|"Only dull people are brilliant at breakfast"
|"The liberal soul shall be made fat, and he that watereth, shall be watered also himself."
-- Proverbs 11:25
The poll was conducted on Election Day by Election Defense Alliance, Protect California Ballots, and ElectionIntegrity.org and was designed and researched with the help of at least one well known exit pollster, Ken Warren of St. Louis University's The Warren Poll, for the express purpose of measuring the accuracy of the reported vote count. It functioned beautifully in general, by confirming the results of most of the issues and races on the ballot. On Proposition 4, for example, which concerned a similar hot-button issue --- parental notification for abortion --- polling results and official election results matched within 2%, well within the expected margin of error.
However, for Proposition 8 only, the official results varied from the Election Verification Exit Poll by an average of 7.75% in the 19 precincts polled. In some cases, the discrepancy was as high as 17.7%. That is, of course, far outside of the margin of expected error and certainly worthy of further investigation by officials.
The creators of the poll, along with the analysts of the results and the authors of the study, seem to have gone out of their way to preempt the usual reasons for questioning and/or dismissing the methodology and findings of such polls. For example, as noted in the group's press release [emphasis added]...Data was collected through exit polls in 19 Los Angeles County precincts on Election Day, 2008. Voters leaving the polls were asked to fill out a simplified paper ballot anonymously and deposit it into a locked box; 6,326 voters did so, a sample larger than that for the entire state of California in the exit poll used by news outlets around the country to predict election outcomes.
The Election Verification Exit Poll (EVEP) itself was designed to address the propensity of voters to lie to exit pollsters, as has been suggested of late when official exit polls failed to match up with official results in recent elections. The EVEP was conducted anonymously. As voters exited the polling place, they were asked to fill out a simplified ballot echoing the votes they had just cast, and to place it into a locked box. Therefore, the ballots did not include any identifying information, and thus, those overseeing the poll --- at either the polling place, or later when the EVEP ballots were counted by hand --- would have no way to tie votes to voters. That process is in contrast with official media exit polling where pollsters directly ask voters to reveal how they voted.
While the EVEP process can't guarantee that voters still wouldn't lie in their anonymous responses, or make errors on their exit poll ballot, the likelihood is believed to be greatly lowered. The discrepancies reported from those EVEP results give cause for serious concern --- particularly as other initiatives on the ballot failed to show similar discrepancies. Only Prop 8 was off by this kind of margin.