Correlation between two sets of data

I am working on a data file looking for the correlation between two variables (numerical). My population is divided into two. There is correlation between the two variables of the two groups (group 1 has a correlation of -0.256 with p<0.000 and group 2 of -0.287 with p<0.000).<br>I want to compare the degree of correlation between the two groups, ie is the correlation between the group 1 more powerful than group 2 or is it the same?
Please tell me which test to use.
I would appreciate your help and I would like to thank you in advance.

Comments

  • I think statisticians will tell you to use the Fisher transform on your correlations and then test for differences. (I am presuming that you are tracking correlations between your groups and some variable across time). In meta-analysis, however, they often don't bother to transform the correlations but just treat them as data in their own right. Numbers, they say, don't know where they come from.

    If you simply have two correlations and wish to test whether they are significantly different I think you have to use the Fisher transform. Excel has a function to do this. A good stats text should have details of the method of comparison.

    Lance
  • I found the following using a Google search:

    homepages.gold.ac.uk/aphome/correlnotes.doc

    There were also several other urls that looked as though they could have useful material - one referred to excel:

    www.stat-help.com/spreadsheets/ Comparing%20Correlation%20Coefficients.xls

    fsweb.berry.edu/academic/education/ vbissonnette/applets/rdiff.html

    Do a bit of searching...

    Lance
  • Thank you for the reply. I have to accept that it was very useful. The links also were interesting.
Sign In or Register to comment.