The codon based Z-test for selection can be done in two places. First, you can use the Tests | Codon Based tests of selection | Z-test (large sample) option to find the probability that the null hypothesis will be rejected, in addition to the actual value of the Z-statistic. Alternatively, if you want to know the difference between s and n (synonymous and nonsynonymous substitutions and their variance, you can go to the Distances | Pairwise menu option and in the distance computation dialog, select an appropriate method (e.g., Nei-Gojobori method) and then choose s-n (or n-s depending on your need) from the Substitutions to include menu. Also, you can choose to compute standard error.