Fair enough point about DnD 3.5 / 4th ed... I haven't actually played either one so I can't comment. The rather amazing success of Pathfinder alone though seems to suggest that the new editions of DnD have lost the plot somehow.
Oh, and, I completely forgot to mention Traveller: very successful SF game in its day. I'm not sure, but I don't think anyone has ever quite captured that success again with an SF game.
I wonder if source books could take the TTA approach: commission concept art of ships / space wrecks / planets and then write the game material around each illustration. Do it backwards from the illustrations first.
Anyway, yes, that'd be more or less the way I'd put the data together, except that it needs to be in long-format rather than tall format if it's to be put into an analysis, and all variables would have to be listed. In an Excel spreadsheet it'd have to look something like:
GAME YEAR SUCCESS SETTING DICELESS D4
DnD 1974 Very.High Fantasy N etc
RuneQuest 1978 High Fantasy N etc
Amber 1991 High Fantasy Y etc
I could set up a shared Google doc, invite anyone who wants to contribute and then do the various statistical analyses down the track once enough data is together. I can't really contribute to the data gathering though... I just don't have the time. Complex analyses on the other hand are relatively easy for me to do (of course we might not actually find anything interesting in the data... it's hard to tell). I spent the last three years of my PhD figuring out how to elegantly analyse insanely difficult ecological datasets full of multiple auto-correlation and statistical interactions. It'd be fun to put that to use with something a but more lighthearted.
Chris