Claussen on 1A Polling

UPDATED 9 p.m.: Rick Claussen, the veteran initiative consultant handling the sweeping campaign for all six California special election measures, is out with an interesting memo on the recent PPIC poll, which shows five of those six in trouble.

Claussen is self-interested obviously. But your blogger has found him to be very reliable in the past. (Most recently, your blogger said Prop 11, last November's redistricting initiative in California, had no chance of passing. Claussen told me I was wrong and then proved it, managing the campaign that got it passed).

In this new memo, dated Wednesday, Claussen writes that the PPIC poll effectively assumes a turnout of 50 percent. He says the campaign's own internals on Prop 1A show it with a lead of 50 to 37 percent. That would make the measure hardly a sure thing, but that's much better than the PPIC poll, which showed it trailing, with 46 percent No and 39 percent Yes:

From the memo:

"In an off-year special election to be held in May, we believe a 50% turnout model is very
high based on historical data. It is more likely that we will see turnout of 30% or lower. 
In the heated 2005 special election (with $150 million spent), turnout was 50%.  
 
"PPIC conducted 2,004 interviews in this survey and of those, 1,525 self-reported that they
were registered to vote and then self reported their past voting record, current interest and
voting intentions on May 19. This narrowed PPIC’s likely voter pool down to 978 likely
voters. This means they used a turnout model of over 50%.  PPIC also uses a random
digit dial and then asks folks if they are registered and then attempts to build some sort of
“likely voter” crosstab.  
 
"Internal polling conducted both by the Budget Reform Now committee and others
working to pass these measures was built on roughly a 30% turnout model using
registered voter lists supplemented by calls to cell phone numbers.  The numbers in these
surveys tell a different story. For example, in our polling we asked nearly the exact same
question on Proposition 1A and found 50% support for the measure to 37% oppose.  
PPIC, on the other hand, found 39% support for the measure to 46% oppose."

What does "nearly the exact same question" mean? According to the campaign, its internals did not include two small changes made to the ballot label by the judge. (Both polls read the ballot label).It is worth noting that the judge's changes, whille minor, removed two words that gave the label a more positive cast. The word "reforms" was struck from the phrase, "reforms the budget process."

I sent the memo to Mark Baldassare of PPIC, who replied in part:

"Thanks, our poll does not assume a specific turnout for the
election. In fact, we never make predictions about the size and
composition of the electorate before an election. We define likely
voters through answers to a series of questions about voter
registration, intentions to vote, past voting, interest in politics, and
attention to election news. Rather than focus on the size of our sample
as it relates to the percent of registered voters, let's consider the
partisan makeup among our likely voters -- Democrat 45%; Republicans
36%; dts/others=19%. Would a low turnout produce a more Republican or a
more Democratic electorate? This will be important, since Republicans
are  opposed to 1A to 1E will Democrats favor each of them...."