Sunday, January 23, 2011

Example #52,349 of Why Intelligent People No Longer Trust "Statistics"

A new "study" by the "Centre for Addictions Research of B.C." at the University of Victoria claims that private liquor stores are directly responsible for increased deaths associated with alcohol. Here's the official press release on the "study" and here's the actual document (h/t & thanks to langmann).

The lead "researcher" on the "study", Tim Stockwell, was interviewed on CKNW this afternoon. You can listen to the interview here, beginning at 6:00.

I was most curious why he kept on interjecting "private" before "liquor stores" so I sent him this e-mail:

On your recent appearance on CKNW, you appeared to state unequivocally that the increase in deaths due to alcohol was directly related to the fact that the increase in liquor stores in the province were private, as opposed to public.

I am most curious to find out how you proved that the same number of deaths wouldn't have occurred if all of the private liquor stores had instead been public ones. Could you direct me to the section in your study that proves this?
Incidentally, a month ago Mr. Stockwell was advocating for higher booze prices for the general public but free alcohol for homeless drunks.


To my surprise, Tim Stockwell did write me back:

Hi Robert
Thanks for your enquiry. Please find attached a copy of our paper. We found in effect for the number of stores regardless of type but then when we controlled for that the percentage of private stores in an area was highly significant as a predictor of deaths - we found the same thing in an earlier paper in relation to sales. The private stores to seem to be better at selling the stuff.
Best wishes
Here's what I then wrote back to him:

Dear Tim,

But that's not what you said on Sean Leslie's show! Or, if you'd like me to talk in strictly legal language, "that is clearly not the impression any reasonable person would have had after listening to what you said".

I am a man of science and math and facts. From my point of observation, no other field of science has been so abused in recent years as that of statistics.

A person in your person has a professional & ethical responsibility to strive for accuracy in all areas of your work, especially when it comes to informing the public. It would be most interesting to see what conclusions a random sample of 1,000 people would draw from your appearance on Sean Leslie's show, then compare them with what you've said to me here, and see what correlations or lack thereof were found. I suspect I know the answer. :-(


No comments: