Tuesday, October 03, 2006

Riley Poll

Mike Riley released a poll today showing a dead heat, Saxton at 39% and Kulongoski at 37%, with 20% undecided.

This is obviously good news for Ron Saxton. The poll was conducted between September 20th and 29th, which was pretty much the same time frame as three other polls - the Hibbitts and the Rasmussen showed Kulongoski with a 5% lead, and the Zogby showed him up by 2%.

Are these differences significant, or just statistical noise?

It seems logical that the difference between the Hibbitts poll and the Riley Poll is due to different methodologies in qualifying "likely voters." Hibbitts polled "2/4" voters, which means he called from a list of registered voters who have voted in two of the last four elections.

Riley used a different system. He called from a list of registered voters who have voted in at least two of three specific elections - the 2006 primary, the 2004 and 2002 general. He then went on to qualify the most likely voters by asking how likely they were to vote. Anybody who responded "somewhat unlikely" or "highly unlikely" were excluded from the sample.

On its face, this seems like a more rigorous way to poll only those people who are actually going to vote. I am not a pollster, although I have talked with Tim Hibbitts about his "2/4" methodology. He defends it as the most accurate predictor of who will vote, especially compared to the method some of the automated polls use, which is to call all registered voters and then ask them how likely they are to vote. As I recall, Hibbitts believes that there is a bias inherent in relying on the respondents to honestly and accurately tell whether or not they are going to vote.

But Riley's method is a hybrid - it uses both the voters' own history (albeit with a modification from the usual "2/4") and the respondent's answer to the "likely" question.

Is that more accurate than the simple "2/4?" I don't know. It would make intuitive sense that Riley's method would be more accurate in identifying those people who will actually vote.

I talked to Mike this morning. He said that "very few" people responded "unlikely" to the qualification question, which means that the difference in Hibbitts result and Riley's result is probably due to the difference in the "2/4" and the "modified 2/3" qualification methodology.

In any event, it seems obvious that we have a very competitive race on our hands.

The C&E reports yesterday showed that Ron Saxton has far outraised Kulongoski, but Saxton has about half as much cash on hand as does the governor. Of course fundraising is hardly finished.

Saxton has started running an ad on the illegal alien issue. It is an effective ad in that it highlights the drivers license issue - which I have to think has broad appeal even among voters for whom illegal immigration is not their biggest concern.

I do wonder if it is wise to change the subject at this stage, after the "tax" ads had worked so well to bring Ron to within a few points of Kulongoski. The Governor just handed Ron the sales tax issue on a silver platter. I'd be shouting from the mountaintops that Kulongoski wants a sales tax.

The question is whether there are still undecided voters who will vote for Ron based on the illegal immigration issue, or does Ron already have those votes locked up? The other question is whether there are undecided voters who will vote Saxton based on Kulongoski supporting a sales tax.

I don't pretend to know the answers - but I'm pretty confident that Felix Schein and the Saxton team do.


John Eyler said...

I like the Riley system I've found that in every election I can remember polling favors the Dems by about 3%.

Capitol 3 said...
This comment has been removed by a blog administrator.
Capitol 3 said...
This comment has been removed by a blog administrator.
Mike Riley said...

The Riley Report can be found at RileyResearch.com

With such a low turnout in this year's primary, we wanted to make sure we heard from those who would most likely vote.

The crosstabs (on our site) show how those who are "certain" to vote responded, versus those only "likely" to vote and the differences are significant.

The "certain" voters appear decidedly more conservative than the "likely" voters. - Mike

Capitol 3 said...
This comment has been removed by a blog administrator.
Capitol 3 said...
This comment has been removed by a blog administrator.