Framer,
SME = Subject Matter Expert = Smarter than me.
So you got the gears in my noggin going...
One possible way to do this would be to pick your data set, then go through and see how many events of the total had an FE, other crewmember, flight check, or jumpseater in the cockpit. (i.e. FE/Total, OC/T, FC/T and
JS/T). Then see whether these numbers are disproportionate.
If you were, say, to pick commercial aviation accidents in the US between 1990 and 2002, we could take it one step further. Shappell and Weigmann have already performed an HFACS analysis on this data set (link to article below), so we could then perform an HFACS analysis of these subsets (or somehow get their data sets) and see whether these events had different percentages of certain errors or violations. Though we could do this with any data set you choose, they did such a large set to validate their protocol.
Soooo.... have I taken a fun cocktail party question and hosed it for life?
J.
http://01f7812.netsolhost.com/pubs/H...0Accidents.pdf),