Re: Comparing different scales


I'm not sure what you want to do. One possibility is you want to
undersatnd the realtionship between them - for example to predict one
from the other. I think that calls for simple linear regression.

Maybe you want a score for each person that captures the discrepancy
between the tests. I think this might be achieved with the residuals
from the regression (maybe standardized, maybe not).

Another possibility is that you are interested in agreement between the
tests. In that case I think you need to look at the Bland and Altman
method.

Thom

.

Loading