Poll comparison

| | Comments (1) | TrackBacks (0)

This morning the Tulsa World finally got around to publishing the poll it commissioned from SoonerPoll.com. I say "finally got around" because polling was completed on September 1, five days ago.

It seems to me that if you're in the business of news, and you have a newsworthy story, you would want to publish it as soon as possible.

It shouldn't take a pollster more than an hour to turn survey results into a top-line number -- percentages for each candidate. It shouldn't take more than a day to generate cross-tabs -- breakdowns of results by demographic groups.

It also seems to me that one week out is an odd time to take a poll, particularly your very first. A poll right after the filing period would give readers an idea of the shape of the race at its beginning -- how big is each candidate's base of support. A poll taken the day before the vote provides a fair test of the pollster's accuracy -- something that can be directly compared to the result of the election.

A poll at one week out doesn't capture any shifting that may happen as a result of the last week of campaigning, as more commercials hit the airwaves, as mailers hit mailboxes, and as voters finally begin to focus on the election. A one-week poll would only be useful if there were an earlier public poll to which it could be compared, as a measure of momentum.

It's interesting to compare it to the Tulsa World's 2006 pre-primary poll. It was taken 9-11 days, slightly more than one week, before the election, but the results weren't released until five days after the poll was complete. This year the poll was taken closer to the election, but with just as much of a gap between completion and publication.

In 2006, the sample size was 500 for each party primary. In 2009, it was cut to 300 for each, increasing the margin of error. There's a reference in the 2009 story that hints at other questions being asked and crosstabs by ideology, but the details weren't provided to the reader.

In 2006 results were provided to the tenth of a percentage point. In 2009 the results presented to the reader were rounded to the nearest whole percentage point. That's a minor point, perhaps, but it still adds to the downgrade in the quality and detail of the information being presented to the reader.

The 2006 poll showed Medlock with 18.8%. His result on election day was 34.1%. That doesn't mean the 2006 poll was wrong, just that it couldn't measure the impact of the final week of mailers and commercials.


Just noticed: While past polls were co-sponsored by KOTV, this poll appears to be solely sponsored by the Tulsa World.

A reader points out that the story on the poll did not disclose the methodology used to determine whether or not a respondent was likely to vote and did not note whether respondents were given a list of names before being asked their preference. And, for what it's worth, the reader also notes that the pollster, Oklahoma City-based Bill Shapard Jr., lists himself on his Facebook profile as a supporter of Dewey Bartlett for Mayor.

Also, Medlock campaign manager Howie Morgan sent out an e-mail reacting to the poll:

This past Tuesday our radio ads hit the air. Also on Tuesday we had our TV ads all over Fox News, CNN, ESPN, HGTV, Lifetime, as well as the local news channels. You have likely seen them over and over all weekend. In those ads Chris pledges to Tulsa voters how Chris Medlock will end the wasteful spending in City Hall and get back to the basics of fixing our streets and fighting crime. His priorities will be in favor of your family budget instead of trying to take more taxes out of your wallet by changing the Broken Arrow Expressway into a Toll Road. And boy did those ads work! Our phones have been going off the hook, our website traffic has tripled, and people are disgusted when they found out that Dewey Bartlett, Jr. is just another tax and spend RINO who wants to take his past history on the Oklahoma Turnpike Board and add one more toll road that taxes Tulsa. We love those ads on TV and radio. If you have not heard them yet, click here to watch and listen to them. (I like the funny radio ad the best.)

Unfortunately for us, the Tulsa World did their poll on this race in the days just before our ads hit the air. So unless 30,000 voters were clairvoyant in knowing about Chris and the election for Tulsa Mayor, there is no way they even knew Chris was running.

Yes, I know you know. But the average voter did not until our ads hit the airwaves.

So when the Tulsa World did their polling last week (on a Sunday? what is that?) we are pretty sure that no one in the general public knew about Chris Medlock and where he stood on the issues.

They have now joined you as an informed voter, and they now know the real story about Dewey, Jr. and his RINO status. So please take that into account when you see the Tulsa World poll today. Their numbers are a snapshot in history, but that history is tainted with $150,000 of 4 weeks of Dewey, Jr. ads and not one TV or radio ad from Chris Medlock until after the poll.

0 TrackBacks

Listed below are links to blogs that reference this entry: Poll comparison.

TrackBack URL for this entry: http://www.batesline.com/cgi-bin/mt/mt-tb.cgi/5268


Anon said:

I did notice the poll 'accuracy' quoted as something like 5.9x% (to the hundreths). A number not generally used (common +/- 4% is the norm).

Could it be the poll was designed for 'normal' and adjusted to fit mood? Downsizing the sample count and upping the error?

And, as you point out, last year's poll was approaching 50% off, for some unknown reason.

I haven't trusted a World poll in years. The timing of this one (the first, too) is odd.

Get out and vote people.

About this Entry

This page contains a single entry by Michael Bates published on September 6, 2009 1:06 PM.

OSU journalism prof to Bartlett Jr: "Sealing court records isn't supporting the public's right to know" was the previous entry in this blog.

Tulsa campaign roundup is the next entry in this blog.

Find recent content on the main index or look in the archives to find all content.



Subscribe to feed Subscribe to this blog's feed:
[What is this?]