Great comment - in hangman, choosing letters solely on the basis of data (no matter how complete the data or how compelling the study) is a highly predictable (and thus highly defeatable) strategy.
Of course, this analysis has a counterpoint. If I try to pick words strategically, I suspect that my words would begin to follow patterns...
To be sure. I recall we'd mostly stick with the two strategies I mentioned (words with most rare letters, or words in very common patterns)... then when you're settling into a pattern (and your opponent's first guesses are "x" or "q" or similar), throw in something like "teat" to throw them for a loop.
We did have one minor rule change -- definitely more than 11 line segments to hang the man. If you had a really good word, and wanted to draw out the torture, you'd start drawing in facial features, fingers...
It also adds a bit of artistic fun to the game; you don't even have to be hanging a "man" if you want to get creative; it could be a giant insect, horse, whatever.
It would still be pretty clear who was "winning" by how long the string of crossed-out letters were by each round.
Of course, this analysis has a counterpoint. If I try to pick words strategically, I suspect that my words would begin to follow patterns...