Thursday, September 11, 2003

previous entry | main | next entry | TrackBack (3)

California polling and California spinning

Despite Mickey Kaus' latest Schwarzenegger scoop, it looks like the Teutonic Terminator has succeeded in phase one of his campaign -- with Peter Ueberroth's withdrawal, Schwarzenegger is now the only viable Republican challenger [What about Tom Mclintock?--ed. According to Daniel Weintraub, he's irrelevant].

Now comes the latest poll from Knowledge Networks:

The new poll of 528 likely and registered voters was conducted August 29th to September 8th and shows Californians favoring the recall of the Governor by 62% to 38% (with a margin of error of +/- 4.3 percentage points). Evidencing a strong split along party lines, 91% of Republicans — compared to 42% of Democrats — favor recalling the Governor.

Among the replacement candidates, the survey found that Republican Arnold Schwarzenegger has a commanding lead over his main rival, Democratic Lieutenant Governor Cruz Bustamante, by a margin of 40% to 28%. Only 3% of Republicans indicated that they will vote for Bustamante; but 19% of Democrats said they would vote for Schwarzenegger....

The Stanford/Knowledge Networks survey is the first to ask voters to choose from the same list of 135 candidates that they will see on election day. Previous polls have restricted voters' choices to the top candidates and have allowed respondents to select "undecided" or similar options.

Click here for the full results. There is one jaw-dropping statistic that is not mentioned in the press release: among Hispanic voters, Bustamante only beats Schwarzenegger 40% to 37%. Since this is well within the poll's margin of error, so Bustamante and Schwarzenegger are in a statistical dead heat for the Hispanic vote. This would be consistent with the Field poll data as well.

Many bloggers, myself included, believed that unless Schawrzenegger started getting specific on policy proposals, he'd wither on the vine. However, since Cruz Bustamante has decided to match Schwarzenegger's vagueness, that pressure has yet to kick in.

One last point about this poll. The only reason I know about it (and apparently beat Kaus, Robert Tagorda and Daniel Weintraub to posting about it) was because someone at the White House Writers Group e-mailed the press release to me. [Does that mean the White House is getting involved?--ed. No, WHWG is a private consulting firm unaffiliated with the government. Do they have a political slant?--ed. The group was founded by Reagan-Bush speechwriters, and perusing the staff bios it's safe to say they lean to the right. By posting about this, aren't you, like, their willing slave?--ed. This is worthy of blogging because of the caliber of the people who ran the survey, and the fact that their survey method mirrors the actual voting process. That's the spin in the two media stories I found on the poll, both of which are less than two hours old]

Two lessons to draw. The first is that the White House Writers Group is smart enough to know how to get favorable information out there -- distribute it to members of the blogosphere!! Second, the Schwarzenegger canpaign may be short on specifics, but they're long on quality consultants.

UPDATE: Eugene Volokh got the same e-mail.

posted by Dan on 09.11.03 at 04:04 PM


Something to bear in mind is that KN doesn't use telephone polling to get their results; rather, they have a large, randomly-selected panel that are supplied with free WebTV boxes and service in exchange for their participation. (Sort of a half-way house between RDD and what Harris Interactive does.)

The stats I've seen show that its results compare favorably with telephone polling (RDD), and it can do lots of things that telephone polling can't (photo recognition, for example), but it's a caveat worth noting nonetheless. Especially since no poll has shown such a huge Arnold lead (in fact, most show Bustamante in the lead).

posted by: Chris Lawrence on 09.11.03 at 04:04 PM [permalink]

I've always thought that Arnold was doing better than most polling suggests for two reasons.

Firstly, most polling doesn't measure voters' intensity very well. The Arnold voter who is mad as hell at Gray Davis is much more likely to turn out to vote than the loyal Democrat who doesn't like the situation the state's in but feels duty-bound to vote "No/Bustamante". The loyal Democrat is much less likely to get up out of his chair and go vote because he doesn't really like Davis.

The other major factor working in Arnold's favor is the way polls break in elections. In most elections, the "undecided" vote breaks against the incumbent by at least a 2 to 1 margin. For this election, Davis and Bustamante can both be considered incumbents.

That's what makes this poll's methodology so interesting -- because it's web-based, there is no "undecided" vote. In my opinion, that is why Arnold's numbers are so much higher in this poll than others we've seen.

posted by: Ryan Booth on 09.11.03 at 04:04 PM [permalink]

I wouldn't put too much credence in this poll, or any other poll for that matter, on the recall. All of the methodologies pollsters usually use to make their polls mirror the voting population aren't really useful for the recall, because they don't really know who is going show up to vote, and the existence of 135 candidates breaks the usual voting patterns. No one knows whether the methodologies the pollsters have to come up with for this recall are any good or not. Since different polls are using different new and untested methodologies, that might explain why they differ from each other so much.

Also, if it's a really close race, the uniqueness of the ballots (in that there's 135 candidates and there's a chance not all the them will show up on the same page) must be considered. By California law the candidates are not listed alphabetically, but listed according to a random drawing, so most voters will have to work through all 135 names to find their candidate. Usually, surprisingly high numbers of people stop voting after the first page, so there will be some people who might write in who they want to vote for if they can't find their candidate's name on the first page. Because of the drawing, Arnold will be listed 6 spots after Bustamente on most ballots, except for 6 districts where Schwarzenegger will be near the top of the list with Bustamente at the bottom. I don't know whether it helps Bustamente or Arnold, but I'm sure some supporters of the loser will be complaining about this after the election.

posted by: Hei Lun Chan on 09.11.03 at 04:04 PM [permalink]

Not to brag, but I have been blogging pretty much from the outset that Arnold would win this election. I said he would do it in a walk and still believe that is so, particularly if McClintock drops out or wilts. The idea of "specifics" in a political campaign is a paper tiger, particularly in this one where "leadership" is more the issue than policy... and Arnold, in his weird way, promises leadership. That Cruz is not offering much specific on his own is not a surprise. He has long been a dunderhead and a political accident. Here is one of my early links on the subject. There are several...

posted by: Roger L. Simon on 09.11.03 at 04:04 PM [permalink]

"In conjunction with the Hoover Institution at Stanford University, Knowledge Networks has used its projectable Web-enabled panel to conduct a survey..."

Hoover Institute. Hmmmmm.....

posted by: Barry on 09.11.03 at 04:04 PM [permalink]

Dear Dan,

Please join with us and bloggers around the world in supporting the Iranian people against a brutal fanatic terrorist clerical regime that continues to support terror and make problems for not only peace but also for American soldiers and others who are trying to assist the Iraqi people!

Join BLOG-IRAN at and help free the Iranian people... LETS BE THE VOICE when everyone else is SILENT!

In Unity & Struggle,

posted by: Haleh on 09.11.03 at 04:04 PM [permalink]

Did you attend Williams, Mr. Drezner?

posted by: d on 09.11.03 at 04:04 PM [permalink]

Post a Comment:


Email Address:



Remember your info?