Thank you for the question. I'll confine myself to one example.
You'll recall that I was talking about the estimate, which I thought was on the high side, of an 80% response rate. Even at an 80% response rate, it's going to be hard to get the level of geographic detail, census tract by census tract.
My hometown newspaper, the Hamilton Spectator, did a detailed analysis by census tract of a number of variables. There's a ton of them, but I'll just choose one.
One is the number of children below the poverty rate. There are a number of census tracts in Hamilton where the child poverty rate is in excess of 60%. The highest is 68%. There are also quite a number of census tracts in which the rate is 0%. So there's this huge span across census tracts. The trouble is that with this change in data there are a couple of possibilities.
One, if the response rate isn't great enough, we might just not get the data at the census tract level. It might just not be accurate enough and we won't know whether progress is being made on this high level of child poverty in particular areas.
It also might be impossible to link it to other variables. For example, we might want to know whether these children in poverty are going to school. That might just be impossible to know at the local level.
If it is available, there will also be some doubt, of course, as to whether the change is in the method of statistical collection.
Thank you.