match between the Gore’s recorded vote, pre-election polls, exit polls and True Vote Model indicates that Oregon was representative of the True National Vote.
In 2004, Bush was the incumbent. Kerry led the state pre-election polls by 48-47% and was projected to win by 51-48%. He led the Oregon pre-election poll by 50-44% and was projected to win by 53.7-45.3%. He won Oregon by 51.3-47.2%, a 3.6% improvement in margin over Gore. Kerry won the unadjusted state exit poll aggregate by 51.1-47.0% but lost the recorded vote by 50.7-48.3%. Kerry had a 53.6% national share in the True Vote Model – a 10 million vote margin. The model assumptions were 98% turnout of living returning Gore and Bush voters and 12:22am National Exit Poll with 8% of Gore voters defecting to Bush and 10% of Bush voters to Kerry.
Bush improved on his 2000 recorded vote share not only in the battleground states as well as in solidly Democratic New York. But Oregon went against the grain and shifted from Gore to Kerry. Kerry’s Oregon margin was 3.7% higher than Gore’s. This was primarily due to Kerry’s 65-13% lead in returning Nader voters and 57-41% edge in new voters. Kerry’s Oregon share was close to his 52% pre-election poll as well as the unadjusted state exit poll aggregate (51.1-47.6%).
Once again, the close match between Kerry’s recorded share (51.3%) and the aggregate of the state exit polls (51.1%) indicated that Oregon represented the True National Vote.
Therefore, we must conclude that the election was stolen in the Battleground states and that Oregon’s vote-by-mail system was virtually fraud-proof.
In 2008, McCain was the de-facto incumbent. Obama led 56-39% in the Oregon pre-election poll and won the state by 56.7-40.4%. He had 58.4% in the post-election survey. The True Vote model indicated 56.0-42.8%. Obama won the national recorded vote by just 52.9-45.6%, a 9.5 million vote margin, but he had 58.0% in the unadjusted state aggregate exit polls and 58% in the True Vote model. The triple match is powerful confirming evidence that the vote-by-mail system worked. Obama won the unadjusted National Exit poll (17836 respondents) by a whopping 61-37%.
We can conclude that Oregon’s votes were counted accurately – unlike the other states.
Again, the close match between Obama’s 2008 recorded vote (56.7%), unadjusted state exit poll aggregate (58.0%) and the National True Vote Model (58.0%) indicates that Oregon was representative of the national electorate.
In the 2010 Midterms, Ron Wyden (OR Dem Senate) won re-election with 57%, exactly matching the pre-election polls and Obama’s OR share. But popular progressive Democrats in other states such as WI, IL and PA were all defeated; their recorded vote shares were far below that of Obama, who won each state in a landslide. How come Wyden won handily but other progressives lost in WI, PA, IL? Did it have something to do with Oregon’s unique early voting system (mail and in-person) and it’s mandated hand recounts?
To believe that Oregon’s mail-in/early voting system miscounted votes, one must also believe that Bush did legitimately win all the other battleground states and therefore that the national and state exit polls that showed Kerry winning were all wrong. But what if the exit polls were correct? What if the votes were miscounted? Then one would have to conclude that Oregon’s system worked. The states used electronic voting machines, punched cards and levers.
True Vote Methodology
The analysis tables provide a reasonable approximation of the National, Oregon and Battleground True Vote shares.
Given 2000 and 2004 votes recorded and cast, the True Vote calculation assumes:
1. Kerry and Gore had 75% of the uncounted (cast – recorded) votes
2. Annual 1.25% voter mortality (5% in the four years between elections)
3. Equal 98% turnout of returning 2000 voters in 2004.
4. Equal Gore and Bush returning voter defection rates (they cancel each other).
5. Kerry won returning Nader voters by 65-13% over Bush based on the National Exit Poll.
6. New 2004 voters is the difference between 2004 votes cast and returning 2000 voters.
7. Kerry won new voters by 59-39% nationally. His DNV share in each state is calculated as:
State DNV share = 0.59* (1+state exit poll share – 0.5197)*new voters, where .5197 is Kerry’s unadjusted state exit poll aggregate (national) share. For instance, in NY, Kerry’s share of 1347k DNV is 894k = 0.59*(1+.645-0.5197).
