Hi Hiliary,
I agree with Ian but just wanted to add a couple of points. The first is that scattergrams are a great visual aid for interpreting correlations and so, along with Ian, I'd suggest that you do that as a matter of course when you conduct this kind of analysis.
The other point I wanted to make is that the only way to compare correlation coefficients (and I assume you're using Pearsons r) is to standardize them using a z transformation. The Fischer's z test will compare two correlation coefficients after you've transformed them (i.e., are they significantly different from one another).
Finally, correlational analysis always assumes that you have sampled from the entire distribution, otherwise your scores will not be normally distributed. This is another reason why comparing two correlation coefficients can be problematic.
Cheers,
Candice
|