Practising with a Stats Question

Sent by Laine, after her session at the Gerstein Librarians Meeting on March 23:

"When I got back to the office yesterday after the Gerstein event, there was a perfect question waiting for me:

The user wanted annual smoking rates for Ontario, from about 2000 to the latest. We checked CANSIM, but each time series from the Canadian community health survey seemed to consist of only one year per series, and it would have required 'gluing' together a series of numbers from disparate CANSIM tables.

So we went to the Canadian tobacco use monitoring survey. Selecting (in
SDA) each annual person file, and doing a cross-tabulation of the smoking status question (Q10 or Q010 in most years) and province, and noting down the % of daily smokers, occasional smokers, and non-smokers.

I encourage you to try the same thing, in SDA, and in to see the differences.

One major difference: in SDA the resulting table is by default weighted to the population, while in Nesstar (ie ) you need to add the weight variable. If you do not use the weight, you can only talk about the prevalence of smoking _in the sample_, while if you do use the weight, you are generating estimates of and can talk about the prevalence of smoking _in the Ontario population_ 15 years of age and over."