The simplifying assumption is that there was zero net defection of returning Gore and Bush voters (they cancelled each other). But the 12:22am National Exit Poll of 13,047 respondents indicates that 10% of Bush voters defected to Kerry and only 8% of Gore voters defected to Bush. The 2004 True Vote analysis shown below indicates that Kerry had a 53.7% national share assuming a net 2% defection as opposed to 53.3% assuming zero net defection. See the Recursive True Vote Model.
Kerry True Vote Sensitivity Analysis
Two groups of three tables display the effect of various model input assumptions on Kerry’s vote share. The margin of error is less than 1.5%.
-New Voters and returning Nader/other voters
Three tables display Kerry’s National, Oregon and Battleground True Vote shares over a 54-63% range of new voters and 61-69% of returning Nader/other voters. Kerry wins all worst case scenarios (54% of new voters and 61% of returning Nader voters).
-Returning Gore and Bush Voter Turnout
Three tables display Kerry’s National, Oregon and Battleground vote shares for 91-99% turnout of living former Gore and Bush voters. Kerry wins all worst case turnout scenarios (91% of living Gore voters and 99% of living Bush voters).
Oregon vs. New York and California
In 2000, Gore won the recorded vote by 48.4-47.9%. In 2004, although returning Nader voters broke heavily for Kerry by 65-13% and new voters by 59-39%, Bush won by 50.7-48.3%. That is not plausible.
Gore won by 47.0-46.5%. With returning Nader and new voters breaking for Kerry, his recorded vote-count margin increased to 51.4-47.2%. That is plausible. Kerry led by 52.2-46.3% in the exit pollster telephone poll. That is plausible.
Gore won by 60.2-35.2%. Although returning Nader and new voters broke heavily for Kerry, his recorded vote-count margin declined to 58.4-40.1%. That is not plausible. Kerry led by 64.5-34.0% in the exit poll. That is plausible.
Gore won by 53.4-41.6%. Although returning Nader and new voters broke heavily for Kerry, his recorded vote-count margin declined to 54.3-44.1%. That is not plausible. Kerry led by 60.1-38.6% in the exit poll. That is plausible.
Why did Kerry’s margin increase in Oregon, a battleground state, and decline in strongly Democratic California and New York?
Why was the exit poll so far off in California (11.6 WPE)? It voted 29% on DRE touch screens, 66% on optical scanners and 4% on punch cards.
Why was the exit poll so far off in New York (12.2 WPE)? It voted exclusively on lever machines.
Why were the exit polls so far off (7.5 WPE) in the Battleground states? They voted on punched cards, levers, optical scanners and DREs.
Florida and Ohio
In Florida 2000, there were approximately 185,000 spoiled punch cards (under-punched and over-punched). According to the Census, 43,000 more votes were cast than recorded. Where did the 142,000 extra votes come from? Bush won Florida by 537 votes.
In Florida 2004, according to the Census, approximately 238,000 more votes were recorded than cast. How many were uncounted? Bush won by 380,000 votes.
In Ohio 2004, according to the Census, 143,000 more votes were recorded than cast. Approximately 300,000 were uncounted (see Was the 2004 Election Stolen? by RFK, Jr.) How many votes were switched? Bush won by 119,000 votes.
Oregon’s Pre-Election Polls Uniquely Matched the Recorded and True Vote
Final state pre-election polls were virtually all Likely Voter (LV) subsets of the full Registered Voter (RV) samples. Likely Voter subsets largely exclude “new” voters: first-timers and others who did not vote in the prior election. The Democrats won ‘new voters’ by an average 14% margin before Obama’s whopping 44%. Projections that ignore RV polls and focus solely on LV polls will inevitably underestimate the Democratic share, especially in heavy-turnout elections such as 2004 and 2008. In 2004, final pre-election projections were based on LV polls which understated voter turnout by 6%. Virtually all online political sites displayed LV polls (not RVs) and failed to allocate undecided voters.
