On 2 Jul 2002 07:46:20 -0700, [EMAIL PROTECTED] (nothanks) wrote: > Hello all, > > I have a client who wants to compare the relationship/association > of two variables (Y1 and Y2) between 2 groups (say Gender). > They (statistically challenged) have left it to me to decide > on method (and how to measure association). > > Y1 and Y2 are ordinal, but are actually continuous variables > binned into 0, .5, 1, 2 and 3. So, I'm o.k. with treating them > as continuous.
Well, do look at the crosstabulation, if not at the plot of raw numbers. Is the association linear? Does the differential spacing (half-point and one-point) suggest that you ought to rescale each interval to 1.0 ? > > Anyways, here's what I'm thinking. > > 1) Fit a linear model Y1 = Y2 + Gender + Y2*Gender, > and see if the interaction term is signficant. > Yes, that tells about different 'slopes'. And the test on the Gender term tells if the 'intercepts' differ. > I figured the differences in slope would give me a decent comparison of the > differences in association between Y1 and Y2, between the two groups. > Keep in mind, I'm not wedded to actually comparing the Pearson correlation. > Hence, You are comparing the regressions, which is less *presumptuous* than comparing the correlations - the latter presumes that variances are equal (usually, neither interesting nor relevant.) > > 2) Use a Fisher's Z' Statistic to compare the Pearson Correlation. [ snip, rest ] <It is better to compare regressions. see above. > -- 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/ . =================================================================
