Friday, January 13, 2006

Jacobs vs Rumler preference poll results in Rumler landslide

Yes, it's just a goofy web poll, but when readers were asked the question, "If the 36th District state senate primary were today, who would you prefer?" the result was both surprising (at least to me) and not even close.

Rumler won by a landslide.

click for larger image

The final true totals with "overvotes" eliminated are as follows:

Rumler - 81 votes or 81.82%

Jacobs - 16 votes or 16.16%

Other - 1 vote or 1.01%

No preference, too soon to tell - 1 vote or 1.01%


All the number crunching and hours of compiling, sorting, and eliminating duplicate votes has been completed, the numbers and votes checked, rechecked, and checked again, and all the numbers balance perfectly.

The poll was put up Monday the 9th at 10:00 a.m., and was taken down Friday the 13th at 2:46 a.m., thus running about 4 days.

The polling service reported the IP addresses of voters and by checking them for duplicates, I was able to eliminate those who voted more than once.

Duplicate vote totals are as follows:

Rumler - 20 "bad" or overvotes from 12 voters
Jacobs - 7 "bad" or overvotes from 3 voters

for a total 27 "bad" votes.

Adding the total "clean" votes to the "bad" votes equals 126. Adding in the 2 votes for each (4 total) which were cast before tracking began, brings the total up to 130, matching the total votes reported by the polling service.

When posting the poll, I had assumed that Jacobs would manage a fairly convincing victory, so the results are surprising.

A few remarks may be in order as far as what this poll means.

The bottom line is that this poll reelects exactly what it is, namely, it reflects the preferences only of those who visit this particular site and who vote in the poll. Also, there was no way to limit voting to only those who both live in the 36th district and are registered Democrats. Therefore, of course, it is far from an accurate reflection of the voters in the district in any scientifically accurate sense.

I'm not sure how many people actually voted in the last Democratic primary for the 36th district (when was that? in '42?) but 126 individuals is a very tiny sample compared to the number of people who can be expected to show up to vote in the primary.

That said however, it is routine for polls to extrapolate poll results to the entire country based on a similarly tiny fraction, actually even smaller. They sometimes project voter preference for the entire country based on only a few hundred respondents, a far smaller proportion of the population than the respondents to this poll vs. the expected voters in the primary.

A few things might be surmised from the results.

First, the Rumler camp appears to have a superior capability to get out the troops to participate. They also might be seen to have a broader network of supporters who are online and who communicate by e-mail, etc.

Secondly, it might be suspected that there are more Rumler supporters out there than conventional wisdom might dictate. Of course, there's no way of knowing, but the results were so lop-sided that it can't be ignored completely. (though some will try very hard to do just that, I'm sure.)

What do you think of the poll results? Are you as surprised as I am? What does it mean, if anything?



Comments have been wild and whacky.


More data for those so inclined:

The vote count reported by the poll (which included repeat voters) is as follows:

Rumler: 103 or 79%
Jacobs: 25 or 19%
Other: 1 or 1%
No preference, too soon to tell: 1 or 1%

The total of votes cast was 130 exactly.

As is evident, the count including overvotes varies by less than 3 percentage points from the adjusted count with overvotes removed.


click for larger image

Two votes each for both Jacobs and Rumler had been cast before I began tracking votes to eliminate repeat voting.

I counted one vote for those who voted more than once, and did not count the duplicates.

The total of "clean" votes were 99 (which is why the percentages almost match the vote count.)

5 comments:

Anonymous,  2:09 PM  

Lotta work for a poll that doesn't mean much.

The Inside Dope 8:28 PM  

Ha... you're right. If I'd have known the time and effort required to list all the IP numbers and sort them and weed out dupes, I don't think I would have done it.

I was just excited by the idea of having a largely "game-proof" poll which at least would reflect as straight a vote as possible.


Even though it's far from scientific, I still think it's pretty surprising that Rumler could pull in such a huge majority. Surely Jacobs has plenty of connections and could have easily roused the troops.

The suggestion that they just didn't care is shot down by the fact that they're absolutly hysterical in comments trying every way imaginable and then some to attack the poll.

They obviously think enough of it to waste a whole day preventing anyone from discussing the thing in a rational way.

Anonymous,  11:18 AM  

Any poll that allows people to select themselves - to volunteer to be interviewed is ripe with bias.

The DOPE's poll consist mainly of Rubler's supporters that are intensely interested in the subject, and it highly unrepresentative of likely Democrat primary voters in the 36th District. As in all nonprobablity samples, one cannot generalize the results of volunteer samples to whole polulations. Since the Dope's procedulral errors do not occure randomly, they are termed sampling bias.

In his book, Polling and the Public, Herbert Asher warns consumers about manipulation - and hence to biased poll results- inherent in sampling techniques, interviewing procedures and the existence of "non-attitudes."

I have tried and tired to explain these simple facts to the DOPE, but he has taken to censoring my comments as he thinks they favor Senator Mike Jacobs.

I hope that you do not have a personal agenda, and will print my comments, as they are highly relevant when viewing this so called "poll"!



Anyone on the ground can tell you that Senator Mike Jacobs is heading toward victory!

The Inside Dope 7:33 PM  

Though I could, and probably should, delete it, I'll let the above stand as a preview of the dozens of comments which were left on The Inside Dope in response to this little online poll.

And the above is an example of them at their best behavior, trust me. That is as close to rational as they get.

The above poster, be they Mike or whoever, has plastered the comments over at The Inside Dope with a veritable frenzy of barely coherant comments.
I encourage all to check them out, but be warned, they're rather disturbing.

The comments are far more telling and of interest than the poll ever could be.

I shouldn't have responded to them, so let me apologize up front for allowing myself to get sucked into the insanity.

The Inside Dope 3:54 AM  

I must add that, as anyone can see, I clearly explained that this poll is not scientific nor can it be said to reflect voter preference at large.

But of course any person with any sense realizes that. Apparently Jacobs supporters don't, as they've gone nuts trying to restate that very fact, despite my stating it about 10 times in the post and comments.

They're a whacky bunch, they are.

  © Blogger template The Professional Template by Ourblogtemplates.com 2008

Back to TOP