You can't calculate percentages using data from different studies and surveys, especially if you don't know the criteria or methodologies involved. The results just won't make any sense.
The data Val is quoting is likely the average from the two government sources, the NCVS and the FBI, but there's some problems with treating these two sources as absolute canon.
Among them is the fact that the FBI data only count reported rapes and even then, only the ones that fit their specific definition of violent rapes. Meaning most if not all incidents involving drugs like ketamine, GHB, Rohypnol, etc or even plain regular alcohol won't be on there. That definition only changed recently (I think this year in fact), so meaningful data from the government won't even be available until next year at the absolute earliest and realistically not for several years.
The NCVS statistics have their own problems, namely that their methods of data collection is so suspect that there's little confidence in their accuracy in the scientific community. This is enough of a problem that the government have ordered a National Statistics panel to convene and develop better methodologies and until that is complete and a new set of data is collected, the NCVS statistics can't be considered anywhere near accurate.
Now, Kilpatrick's study, the source of the 30k pregnancies estimate likely have its own problems, but since I no longer have access to institutional full text databases, I can't check on their data. You'd need to know their sample sizes, how the sample was drawn and their other methodologies to make any meaningful comment on the study's validity. But until their results are colloborated by other independent studies, I'd consider their results suspect as well.
Bottom line, the data you guys want to argue about doesn't even really exist in any meaningful sense at the moment, but ignoring that, it's still sheer idiocy to attempt mathematics on studies that were not even counting the same populations.