Kevin Wang wrote:
Has anyone seen this paper? https://injepijournal.biomedcentral.com/articles/10.1186/s40621-018-0174-7
Quite an interesting reading!
It is! I don't have much experience with survival analysis, but this part caught my attention:
"The analyses revealed worse survival for characters who ... had not switched allegiance during the show (P < 0.001)... After adjusting for other factors, whether or not a character had switched allegiance during the show and how prominently a character featured in the show were revealed to be independent predictors of death."
Naively, I would expect that substituting almost any uncommon life event for "switched allegiance" would give a similar finding. The longer somebody lives, the more opportunity they have to switch allegiance, get married, visit King's Landing, etc. etc. Applying Bayes' theorem, it would follow that (life expectancy conditional on experiencing X event at some period during one's life) will generally be higher than unconditional LE.
Do the methods in this analysis take that sort of consideration into account?