#48#
As I mentioned last week, I was in Seattle on personal business and missed covering some of the trial in detail. I stayed over Tuesday in order to watch the Mariners at Safeco Wednesday, and ended up spending two nights at the home of a gracious friend. Unfortunately, I did not bring along a sizeable wad sufficient to entertain myself in the Epicurean style. Having seen many of the pricier offerings (like the Experience Music Project, one of the all time best museums but a shocker at $20), and wanting to see more of the non-tourist Seattle, I still felt guilty about doing nothing for two days while the trial was wrapping up and traffic was peaking.
So Wednesday morning I woke up and decided to mix the two. I sat listening to the morning session of the Whoopass in Wenatchee, composing questions for an intercept survey. I narrowed the substantive questions down to three, and added four demographic questions (gender, voting status, age, education). And with that, I said goodbye to my rather bewildered host and headed out.
Let me get a general disclaimer out of the way first, and at the end I'll be a little more detailed with the caveats. I would call this a mezzo-scientific survey, and I've also made up that phrase just now. For those not familiar, I have a survey research background, having spent 10 years working for a university polling research house. I spent much of that time doing data analysis and application development, but I also picked up quite a bit about polling science and constructed some of my own questions that were fielded.
So I'm fairly confident that I know what the knowns and unknowns are, and I'm smart enough to know that this is a survey of limited predictive power. Where possible and manageable, I followed scientific principles that would guide journal-quality research. For instance, I did not vary my question phrasing except to rotate the order of some answers. I did, however, use varying and on-the-fly leaner questions, trying to get an in-category response. It should also be said upfront that collecting by myself--trying to enjoy the city at the same time--I knew I was not going to harvest enough responses to make bold claims to representativeness. Bold, unsupported claims are for Dale Forman.
It was clear that unless I left Seattle proper, I would be able to say almost nothing at all, since downtown King is a solid wall of Gregoire voters. (On the other hand, polling during the late lunch and afternoon hours exposed me to quite a few people in town on business, not necessarily resident--and I also got several from the parking lot of a pre-game beerstand, which certainly drew from the whole metro area and beyond.) So to balance things out, it was recommended that I visit Bellevue, an Eastern suburb across Lake Washington and out by Redmond and Microsoft. In retrospect, that seems to have been an excellent choice.
So, to the results. One thing that occured to me only on the way home was that I could have devised a quick coding system and tallied each answer on a separate line, so that I could cross-analyze answers (eg, Males who said Gregoire vs Females). In order to make the interview as painless as possible, I opted for a faster tick-mark tally system while receiving responses. I did however keep separate records for the two days and thus can break down the data by geography. There are somewhat more downtown records than Bellevue records, which I'll explain later.
The first question, below, came after the contact request: "Hello (or excuse me)--could I ask you two or three (brief) questions about the Governor's Election (Contest) Trial in Chelan County (this week?)" If I got a refusal here, or the person could not answer in English, or they begged off for not being locals, I moved on. (Towards the end of the day I started to ask of refusals "are you sure?", which is really dumb and of course changed no one's mind.) Of those who didn't refuse, I moved on to Question One:
Question/Response
|
Survey Location
|
|
|
How closely would you say you are following the trial?
|
Downtown
|
Bellevue
|
TOTAL
|
Very closely
|
3 (6%)
|
2 (7%)
|
5 (6%)
|
Somewhat closely
|
13 (26%)
|
15 (52%)
|
28 (35%)
|
Not very closely
|
12 (24%)
|
7 (24%)
|
19 (24%)
|
Not at all
|
22 (44%)
|
5 (17%)
|
27 (34%)
|
Refused
|
16
|
17
|
33
|
Follow Trial
|
28
|
24
|
52
|
Non-Refusals
|
50 (100%)
|
29 (100%)
|
79 (100%)
|
TOTAL
|
66
|
46
|
112
|
Notice the three total lines. The "follow trial" number represents those who had followed the trial at least a little bit. For those answering "not at all," I took their demographic info and bade them well. The "non-refusals" number adds those folks in, and the "total" adds in the outright refusals, for a total number of contacts per day. The percentages in the cells are based on everyone who answered the question. The Bellevue respondents were definitely following the trial more, which I think stands to reason. I had only five people all day say they weren't following it at least a little bit. By contrast, you can see that every other person downtown cared not a whit about the trial. My guess is that some of those folks are Gregoire voters and think the trial is a bunch of crap, but others are simply so disassociated that they simply aren't paying attention.
