And then we add in all the variables around that MA training: focus, calendar time, average time per week, intensity, age started, experience of the instructor, etc. Given the number of assaults/attempted assaults that occur (and ignoring how many of them never involve a police officer, at all), there simply wouldn't be a large enough "population" in the study to control all those variables. It would have to come down to something very simple - has training (more than 6 months) or not.To the OP, the issue with finding more than general empirical evidence, such as the psychological dynamics of the offender and victim, the pre-attack cues, general tactics of a street robbery etc. is because there are simply way to many variables to consider in terms of the attack itself. Environment, training, health, age, even the clothing worn by attacker and defender matter. Now would it be possible to comb through a butt ton of police reports of a specific "typical" city and look at all of that but there are two problems with that. First, it would cost money and the chances of getting a grant to pay for skilled researchers to study "the efficacy of martial arts based self defense training in street attacks" is slim to none. Second I don't know of any officer who actually asked a person who was a victim "do you have self-defense, martial arts training" so it likely wouldn't be in a report unless the victim volunteered the information.