I mean, you're not wrong. But think about how ridiculous that is that coaches would give up on a strategy that would improve their expected win percentage by a substantial amount just because there is a slight chance they would get criticized if it goes wrong.
Actually, though, while I think "I don't want to get criticized" explains a lot of overly conservative strategies coaches employ, I'm not sure that's what's driving this scenario. Like I said, it doesn't appear they even consider going for two. Here's what I think it is: When teams are down by a "round" number (14, 10, 21) there is a natural human inclination to want to get it back to a tie game, even though tying it up only gets your expected win percentage up to 50%.
I'll give you an example (can't remember specifically when this happened, but I definitely remember seeing it within the past couple years): Team is down 10 in the 4th quarter, facing 4th and goal. Leaving aside the question of their likelihood of converting the fourth down, a lot of teams will kick the FG under the logic that "We're going to need a TD and a FG, so might as well get the FG now." The problem with this thinking is that by kicking the FG, you're passing up the opportunity to score a TD, which will keep alive the possibility that you could score 
another TD on your next drive and win the game in regulation.
Not sure what causes this blind spot, but for some reason, our brains just aren't wired to think that way. Perhaps it's related to 
loss aversion. Once you're losing, your sole goal is to no longer be in that position. So you focus on getting out of that hole to the exclusion of all else, even if that ultimately impacts your chance of winning.