Defects in Advertiser Poll on Akaka-Case race
HonAd: Poll shows Akaka has edge in Senate race
The article relies upon a poll conducted by Ward Research, a Honolulu-based company. In the past, Ward has been viewed as one of the more credible polling firms in Hawaii, but the methodology they used in this poll has caused me to re-think that assumption. For a complete sense of the problems, you would have to read their “poll” on the Second Congressional District race, which was really more of “name recognition” survey.
Here: HonAd: Second CD Congressional Poll
On the Republican side, for example, Bob Hogue, a former television sportscaster, has high name recognition and has an affable public image. So he scored high. But few people who monitor politics in the State legislature would give him very high marks as either a lawmaker or as a politician. He is generally regarded as a bit of a “dim bulb.”
His Republican opponent, Quentin Kawananakoa, on the other hand, is also very likeable, but is bright, articulate, polished without being “slick.” Because he has not been in the public eye, he has pretty low name recognition. There is no way that the poll offers a sense of how that race would turn out. Even as a “snapshot” of current opinion, it measures name recognition, not an intention to vote for the candidate. Of course, the sampling of likely GOP Second CD voters was very small (90-something) and the margin of error was recognized by the pollsters as very high.
The sampling for the Akaka-Case poll was probably adequate, except that they allowed voters to self-identify as “likely Democratic primary voters” in a way that makes the whole project suspect. I have not seen the actual questions, but my hunch is that early in the poll, they asked respondents if they were likely to take a Republican or a Democratic ballot in the September primary. 30% (this is from my memory) said that they would be voting in the GOP primary, so they were not asked about the Akaka-Case race.
In many states, it would be reasonable to assume that such a response would be roughly reflective of what is likely to happen in the primary. But Hawaii has a “semi-open” primary. On Election Day, voters are presented with the ballots of all political parties. They can choose in the privacy of the voting booth which primary they wish to vote in. If the polling question was asked along the lines of my hunch, the resulting data would tend to greatly overrepresent likely GOP voters. Many Republicans, asked such a question months before the primary, would reflexively, and proudly, reply that they will be taking a Republican ballot. But in truth, the percentage will be much smaller on Election Day.
This is not because GOP voters are intentionally misleading the pollster. It is because they have not had to fully process their thoughts prior to being asked the question. The GOP primary rarely has any interesting or meaningful contests. In the majority of races, the Democrat is going to win in the general, so serious and competent people have little reason to interrupt their lives to run as a Republican. The Party will sometimes put up a credible candidate and most reasonable Republicans within the district will assent to that choice, so when there are primary battles on the GOP side, only one candidate is regarded as credible, while the other is a longshot eccentric.
As a result, Hawaii Republican voters tend to vote in large numbers in the Democratic primary instead. Our voting system allows this to happen. Most of these GOP voters are sincere in their choice, rather than trying to “game the system” for partisan advantage. They generally vote for the most Republican-like of the Democratic candidates. In this race, that would clearly be Ed Case.
The Ward Research poll appears to have failed to guard against the problems this might cause for their polling data. One way to minimize the effect, would be to ask questions specifically about the Akaka-Case race, causing the respondent to think through their feelings on the race, BEFORE asking them which ballot they would take. Another safeguard would be to ask them to pick which statement is more descriptive of their past voting behavior:
1) In the primary, I always vote in the Democratic primary
2) In the primary, I always vote in the Republican primary
3) In the primary, I vote in whichever primary has the most interesting races.
Or something along those lines. Only then, and with the Akaka-Case contest fresh in their minds, should they be asked if they are likely to vote in the Democratic or Republican primary.
As a result, I believe the percentage of Republican voters planning to vote for Ed Case was greatly underreported. A lot depends on whether the GOP primary battle in the Second CD between Hogue and Kawananakoa heats up enough to keep GOP voters in the GOP primary. I do not know if that will happen. But even if it does, that would only suppress some GOP crossover in half the state.
I have not checked the historic record, but I would guess that only about 20% of Hawaii’s primary voters will vote in the 2006 GOP primary, meaning that the polling sampling has excluded about 10% of the voter base and that this 10% is self-identified as Republican! My hunch is that about 80% of those GOP voters will vote for Case, meaning that support for Akaka may have been overreported by about 8% in this poll and Akaka’s alleged 11% lead is shaved down to 3%, well within the poll’s report 5.3% margin of error.
To my mind, once you rework the Ward Research data, you have to view the Akaka-Case race as too close to call.
We’ll have to work to change that!
