Also, this assumes a similar demographic distribution to identified terrorists to unidentified ones. It's quite possible another class of people are more likely to be terrorists murders etc but don't show up in the data because for whatever reason they're better at hiding it.