About an hour ago I spoke with Michael Kozdron, a probabilist in the statistics department at the University of Regina. He isn’t the U of R statistician the clerk’s office worked with on the statistical sample portion of the petition verification process.¹ But he’s looked at everything the clerk’s office has provided and he’s confident that the statistics are 100 per cent satisfactory.
We spent a very long time on the phone and he walked me through all the statistics and while I couldn’t teach a math class on any of this, I’m now confident too that at least this part of the clerk’s report is correct:
The above represents a confidence level of 95% and a margin of error of 3%. That means that the proportion of the verified answers is 87.4%, so there is a 95% confidence that the true proportion of verified (valid) signatures lies between 84.4% (16,941) and 90.4% (18,145)
And Kozdron points out that even if all the people who didn’t respond to the phone survey had turned out to be legitimate, verifiable petition signers, the petition would still not have passed the threshold.
Okay. So the math on that portion of the verification all checks out.
However, Kozdron could not comment on the criteria used to exclude signatures during the clerk’s first pass through the petition and he certainly would not comment on the decision to exclude petitioners who didn’t include a year when they wrote the date.
I will note though that I can multiply. And as the proportion of verifiable signatures is 87.4% and as 2,834 names were removed for not including the year, that would mean that with a 95% confidence level and a 3% margin of error, between 2,391 and 2,591 names could be argued to have been inappropriately excluded. And had they been included, the petition would have passed the threshold needed to force a referendum as it would have had between 19,332 and 20,732 valid signatures on it.²
¹ For the record, I called Michael because I wanted to get a second opinion and I know him. My wife’s a math prof at the U of R and Michael and his family were the first people we met in Regina. They’re great! And I knew I could trust Michael to do a thorough and honest read through of the clerk’s report. And I knew he’d be able to explain the math to me so that I’d get it.
² That paragraph is all my math. Not Michael’s. And… Yeesh. I’m all worried about attribution today… In comment fields on other posts, expert commenter Brad has gone through a lot of these numbers and come to the exact same conclusions. Look it up. And as far as I know, other people have comments on here that do as well. For the record: I don’t have time to read every comment that winds up on our blog.