The second type of exit poll, called a parallel election by Lynn Landes, and done by various citizen activists in several states in every election since 2004, involves getting as many respondents as possible to fill out a facimile "ballot", and having them write their names in a poll book at a table outside the 100 foot mark (back before we were allowed within the 100 foot mark). The names and signatures, which affirm that they voted the same way in the parallel election as they did in the official election, are also useful in determining what is each participant's political party. A listing of all the voters, with registered party, on the election day can be obtained later from the Board of Elections. Knowing their political party allows for correction for oversampling one party. One could also do this in any type of exit poll, though it would have to be specified in the question, which election or year you were asking their party affiliation, since in some states such as Ohio, a voter can change their affiliation when they vote at the polling location. For instance, the survey question might ask," In which party were you registered for the 2004 election?" or the 2006 election? or both. We want to be able to show we did not oversample one party, or that we corrected for the oversampling. In a Primary, this is not an issue, since we are not comparing candidate results across parties, though the crossover vote might be interesting to survey.
The third type, promoted by Mark Adams of ProjectVoteCount.com, uses actual affidavits as surveys. In Florida, a written affidavit, even though not notarized, is legally binding. The plan is that with diligent poll taking, a high number of participants would provide enough votes for a losing candidate or issue to show fraud. The affidavits would then be evidence in court. It would be possible to gather enough surveys to do this, but a high percentage is required, 80% or higher probably. With such a high percentage of participants, even if more total votes than the official count are not found, the percentages of the candidates totals are still significant to show probably fraud, if they vary greatly from the voting machine count percentages.
Both professional pollsters and unpaid citizen pollsters are now using a blend of these types of exit polls. In the Pennsylvania Primary on April 2008, we used a survey that asked race, age range, and gender, but not party, and not their name. It also asked if they liked their current method of voting--paperless touch-screen machines--or if they preferred a voter-verified paper ballot. Another question was if they had any trouble voting. The surveys were collected in several locked file-type boxes with a slot that allowed the whole survey, without folding, to be inserted. Pollsters usually placed the surveys in the boxes. In our parallel elections we asked the voters to place their own ballots in the boxes, to demonstrate chain of custody. Professional Edison/Mitofsky pollsters were observed in the SC Primary using voter filled-out surveys, with every voter asked if they would like to participate.
The primary difference between professional and citizen-run exit polls, is one of purpose. Professional pollsters seemed in 2004 to see their function as providing early results to TV commentators, but since results are obtained so fast now from early reporting precincts, their function now seems to be simply to provide demographic data on what groups of people, such as Evangelicals, Hispanics, white women over 50, young parents, and so on, are voting for which candidates. Professional pollsters no longer pretend to be providing a check on the accuracy of elections, as exit polling has done for years, all over the world , in countries that claim to have democratic elections. Citizen exit pollsters are taking over the function abandoned by professional exit pollsters. Citizen exit pollsters can work with statisticians in local universities to design surveys and to use polling techniques that provide the most scientifically significant results to show whether an election is accurate or suspect. I recommend that surveys ask political party preference in the last two federal elections, eg, 2004 and 2006, age range, gender, and race, and that at least two contests are included, one as a control, one as the main race, e.g., President in 2008. The surveys should be standardized, which we did not do so well in the Ohio primary exit poll, due to last minute organizing.
Other questions such as attitudes toward electronic voting could also be interesting. Precinct location is necessary. I believe I would make names optional, or not ask for them at all, unless you are fairly certain that one candidate will be cheated, and you can convince the voters that the secrecy of their ballot is not as important as stopping a fraudulent election. A high rate of participation is best, though 30-50% participation should still yield valid percentage results. Pick a size precinct that your available exit pollsters can cover 80%. We demonstrated in PA that five diligent exit pollsters, working all day, and taking breaks in lulls, can poll about 70% of 2200 voters. We also had the help of about 4 others sporadically. Consult a statistician for the validity of your percentages polled. Diverse locations would be good since we know that high concentrations of either major party can be targeted for fraud, by swapping votes. But also having two polling locations close to each other will verify each other. Multi-precincts provide lots of voters, but people don't know their precinct numbers or will write the wrong one. You can lump the data, however. Ask whether they voted provisional or regular ballot, because provisionals should not be in the totals to compare to the machine counts. Don't tell officials where you are going to be, or the citizens either till the last minute, or the word will get out, and any machine rigging may be "unrigged." One person should write down demographic data for the ones who refuse to be surveyed, or who otherwise "get away."
4. BE EXTREMELY NONPARTISAN
Be extremely nonpartisan, which includes dressing neutrally, and treating everyone the same, with respect and eagerness to have them participate. Do NOT talk about your political preferences. I would not even use the event as an opportunity to hand out literature about the failure of electronic voting machines. Republicans, at least in Ohio, tend to like touch screen voting and Democrats optical scanners. If you are promoting OpScans or hand counting, you may skew your results toward one party's voters. Best to remain neutral. It would be better not to poll your own neighborhood, since people may know how you vote.
5. LOOK for ANOMALIES
In your analysis of the results (after counting and tallying in the presence of other citizen witnesses, who should also help with the counting), you are looking for anomalies. You should first be able to establish accuracy in some of the races, for credibility--hopefully not all of the races will be rigged! You are also looking for skewing of results on one direction. No one will believe fraud has occurred if there was significant rigging on both sides, though in today's political climate and the proven insecurity of electronic voting, anything is possible! Of course you are also looking for your totals for a candidate or issue surpassing the machine counts. If this happens, you will wish you had asked for affidavits, but that may also have affected your ability to get high numbers of participants. Catch 22.
No exit poll is perfect. Most of the ones I have participated in do not prove fraud, but many certainly indicate areas that need a recount or audit. BUT, strong indications of fraud, across many counties and precincts, if we do many exit polls across the country in 2008, will be enough to convince many more citizens that our elections are bogus, and may trigger a recount by a third party candidate. Exit polls also may scare riggers away from manipulation.
Sign Up
We need citizens to sign up in their states now. One person in each state needs to take the lead. If we, mainly Victoria Parks, can organize eight counties (22 precincts) in less than two weeks, with the help of Ron Paul meet-up folks, our progressive lists, and ProjectVoteCounts's website, then so can others. Citizen oversight is the watch-phrase for honest elections.
People say,"An honest elections cow, I don't get it." So I say back, "Well, it's obvious I am a cow, see, udders, horns, spotted fur, big head, floppy ears. And I happen to think that making our elections honest again is about the most inportant thing we can do for our supposed Democracy. Thus, Honest Elections cow. What's so hard to get?" moo
