On 29 Jun 2002 06:33:53 -0700, [EMAIL PROTECTED] (Holger Boehm) wrote: > Hi, > I�m presently working on a statistical problem in SPSS: > For two populations P1 and P2 I have calculated the coefficients of > correlation R1 and R2 between a quantity ALPHA and BETA. > In SPSS, how can I perform a statistical test to tell whether the two > coefficients significantly differ from one another?
The correlation is ordinarily "not a good thing to test" - when you can test the regression coefficient, without the (almost-always) extraneous assumptions about equal variances. Since it is the same A vs. B, you should be able to pool the two sets of data. Is the residual SS of the combined regression a lot bigger than the sum of the residuals of the separate regressions? - look for the Chow test (one name for it). I think there is more detail in my stats-FAQ. -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