It was the base of people who were paying at least a little attention who got the next two questions. Question Two was designed to get at a visceral feeling for the trial, whatever the reason:
There was a marked difference in opinion about the trial, depending on where I asked. It was not for a few hours downtown until I found someone who thought the trial had been "good for Washington." On the other hand, in Bellevue over 60% of the group thought it was a good thing, and only a quarter figured it was bad. Again, I didn't probe the source of their feeling; some Gregoire backers may feel the trial will be apt humiliation and prevent this kind of thing in the future. However, my editorial impression is that the Bellevue group predominantly found it a good thing because it was exposing the process to public view. And I'll speculate even further, and report that it seemed to be so, because they did not what like they saw exposed.
By contrast, the most-cited comment when people downtown selected 'bad' or 'very bad' was that it was a waste of time and effort. Here again, however, that sentiment was split between a very specific frowning upon the Republicans and a more general "government is all a boondoggle" feeling. This would be as good a time as any, to disclose that I did not endeavor to skip over those who appeared homeless or temporarily indisposed. Most of them were more polite to me than some of the more privileged, I can say.
Question Three was the million dollar query:
18 (64%)
9 (38%)
27 (52%)
2 (7%)
4 (17%)
6 (12%)
6 (21%)
9 (38%)
15 (29%)
2 (7%)
2 (7%)
4 (8%)
28
24
52
As you can see, 52% of the total number of respondents wanted to keep Christine Gregoire in office. Even among the Bellevue group, keeping her got as many responses as setting the election aside. So the headline is obviously that a majority of Washingtonians would prefer the status quo, but to me the most interesting thing is that Rossi gets just 12% of the tally. I talked to some definite suits downtown, and some fairly firm Gregoire spurners in Bellevue--but very few people actually wanted to see Rossi made governor. What does that indicate? As far as we can surmise, to believe that Rossi should be governor, you have to believe that evidence indicates Gregoire didn't deserve to be in office--or that she's so filthy that even without evidence she should be gone. Whether in the central core of Seattle or the toney shopping boutiques of Bellevue, the strength of the Sharkanskanite movement seems awfully fringy.
It should be noted, however, that over 4 in 10 respondents think Gregoire should not be governor--at least until some kind of do-over. (When respondents said they wanted a revote, I told them one was not possible unless this result was first set aside. When I re-asked the question, not surprisingly each of those chose "set aside" as their response.) That's really quite a lot of doubters, and that's from a population entirely within the Seattle Metro area (at least for the day). Findings like this reinforce the idea that the election contest is less than half over, and the way the result affects the mood of the electorate will be the other half.
Below are the demographic responses, collected for anyone who answered at least one question:
No surprises here, except maybe that the downtowners were no less educated than the Bellevueites--scarily equal, in fact. Some other notes: my sampling technique was what I'd call assisted-random. Downtown it was slightly more difficult to get a full response from women; they were harder to pin down. In order to remain as friendly an unobtrusive as possible, I sought out people mostly at rest--waiting for a bus, on a park bench, having a coffee, etc. If they were walking at anything more than a ramble, I let them pass. And I think because it was harder to find a woman who was not generally in motion, I had to work harder to keep a reasonable balance in my tallies. So at times I ignored men and sought out women at random. At Bellevue Square and the Bellevue Downtown Park, I had to do the opposite, and oversample for men. I did not alternate, however, and for the most part the respondent was truly random.
