Spearmans Correlation Doubt

pedro90

New Member
Hello! I have some doubts about Spearmans's Correlation and I would like to ask your help!

I have 5 variables (3 independent and 2 dependent) and I want to check if each one of the independent is correlated with each one of the dependents.

Once not all the variables showed normality behaviour (i confirmed this using shapiro test;Q-Q plot and histogram analysis) I did Spearmans instead of Person's.
Lets call to the independent variables A;B and C
and to de dependent D;E (My E variable is defined as D/A).
I did not find correlation between A-D; B-D nor C-D. However I found a correlation statistically significant between B-E and C-E...

The question is, am I committing any kind of violation with the D/A division to "generate" E? Or can I trust that the result of the test is trustworthy?

(PS: A;B and C are independent!)

Thank you so much!

noetsi

Fortran must die
I was not aware that Spearman assumed a variable was independent or dependent unlike say regression. If your dependent variable is defined in part by one of your independent variables (which E is) then you have forced it to be related to the IV. That can cause a problem with the calculation of parameters in regression, but I don't know if this is true of spearman's.

What is the form of your variables (interval, ordinal, or binary)? This has significant impact on your results. If any of the variables are binary (say yes no) then spearman's is not ideal.