I need to compute point biserial correlations for some data. However, I cannot find a procedure in any of the major stats packages that does this.
The point biserial correlation is just the Pearson correlation with one of the variables being dichotomous. A special formula exists, but its purpose is to ease the burden of those who have to do the calculations by hand.
So, on a computer, just use the Pearson correlation procedure:
In SAS, use: PROC CORR.
In SPSS, use: CORRELATIONS.
If you have further questions, send E-mail to stats@ssc.utexas.edu.