Re: Pearson's correlations and dichotomous pairs?

Posted by Art Kendall on Nov 16, 2006; 1:34pm
URL: http://spssx-discussion.165.s1.nabble.com/Pearson-s-correlations-and-dichotomous-pairs-tp1072140p1072144.html

I concur.
In addition, even version 1 (1969?) would give the correct answer, not
just pc version 13 on.

Art Kendall
Social Research Consultants



Statisticsdoc wrote:

>Stephen Brand
>www.statisticsdoc.com
>
>Robert,
>
>The phi coefficient and Pearson's R are algebraically equivalent when you
>are computing correlations between dichotomous variables.  So, yes, SPSS
>will give you phi, but not because it "switches" to a different formula, but
>because the results are the same.  Phi and many other formulae are
>computationally simpler than Pearson's R but yield the same coefficient.
>Pragmatically, you never need to worry about SPSS giving you something other
>than the Phi coefficient when you are dealing with a pair of dichotomous
>variables.
>
>HTH,
>
>Stephen Brand
>
>For personalized and professional consultation in statistics and research
>design, visit
>www.statisticsdoc.com
>
>
>-----Original Message-----
>From: SPSSX(r) Discussion [mailto:[hidden email]]On Behalf Of
>Robert
>Sent: Wednesday, November 15, 2006 5:39 PM
>To: [hidden email]
>Subject: Pearson's correlations and dichotomous pairs?
>
>
>It is my understanding that when Pearson's Correlation is used, SPSS (v13
>and later) will calculate the exact correlation regardless of whether the
>variables are continuous or dichotomous.  So in case of a pair of
>dichotomous variables, SPSS will "automatically" compute using phi . . .
>can anyone confirm if this is true or not?
>
>Many thanks,
>Robert
>
>
>
>