This is an archive of the old Stones Cry Out site. For the current site, click here.

« A Response to Cindy Sheehan | Main | New Voters and the 2008 Election »

August 24, 2005

San Diego's "Parallel Election"

Diebold’s machines scanned and tallied every ballot cast in my home town of San Diego in a recent election to replace resigned Mayor Dick Murphy. Controversy surrounding electronic vote tabulation machines, Diebold’s machines in particular, prompted the group Citizens Audit Parallel Election to hold a “parallel election” in an attempt to verify the official vote count.

The Citizens Audit Parallel Election found a 4 percent “red shift” from the official vote count, away from progressive candidate Donna Frye. Carl Luna, a professor of Political Science at San Diego Mesa College and lecturer at the University of San Diego, wrote a remarkable editorial on the findings last week. Luna wrote:

Math is non-partisan

A team of statisticians from California State University- Northridge - have analyzed the data from CAPE, concluding that the probability of luck or chance as the cause of the observed four percent deviation is less than one in 1,300 - or .000678.

Unless you’ve had your head under a rock since November, you’ve heard similar probability based proclamations, as if sampling error is the only source of survey error.

The Citizens Audit Parallel Election paid for a hand count in full view of its representatives of a select number of precincts. That recount of the hand marked ballots matched the Diebold count, almost exactly.

A partial recount yesterday to test the accuracy of scanners that read ballots and tallied votes in the San Diego mayor's race July 26 revealed results that were nearly identical to those of the machines.

For example, a discrepancy of perhaps one vote occurred in a few precincts.

I agree with Jerry Ewig of Democracy for America.

The recount "does not prove the system is accurate and correct," said Ewig, who lives in Temecula. "We're still in need of a system that is verifiable and transparent to the people." (emphasis added)

But what about that parallel election (aka EXIT POLL) that called into question the election result? I am reminded Plissner and Mitofsky’s 1982 Public Opinion aritcle, “Voting Twice on Election Day”. Coming only two years after the first network projection based on exit poll results, the authors acknowledged potential problems with their “parallel elections.”

Voter polls and the more traditional sources of election predictions (precinct sampling) are subject to a mathematically definable sampling error. On top of that, voter polling presents a number of additional challenges. Things can – and have, at least in our experience – gone wrong. For example, there is no guarantee that voters who respond to the poll are like those who refuse to answer (pg. 15).
Something clearly went wrong with San Diego’s “parallel election.” In line with every presidential election exit poll since 1988, the poll was clearly biased toward the Democratic candidate. But why? That’s a question the polling industry should make every effort to answer, and rectify.

Free, fair, and falsifiable elections should be goal of every American, regardless of partisanship. But we know that we are a long way from this goal if we have to rely on exit polls as a verification tool.

If you are interested in election reform issues, the Election Audit Institute is a fine new venue for discussion.

Posted by Rick at August 24, 2005 11:58 AM

Trackback Pings