Objectives:

     1) correcting partial correlations for attenuation

     2) partial correlations with confidence intervals

Objective 1: correcting partial correlations for attenuation

Last week, we learned how to do partial correlations. This week, we're going to learn how to correct them for attenuation. Remember, you must correct for attenuation first!

 To correct for attenuation, you must know the reliabilities of the variables. Here is the formula that you can use to correct for attenuation a correlation between variables x and y.
 

r corrected = corrected correlation between variables x and y

rxy         = original correlation between variables x and y

rxx and ryy = reliabilities for x and y, respectively

Objective 2: partial correlations with confidence intervals All right, now that you've corrected for attenuation, the next step is to run a partial correlation for the variables in objective 1. Remember to use the corrected correlation.

Let's get a little fancy here and partial out two different variables (AGE and SEXCB). Refer back to last week's lab if you do not remember how to do this. The second partial variable goes right after the first.

You should now have a partial correlation, so let's finish up by creating the confidence interval. The formula is in Ralph's notes on page 37-3 (equation 37-6). Here it is for your convenience.
 

Se = ((1- r2)/ (N-2)) * 1.96          (please note that the equation is not written in SAS code)