CA-10: Garamendi Poll Riddled With Errors

X-posted at The Progressive Connection

The more I find out about the CA-10 poll John Garamendi released on Monday, the worse it smells. The press release about this poll from the Garamendi campaign gave out limited information as to how the poll was conducted, which raised a number of questions. However, Peter Charles left a comment at Calitics where he shared more information about the details of the poll. Those details exposed three glaring errors that jumped out from the information that was provided to the poll’s participants.

When the participants were given bios of the three Democratic candidates, here’s what they heard about Mark DeSaulnier, Joan Buchanan, and John Garamendi:

3a. Democrat State Senator Mark Desaulnier has served in the state legislature since 2004. Before that he served on the Contra Costa Board of Supervisors from 1996 to 2004. His top priority issues will be rebuilding the country’s economy, implementing clean energy programs, and regulating Wall Street banks. He is endorsed by Congress members Ellen Tauscher and George Miller, local firefighters, teachers, police and environmental groups.

3b. Democrat State Assemblywoman Joan Buchanan was elected to the State Assembly last November. Before that she served for 10 years on the San Ramon (ruh-MOAN) Valley school board. Her top priority issues will be more jobs and improving the economy, increasing renewable energy programs, and reforming public education. She will likely be endorsed by local elected leaders, school board members, teachers, and civil rights and womens groups.

3c. Democrat John Garamendi is California’s Lieutenant Governor. He has lived in the Sacramento portion of Congressional District 10 for 30 years. He previously served as Deputy Secretary of the Interior for Bill Clinton. He is running for Congress to continue reforming health care, rebuild our economy around clean energy, and reform bank and credit card laws. He will be endorsed by local nurses, firefighters, teachers, police officers as well as former President Bill Clinton and Al Gore.

The problem here is that the pollster misrepresented all three candidates, and he did it in a way that predictably favored the candidate who hired him.

As we’ve noted extensively at The Progressive Connection, John Garamendi does not now, nor has he ever “lived in the Sacramento portion of Congressional District 10.” That’s just an outright falsehood, as are the representations made about both DeSaulnier’s and Buchanan’s record of public service.

Mark DeSaulnier joined the state legislature in 2006, not 2004. Before that, DeSaulnier served on the Contra Costa Board of Supervisors from 1993 to 2006. Thirteen years, not eight. Four terms, not two.

And Joan Buchanan served on the San Ramon Valley school board from 1990 to 2008.  Eighteen years, not ten. Five terms, not three.

You have to ask yourself this question. If a pollster will lie about both his own candidate’s and the opposing candidates’ biographies, what else will he lie about?

And those aren’t the only problems. To get into the really wonkerific world of why Garamendi’s polling sample is all wrong, flip it…

This is how the Garamendi press release described the polling sample:

The poll was conducted between May 1st and May 4th by Jim Moore and JMM Research. Interviews were conducted with a 400-person sample from the 10th congressional district. Turnout was projected at 30 percent, and likely turnout was projected to be 55 percent Democratic, 33 percent Republican, and 12 percent independent. The poll had a +/- 5 percent margin of error.

For starters, a +/-5 percent margin of error is large. For those of you who are new to polling, +/-5 means that you need to give a five percent leeway to each candidate in the race. Relative to this particular poll, that means that Garamendi’s numbers could be as much as 5 points higher or lower, AND DeSaulnier’s and Buchanan’s numbers could also be as much as 5 points higher or lower. So while, at one extreme, it could reflect a blowout with Garamendi at 29 and DeSaulnier/Buchanan at 8/6, the converse could also be true. You could also have Garamendi at 19 and DeSaulnier/Buchanan at 18/16 — or a dead heat.

But there are also real and significant flaws in the poll’s underlying assumptions.

To explain this part, I’ve gathered the following data from the March 2009 SoS registration statistics, along with the CA-10 primary results from June 2008, and the Proposition results from CA-10 in the June 2008 primary.

Voter Data for CA-10

Total Voters % Total Dems % Total Reps % Total DTS % Other %
March 2009 Registration for CA-10 367,306 100% 173,498 47.24% 106,275 28.93% 73,015 19.88% 14,518 3.96%
June 2008 CA-10 Primary Results w/win % 85,814 N/A 55,427 64.59% 30,324 35.34% N/A N/A 63 .03%
CA-10 Voter Turnout as a % of Registration 85,814 23.36% 55,427 31.95% 30,324 28.53% N/A N/A 63 .43%

Got that? 23.36% voter turnout in last June’s Congressional primary.

But what’s fascinating about this is that Democratic and Republican registration amounts to 76.17 of the total voters; DTS and the others account for the remaining 23.84%. Now, if you were registered either Democratic or Republican for the June primary, you automatically received a partisan ballot. If you were DTS, you would have received a partisan ballot only if you specifically requested it; otherwise, you would have received a ballot that did not allow you to vote in, specifically (for our purposes), the Congressional race. Of the remaining parties, only Peace & Freedom fielded a candidate, so the members of the other parties would also not have had a Congressional vote.

Now, there were two propositions on the ballot as well, where all voters could cast their vote, regardless of party affiliation. It’s interesting to note that the total number of votes cast on Props 98 and 99 (regarding eminent domain) in CA-10 was significantly higher than the number of votes cast in the CA-10 Congressional race.

Total Voters % Turnout
CA-10 Congressional Race 85,814 23.36%
Prop 98 113,139 30.8%
Prop  99 112,584 30.65%

So we can assume that at least 113,139 voters turned out in CA-10 in the June primary. But of those 113,139, only 85,814 cast a ballot in the partisan Congressional race — or 75.84% — almost exactly equivalent to the 76.16% who are registered with the two major parties. That would tend to prove the notion that DTS voters seldom pull Democratic or Republican ballots in primaries.

All of which is a really long-winded explanation for why I don’t see any way to justify setting up a poll to reflect a 12% DTS participation in the coming CA-10 special primary.  That pretty much leaves us with a deeply flawed +/-5 MoE poll that, in all likelihood, overestimates turnout and takes 12% of its data from people who are demonstrably unlikely to vote in the upcoming special election (remember, the margin of Garamendi’s lead is 11%), while undersampling Democrats by 10%.

Obviously, the Garamendi campaign was hoping to make a big splash with this poll and establish its candidate as the strong frontrunner. Instead, because the poll is so thoroughly riddled with errors of both fact and judgment, they’re just looking desperate.