Re: correlation _ Pearson

On 2/7/2006 2:02 AM, Thierry wrote:

I want to work with multiple regression. So, first step it is to see if my variables are correlated : I have one dependant variable (Y) and five independant variables (X1, X2, X3, X4 and X5). X1 and X2 are continuous variables but X3, X4 and X5 are categorical variables. So, with SPSS I can transform these categ. var into numerical and after do a test of corellation to see the values of Pearson coef.
can I do this ?

Yes, you can.

Is it a good idea? I don't think so. Turning categorical variables into numerical variables doesn't change the fact that they are categorical. It simply provides a trick so that regression software/algorithms can be used to handle what is really an Analysis of Covariance (ANCOVA).

I am not sure what purpose it would serve to compute a Pearson correlation on a categorical variable with a continuous variable. Perhaps you could explain further what you want to do.

I do see a purpose, for example, in determining if the mean of level A of X3 is different than the mean of level B of X3. If so, ANCOVA or the regression equivalent will provide a statistical test (under the usual assumptions).

--
Paige Miller
pmiller5@xxxxxxxxxxxxxxxx

It's nothing until I call it -- Bill Klem, NL Umpire
If you get the choice to sit it out or dance,
I hope you dance -- Lee Ann Womack
.

Relevant Pages

  • Re: Principal Component Analysis
    ... If you just have numerical and binary data, Pearson correlation will not be ... If your categorical variables are ordered-categorical, ... then place these in a single matrix and analyze that matrix by PCA. ... comparison of my variables and then performing hierarchical clustering ...
    (sci.stat.consult)
  • Re: Principal Component Analysis
    ...  If your categorical variables are ordered-categorical, ...     a. ... then place these in a single matrix and analyze that matrix by PCA. ... redundancy among your variables besides correlation coefficients. ...
    (sci.stat.consult)
  • Re: correlation _ Pearson
    ... var is correlated with the others indep. ... Turning categorical variables into ... I am not sure what purpose it would serve to compute a Pearson correlation ... I hope you dance -- Lee Ann Womack ...
    (sci.stat.edu)
  • Re: anovan not estimable interactions
    ... interactions from the "terms" variable before my script runs the anovan. ... Is it possible i encounter these sorts of trouble because of correlation ... between my categorical variables? ... if I were doing this I'd just use anovan with ss type 1 to do this. ...
    (comp.soft-sys.matlab)