There
appear to be serious methodology problems with the Los Angeles Times/USC
poll released over the weekend that show Jerry Brown beating Meg
Whitman 49 to 44 percent. This is a much larger Brown lead than other
contemporary polls, such as Field that shows the race tied, and Survey
USA that gives Brown a three point lead.
The major problem seems to be in the way the L. A. Times/USC
measures likely voters and party preference. For one thing, the poll
shows Brown doing better among likely voters than the total electorate.
If we know one thing about this election it is that the turnout will
have a higher Republican tilt than the overall electorate. In the
primary, with 33 percent overall turnout, 44 percent of Republicans
went to the polls while only 32 percent of Democrats. The upshot was
that in actual votes cast, the Republican turnout was practically equal
to the Democratic turnout. This is partially explained by hot contests
on the GOP side, nevertheless it is dramatic turnaround from 2008 when
Democratic turnout far exceeded GOP turnout.
The 2008 general election is a good place to start.
According to the California Exit Poll conducted by Edison Media
Research and Mitofsky International, the 2008 electorate self
identified as 42 percent Democratic, 30 percent Republican and 28
percent independent/other. That turnout model produced a landslide
victory for President Obama in California in 2008.
The turnout model used by the L.A. Times/USC poll suggests an
even bigger Democratic landside in 2010. While the party registration
of likely voters is 44 percent Democratic and 36 percent Republican
(about right), the likely voter self identification is 43 percent
Democratic, 28 percent Republican and 28 percent independent. That is
a heavier Democratic bias than 2008 – does anyone expect the Democratic
turnout in 2010 to EXCEED 2008? Hardly likely, given the mood of the
electorate.
There are other reasons to believe the L.A. Times/USC poll has
an unrealistic Democratic bias. The favorable/unfavorable numbers for
President Obama are 56 percent favorable, 37 percent unfavorable. Those
are higher favorability numbers for Obama than other polling shows in
this state.
This poll also shows that Latinos make up 19 percent of the
electorate and 15 percent of likely voters. But in 2006, the last off
year election, Latinos accounted for only 12 percent of voters,
according to the Los Angeles Times own exit poll that year, so to
project a 15 percent share in this off year’s election seems quite high.
The L. A. Times/USC poll also shows Brown receiving a much
larger share of the Latino vote than other polls, suggesting that this
poll may have not only over sampled Latinos, but oversampled Democratic
Latinos.
Los Angeles Times polling has a somewhat checkered history.
In August 2003, a Times poll showed Lt. Gov Cruz Bustamante beating
Arnold Schwarzenegger in the governor’s recall by 13 percent. That was
ridiculous; Schwarzenegger went on to win the governor’s race in a
landslide.
Perhaps the results this November will be the easy Jerry
Brown win the L.A. Times/USC poll suggests; at this point we simply do not
know. But to suggest that the Democrats will have a greater turnout
advantage in 2010 than they enjoyed in their landslide year of 2008
does go against everything that we do know.