Curse you, bell-shaped curve
Toxic, unfortunately the performance of projectiles in tissue and ballistic gelatin is not consistent, although it is predictable. By that, I mean that you can reproduce results, but the penetration through gelatin will fall into the typical bell-shaped curve.
The best example of this is Marty Fackler's work with the 5.56 (.223) NATO ball M193 round. I'll see if I can attach the image. About 70% of the time, the bullet would produce a wound cavity with its largest size at about 4 to 10 inches of penetration. However, about 15% of the time, the cavity would be about 2-3 inches more shallow or deeper. (The chart X axis is in centimeters, not inches). This was with "bare" gelatin. If you introduce a barrier such as windshield glass, your standard deviation will likely become much greater.
There are a number of self-proclaimed wound ballistics experts populating the internet. I'm not an expert, although I do teach the occasional class. I point out that:
1) The quality of projectiles today is generally far better than it was twenty years ago, thanks to Fackler demonstrating the utility of calibrated 10% gelatin,
2) You should select a projectile that will handle your typical mission requirements (in terms of accuracy and barrier penetration). If this isn't covered by the FBI protocols, you may need to test it yourself.
3) Shot placement will always trump bullet performance.
I spend quite a bit of time with CCI/Speer engineers, and they do good testing and are usually glad to share their data.
Fackler, ML: Civilian gunshot wounds and ballistics. Emerg Med Clinics North America 16:17-28, 1998.