There are two possibilities:

1) Take the sum of the cross sections of basic AO's weighted by squared participation coefficients (a popular routine in the UPS business)

2) Calculate the transition matrix element from the whole MO=Sumi(Ci*AOi) as

=Sumi(Ci*)    (el is the wavefunction of the photoelectron)

and then take the square of it.

I feel that (2) is the way to do it. What do you think?