Mainstream pollsters allocated 65-90% of undecided voters to Kerry. His projected national LV poll share was 1-2% lower than the projected RV share. In New York and California, pre-election poll projections were a virtual match to the recorded vote-count share. But they were 5-6% below Kerry’s exit polls and True Vote shares. The same LV/RV mismatch occurred in 2008. Obama had a 53% projection based on LV polls but had 57% based on RV national polls after allocating undecided voters.
Voting by mail results in high turnout, so the pre-election polls are RVs by definition. Kerry led by 50-44% in the final poll. After the undecided voter allocation (UVA), he was projected to win by 53-45%, matching the True Vote Model and within 1.6% of his recorded share. In the final weeks prior to the 2004 and 2008 elections, national LV polls were displayed on political websites; many did not allocate undecided voters.
By virtue of its vote by mail system, Oregon’s pre-election RV polls undermine the media’s objective of fooling voters into believing bogus vote counts. The media primes voters before the election with LV-only projections and then covers up the fraud with final exit polls that they always force to match the vote miscounts.
1988 – 2008: Patterns of Discrepancies Before and After Voting-By-Mail
Before Mail-In Ballots
1988 – Bush was Vice President. Dukakis had 51.3% in Oregon and 45.7% National.
He did 3.2% better in the OR exit poll.
1992 – Bush was President. Clinton had 42.5% in Oregon and 43.0% National.
He did 5.1% better in the OR exit poll.
1996 – Clinton was President. He had 47.2% in Oregon and 49.2% National.
He did 2.2% better in the OR exit poll.
After Mail-In Ballots
2000 – Clinton was President. Gore had 47.0% in Oregon and 48.4% National.
2004 – Bush was President. Kerry had 51.3% in Oregon and 48.3% National.
2008 – Bush was President. Obama had 58.4% in Oregon and 52.9% National.
Is it just a coincidence that when Clinton was the incumbent, there was just a 1.7% deviation between the Oregon and National vote shares?
Is it just a coincidence that when Bush was the incumbent, there was a 3.5% deviation between the Oregon and National vote shares?
If the True Vote Model is correct and Oregon reflects the national electorate, then what does that tell us about the electoral system?
Oregon County Vote Change Correlation
Since Oregon switched to mail-in ballots in 1998, there has been a noticeable decline in the volatility of changes in county vote shares from election to election. Before the switch to mail, there was a 0.93 correlation between 1996 and 2000 county vote share and a 5.0% standard deviation. After the switch, there was a near-perfect 0.98 correlation between 2000 and 2004 county vote shares and a lower 2.2% standard deviation in percentage vote change. There was an even better 0.99 correlation for 2004 and 2008. along with a very low 1.5% standard deviation in percentage vote change. The system is getting better and better.
The statistical analysis makes intuitive sense. Since the battleground states closely mirror the national electorate as by definition, Oregon’s recorded vote share should have been close to the other battleground states. But it was the only state that deviated sharply to Kerry. Oregon’s voting system is transparent. Optically scanned machine counts are verified by random hand-counts. Washington has also recently implemented a mail-in system.
Touch screen voting machine precincts avoid paper ballots; votes can be switched locally or at the invisible central tabulators. Optical scanners are a step in the right direction, but the system is ripe for fraud without a system similar to Oregon’s mandated random hand-count of selected precincts. Punch card machines can be rigged to void votes by double and triple-punching the ballots after the polls close – as occurred in Florida 2000. Corrupt election officials are quick to blame “stupid” voters for not properly filling out h the ballots.
Lever machines in NY, CT and PA did not use paper ballots; too few machines are placed in heavily Democratic precincts; defective machines that break down cause voters to leave the precinct; levers were “stuck” for Bush in 2004; lever gears can be shaved. Most important, tabulation of the votes is done on computers.
