I have a Stats Can paper printed at home, came out a few months ago. Forget the title, so I had trouble finding it via Google, but will try to link to it later today.
In summary, StatsCan takes in a variety of factors in defining their CMA. Notable for this conversation is that density is a major consideration.
Hence, for Winnipeg's CMA, places like St. Andrews and beyond are not included.
IMO, when you're driving out of Winnipeg, a bit past the Perimeter Highway, and you've got farm fields on either side of the highway... guess what!?! you're no longer in Winnipeg's
metropolitan area!
I know some people disagree with this, because they want the Winnipeg population to appear bigger than it is. But, that is just the way it is. If you really want to make the city appear more populace than it is, you can sneakily refer to the Capital Region.
As for StatsCan data sets, if you are conducting your own study there is no reason why you cannot define your own versions of CMAs, Urban-Rural, etc. SPSS is your friend.
IMO, inside the red line is what is (roughly) reasonable to consider as Winnipeg's metro population: