Archives

A sample text widget

Etiam pulvinar consectetur dolor sed malesuada. Ut convallis euismod dolor nec pretium. Nunc ut tristique massa.

Nam sodales mi vitae dolor ullamcorper et vulputate enim accumsan. Morbi orci magna, tincidunt vitae molestie nec, molestie at mi. Nulla nulla lorem, suscipit in posuere in, interdum non magna.

KLRvu polling methodology update

Well, we’ve had quite a few of blogposts on the KLRvu pollsters and their polling numbers for Guelph throughout the blogosphere (including yours truly), and even one by David Akin looking at them. All this publicity no doubt thrills Allan Bruinooge, the head of the polling company, to no end.

I mentioned on Friday that I’d written the polling firm asking them of the 3396 households they polled in Guelph, how many of those refused to answer, hung up, etc. I received this response in email from Mr Bruinooge, and to be fair to him, I’ll quote his reply in full. He explains that the 3396 figure is the final number of respondents out of the numbers that they called:

To your question of the 3300+ households that have taken part in the poll they are 9% of the total population sample. If you do the math, what is total population sample if we know the number of respondents are 3396 which is approx 9% of it? 9%=.09 .09/3396 = 9/339600 which equals 37,733 is the total population of the sample. Hope this helps, I do apologize for not putting the total pop in the press release instead I was using the text:

“This poll was completed in two rounds on July 27th and August 13th,sampling over 3396 random residents of the Guelph constituency representing nearly nine percent of all households with a 95% confidence level and a margin of error of 2%.”

…and that was to mean all the phone numbers with households available in the federal riding of Guelph.

“The response rate of 9% is typical of the polling methodology .”

So, according to Mr Bruinooge, they called over 37 000 Guelph households, and 3396 of them responded, for the party percentages that they eventually listed in their news release. He also claims in that last line that 9% is typical for this type of polling. I’d be rather interested to see what SurveyUSA and Rasmussen – two major polling firms in the US who are polling the Congressional and Presidential race using a similar robo-calling method as KLRvu does – claim to be standard response rates to their polls.

So, it’s an interesting new tidbit of information that I hadn’t seen listed at some of the other blogs who had mentioned this story, so I thought I’d bring it up. That doesn’t mean I’m no longer sceptical of the numbers, of course, but I do thank Mr Bruinooge for responding to my question.

3 comments to KLRvu polling methodology update

  • Allan Bruinooge said he did the poll on his own dime.

    In 2002, the average response rate was 12%:

    http://www.mria-arim.ca/COMMITTEES/ResponseArticle01.asp

  • Ti-Guy

    If you do the math, what is [the] total population sample if we know the number of respondents are 3396 which is approx 9% of it? 9%=.09 .09/3396 = 9/339600 which equals 37,733 is the total population of the sample.

    That’s a bizarre way of answering your question. Shouldn’t KLRVU have a tally of the number of calls placed? They knew the response rate, so obviously they must have that figure, which is the one you should have been provided with. All you were told was that we can figure out the number yourselves.

  • Buckets

    An important as-yet-unanswered question is who commissioned the poll.

unique visitors since the change to this site domain on Nov 12, 2008.