In NY, Gore, Kerry and Obama each enjoyed a 7% higher late (paper ballot) vote share than they did on Election Day levers. What does that tell us?
Here is an amazing statistic that very few are even aware of: Obama had 52% of the 121 million votes recorded on Election Day but he had a whopping 59% of the 10 million (paper ballot) votes recorded after Election Day. What are the odds of the 14% discrepancy? It’s like a 10 million sample-size exit poll.
Kerry won new voters by 59-39% and returning Nader voters by 65-13%. In order to believe the recorded vote, you must also believe that returning Gore voters defected to Bush at a much higher rate than Bush voters to Kerry. But according to the 12:22am National Exit Poll, 10% of Bush and 8% of Gore voters defected.
The above analysis indicates that Oregon’s mail-in system works just fine. It would be greatly appreciated if interested readers can find flaws in the assumptions, logic or the math and present contrary statistical and/or anecdotal evidence. The analysis should be forwarded to Oregon’s election officials who may then decide to scrap vote by mail and convert to HAVA-compliant DREs, Optical scanners, Punch cards or Lever machines.
Readers who believe that vote-by-mail systems are vulnerable to election fraud and/or voter fraud, should lobby state officials to oppose Oregon’s vote-by-mail system and keep their unverifiable voting systems.
Those opposed to 100% paper ballot voting by mail or hand-delivery cite advantages in precinct voting. These include a) voters meeting friends and making new ones, b) taking time off from work to vote, c) projecting a patriotic image by voting in full view, d) looking smart by touching the computer screen, e) exercising their legs while waiting to vote and f) getting free coffee.
If you believe the recorded 2004 vote was accurate in the battleground states, then you must also believe that…
1- Bush won a fair election.
2- The electronic and mechanical voting machines accurately counted the votes.
3- There was little or no fraud.
4- Election reform efforts are meaningless.
5- There is nothing wrong with our national voting system.
6- Oregon’s voting system was rigged for Kerry since it was the only battleground state he won that shifted sharply to him from Gore.
7- Pre-election state and national polls that projected Kerry would win by 51-48% after undecided voters were allocated were wrong.
8- Unadjusted and preliminary state and national exit polls that had Kerry winning by 5-7% were wrong.
9- The Oregon telephone survey that showed Kerry winning a 52.3% share was wrong.
10- Either returning Nader voters defected to Bush and/or he won a majority of new voters and/or more returning Gore voters than Bush voters defected.
11 -The National Exit Poll which had Kerry winning returning Nader voters by 65-13% and new voters by 59-39% were wrong.
12- The Oregon vote must have been padded for Kerry (51.4%) and Obama (56.7%).
13- Election officials in Florida, Ohio, NY and other states did a great job in making sure that the voting machines were not tampered with.
14- A problem with vote by mail is the elimination of exit polls. Oregon needs exit polls even though they are usually wrong.
15- The True Vote Model is flawed since it closely matched the unadjusted National, Oregon and Battleground exit polls.
16- There is nothing wrong with the standard policy of forcing final state and national exit polls to match the recorded vote.
17- Final 1992, 2004 and 2008 National Exit Polls were correct: there were millions more returning Bush voters from the prior election than were alive.
Election Model Forecast; Post-election True Vote Model
2004 (2-party vote shares)
Model: Kerry 51.8%, 337 EV (snapshot)
State exit poll aggregate: 51.7%, 337 EV
Recorded Vote: 48.3%, 255 EV
True Vote Model: 53.6%, 364 EV
Model: Obama 53.1%, 365.3 EV (simulation mean)
Recorded: 52.9%, 365 EV
State exit poll aggregate: 58.0%, 420 EV
True Vote Model: 58.0%, 420 EV
2012 (2-party state exit poll aggregate shares)
Model: Obama 51.6%, 332 EV (Snapshot)
Recorded: 51.6%, 332 EV
True Vote Model: 55.2%, 380 EV