-K
Andy Winer said,
July 21, 2006 at 6:44 am
Interesting post K. With the entry of Jerry Coffee, the GOP now has a contested Senate primary as well as a contested primary for CD2. In 2002, when the GOP had nothing more than a contested race for LG (Aiona vs. Tanonaka), approximately 30% of the primary voters pulled a Republican ballot. History suggests that this year could be similar….
If your analysis is correct, however, then Dems (and Progressives in particular) must make sure that they participate in the Dem primary to vote for Akaka. Case’s initial tv ad tells you that he is attempting to avoid any further discussion of the War (or the Patriot Act or his support for Bush tax cuts). Instead, he wants to focus on vague notions of “transition” or “building seniority”.
The interesting thing about this latter argument is that he made exactly the same point in 2002 when he declared it was necessary to run against John Mink when Patsy died. Case said it was necessary to elect him early in order to build up a few extra weeks of seniority. That way, he would have a leg up on his classmates elected in November 2002. With 20/20 hindsight, that opportunistic slogan was nothing but shibai.
Case got his extra few weeks of seniority, and he has done virtually nothing to distinguish himself since getting there. Now, less than 4 years later, he comes back to Hawaii’s voters essentially making the same argument–that we need to build up seniority by electing him while he’s “young”. Let’s not let Hawaii’s voters get fooled again.
Electing Case puts a Blue Dog Dem who doesn’t play well with others into the Senate. In essence, we have the worst of both worlds. A guy who can’t work with others to help Hawaii get the attention it needs and a guy who supports the Bush Administration on the War, the Patriot Act, on prescription drugs, and on tax policy. What could be worse?
If this poll scares you and others, then we must re-double our efforts to get the word out on this race. Case’s transition is nothing less than Bush Administration light.
I hope you will continue to draw attention to this race.
Andy
koleaula said,
July 21, 2006 at 10:19 am
Hi Andy!
Thanks for your post. I thought I was being so clever with my analysis, but I prefer your historical evidence. (I hope I get half-credit for suggesting the historical record might have some instructive data for testing my hunch, even if I didn’t bother to dig it up myself?)
Forgive my ignorance, but who else is running against Jerry Coffee in the GOP primary? Note: I won’t be critical if you forgot for a moment and were thinking that Ed was running on the GOP side. Is there really a contest that might cause GOP voters to stay on their side of the fence?
The Second CD Republican primary is, of course, a double-edged sword– the contest affects half the state and may suppress Republican crossover in the Second CD, which it MIGHT be assumed, would help the Senator. In truth, we gotta worry about the 2nd CD GOP race attracting native Hawaiian voters, eager to vote for Quentin Kawananakoa and send a prince back to Washington as sort of a reincarnation of Prince Kuhio. That would cost Senator Akaka some Hawaiian votes and work to Case’s advantage. But I assume the campaign has polled for that and you can share the results with us on this obscure little blog? (Not everybody uses technorati to locate hard to find blog posts).
What about this scenario? Donors sympathetic to Akaka, who have maxed out their contributions to the Senator, pump some money into Bill Aila’s campaign, in an effort to raise the profile of the campaign. Hawaiian voters in the Second CD would have an extra incentive to take a Democratic ballot, so they can vote for BOTH Aila and Akaka. That boost might be worth a point or so for the Senator’s campaign. And it would probably be a relatively cheap way to gain a full point.
I’m sure you are aware that there are a lot of Hawaiians who say they are supporting both Quentin and Senator Akaka. There are influential folks in the Hawaiian Civic Clubs saying this. Ooops, come primary election day, they can only take one ballot and only vote for one of the two. With Aila, they get a two-fer. GO Aila!
Back to your earlier report on 2002. That was the year Case ran against Mazie in the Democratic primary and, still yet, 30% of the voters stayed on the Republican side of the fence? My Gawd, what would it take for them to crossover?
I was arguing that the Ward Research poll was not treating the voters as intelligent voters, who would think things through and logically crossover to the interesting races on the Dem Party side. In 2002, it seems like the Lingle supporters had a newfound optimistic identification with their Republican identity and stuck with the Party self-identification, even when it was NOT rational for them to do so.
So, I was assuming that the poll did NOT accurately reflect the reasoning process of the voters. But looking at the 2002 case, it appears that the voters follow similar foolish ways of processing information, even on election day. Perhaps I am assuming that the voters are more rational than they really are?
So if this election is somewhat analogous with the 2002 election, we Akaka supporters can relax, as Akaka DOES have an 11 point lead and the Advertiser poll is roughly accurate.
OK, I’m going to the beach tomorrow….