Someone here at the university is trying to tell me I cannot validly
do what I am doing. There is a set of 8 behavioral symptoms and I
have the prevalences for each in two Canadian provinces. I have
observed a very close proportionality between the prevalences of the
8 symptoms in each of the two provinces, although the overall level
is a bit higher in Quebec than in Ontario. The collinearity is very
obvious, even in a paired bar chart. I have computed Pearson and
Spearman correlations between the two sets of 8 paired prevalences
and found extremely high correlation coefficients, even on the
Spearman. Is there any reason not to treat these prevalences as
elementary observations? This person seems to think there is but I
can't quite figure out why.
David Klein
|