I'm sitting in stats right now, actually typing this up in the command window of Stata. We had a pretty good discussion in theory today. But the best part was this quote:
"Lets just stipulate that there can be issues of sample truncation where people get killed before they can get into the dataset."
My kind of research.
(Just for clarification, the study in question was Gould's look at the participants in the 1871 Paris Commune. Arrest records were used as data sources, but, of course, those who were killed were not arrested, so they were not a part of the data set. Our class kind of wandered into a long discussion about the limitations of the data, and this was an attempt to crystallize those issues and to get us beyond them.)