I did avoid people who I heard speaking a foreign language, but did ask some who turned out to be unable to respond so I could understand. Given the fairly caucasian nature of the Pacific Northwest, I also tended to oversample a bit for ethnic minorities, which was a significant cause of whatever language problems I had. I don't think this skews the results to any great extent, but I had completed my first Bellevue interview before I realized I was still using the downtown response sheet. There is thus one Bellevueite mixed in with the downtowners.
I feel comfortable enough to refer to "Washingtonians" in the headline, because 75% of the respondents were registered in Washington. It's theoretically possible that all of the other 25% were not in fact from Washington, but it was more likely for people like that to say "I'm not from here." Most of the non-registereds were those who begged out in that fashion. The age boundary was fairly arbitrary, but nicely split. I did not really attempt to balance out the ages, although I did seek out the very old and the very young (making sure they were over 18) once or twice, however.
When downtown, I intercepted people in a rectangle between Jackson and 4th, and Pike and 1st. I also worked my way down Pioneer Square, and then towards Safeco Field, finishing as I said with several interviews at the Pyramid Beer Tent-only two of which were conducted after having drunk any beer, and that was because they showed great interest in talking about the trial and what I was doing. Those two were at the tent together, and they were one of two pairs for whom I surveyed both people. The other pair were the final two interviews. Otherwise I avoided doing more than one person in a group, to keep it more random. The pair I interviewed at the beer tent were extra responses I took after making my quota of 25, which is why I have 28 (27 plus the Bellevue woman included with the downtowners).
Some caveats on interpreting the data: the counts of responses are low, about as low as they can possibly be and have any meaning at all. This is the primary reason I call the study mezzo-scientific. I wouldn't stake my job on these numbers as an accurate prediction of opinion on the trial--but I don't necessarily think they're wrong, either. There were definite patterns, and while the sample is small, I believe I did a pretty good job making it broadly representative.
We'll see what Judge Bridges has in store for us tomorrow morning. I've given my prediction; so has most everyone else. After the verdict, perhaps a media outlet will hire a pollster and validate my findings...or not. But from personal experience, I can now tell you that most folks just want this thing to be over. I think they'll be disappointed, unfortunately.
PS--yes, I know I am no HTML Tables magician. Freeware!
--TJ
Interesting exercise. The numbers are of course far from scientific, but they give a gauge, and that's one that shows the Dems will have a lot of mending to do after this election/case, even if they win the case.
Posted by: Daniel K | June 06, 2005 at 00:01
Your poll was did not use a representative sample of King County let alone the State. Thus, it is meaningless.
A scientifically conducted poll says that the only people that think Gregiore won the election are hard core Democrats. Except for the 6 percent who had not formed an opinion, everyone else in the State thought Rossi had won.
And as all Dems know, Gregiore will be gone in a flash in a re-vote.
Posted by: JC Bob | June 06, 2005 at 06:33
JC - TJ is quite up front about the limitations of the his poll, but the one I believe you are referring to was scientifically designed to get the result you report. I think I would prefer a limited poll that is well designed, to a hack poll.
In any case, the Republicans have been rather up front recently that this whole trial is a cynical PR exercise in which they have hyped up a normal (or better then normal) amount of election errors and generally dragged our democracy through the mud for political gain. If you want to talk about improving our elections then fine, but to abuse the system is not.
Posted by: Nindid | June 06, 2005 at 07:09
JC Bob--I didn't ask anyone who won the election. I asked their preference for the outcome of the trial.
Posted by: Torrid | June 06, 2005 at 07:27
Note the much lower percentage of registered voters downtown. That's disturbing to me.
Posted by: Ben Schiendelman | June 06, 2005 at 07:28
Ben, I wouldn't say it's MUCH lower (83 vs 70), especially with these response totals. And as I noted, a fair number of the non-registereds appeared to be tourists downtown. They were invariably NOT part of the substantive sample of 28 who had followed the trial at least a little.
Posted by: Torrid | June 06, 2005 at 07:32