You're referencing poll data and using that to project human behaviour across demographics. Not sure that's good science. Secondly you don't factor in that often capital punishment has majority/plurality support in most nations.