Semipartial correlations for simple slopes


I'm in need of some help for calculating sr2 for simple slopes. Here is my situation:

I'm using SPSS.
I have three IVs:
x1: dichotomous and coded -1,1
x2: dichotomous and coded -1,1
x3: continuous centered variable

I have a 3 way interaction that I've calculated the simple slopes for. I've broken it down so that I'm looking at the effect of x1 at levels of x2 and x3 (1 sd above and below the mean of x3).

I need to be able to calculate the semipartial correlation for the effect of x1 when x2 = -1 an x3 is 1 SD above the mean.

I found a tutorial online that calculated semipartial correlations by first having SPSS output the residuals for each of the terms in the regression equation (in my case this is b1x1, b2x2, b3x, b4x1x2, b5x1x3, b6x2x3, b7x1x2x3) and correlating y with each of those.

Where I'm getting stuck is how to then use this method for calculating the sr2 for the simple slopes (i.e., get the sr2 for the simple effect of x1 at given levels of x2 (-1) and x3 (1 SD above the mean). My first thought was to select cases in SPSS for the levels of x2 and x3 I'm interested in, then output the residuals for x1, and then correlate y with those residuals... but I'm pretty sure that isn't right.

Any help?