Here's an interesting thing I thought about. Let's say everything a presidential candidate does can gain them voters or lose them voters. Let's say that a candidate is listing off their policy views sequentially.
Each policy view wins them or loses them votes. Ones that win votes are great. They increase the probability that the candidate will win and be able to implement their policy views.
But what about views that lose them votes? Well, they'll lose votes, which is bad, but at least they're being honest. But more than losing votes, they'll also be decreasing the probability they'll be able to win and put their views into practice. So being more honest about some issue actually leads them to put their other, potentially more popular, views into practice. And declaring one policy view has a multiplicative effect on the implementability of the rest.
Maybe this is a game theory-esque thing or something else. But it seems to stem partly from people making decisions with incomplete information and partly from hidden, undisclosed information.