Comparing results of two different tests measuring correlational relationship?

#1
Hear this.
I have three scales Scale A, B, and C.
I want to compare the strength of relationship between Scale A and B, and between Scale B and C.
Details:
Scale A: Actually, this is not a scale. This is just 1 question with 4 nominal answers. What type of family do you have? Authoritarian, democratic, inconsistent, undefinable.
Scale B: RWA scale. Scores between 0-160. Lower the better. No decimals.
Scale C: CTQ Scale. Scores between 0 100. Lower the better. No decimals.
Can this be done; is this legitimate comparing coefficients from two different types of tests? Are they even two different tests or are they just basic correlation spearmen etc.?
 

j58

Active Member
#2
For A vs B, I think I would regress B (DV) on A (coded as dummy variables (IVs)). This would provide R^2, the proportion of the variation in B explained by A.
For B vs C, you can compute the Pearson correlation. If comparability with the above is desired, square it.
 
Last edited:
#3
Thanks for the response. So I can use liner regression for the first one and Spearman correlation for the second one. And when I square the Spearman's coefficient, I'll be able to compare the strength of relationship between the two. But.. Spearman correlation is noted by
(rho).
 

j58

Active Member
#4
The square of the sample Pearson correlation coefficient, r, is equal to the proportion of the variability in one of variable explained by the other. If you don't believe this, calculate the correlation between B and C, then regress B on C, and compare the two r^2 values.
 
Last edited:
#8
Hi its me on this particular message again.
Can you elaborate on how to find the r^2 on the relationship between A and B. Coding A as a dummy variable, how?
Many Thanks.
 

j58

Active Member
#9
I assume you're using statistical software to do your analysis. If you regress B on A (coded as dummy variables), your software should report R^2.