Maybe some one can help me out on this. I am sure there is a simple answer.
I have data on 2 categorical variables which are matched. Var2 is an
estimate of the popluation parameter Var1. I am trying to measure the
variation of one variable from another and compare across categories. In
some categories (like B below) var1 has a constant value and the
coefficient of variation (standard deviation/mean) works well where the
mean and standard deviation are calculated for var2. In categories such as
Cat A, where Var1 is not a constant value I have calculated the correlation
between Var1 and Var2 but this has not worked so well. For categories with
a small number of observations on Var1 and Var2 there is a lot of sampling
variation in the coefficient. Can anyone think of a measure I could use for
both types of categories.
Cat A
Var1 Var2
493 434
461 473
409 385
Cat B
Var1 Var2
10 0
10 20
10 5
Thanks
Glenn Treacy
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|