Research has suggested that the prevailing economic conditions are a strong predictor of the incumbent's reelection chances.
I don't think that people are optimizers when it comes to elections. That is, people don't immediately begin to evaluate the various choices available and to compare and decide on the best candidate(s) or party to vote. That's cognitively expensive. Elections are not about choosing the best person(s).
Instead, I think the very first question people ask themselves is, 'Is everything good?" If nothing is really wrong, then most people would consider that acceptable and vote for the incumbent. It doesn't matter if the other candidates have a possibility of being superior; if it's not broken, why fix it? That's the power of the status quo. It isn't exactly wrong; cognitive costs and information costs do exist, even if it trivializes elections.
That's why challengers face an uphill battle; something needs to be wrong for them to have a good chance. Only when people have doubts about the incumbent does the second question come in: 'Which of these choices is better?' As a gatekeeping rule to conserve cognitive resources, this seems to be a valid strategy for making decisions, though it is somewhat reactionary.
Of course, there are still many who rely exclusively on the first judgment. If things are good, they vote one way; if not, they vote the other. It suffices to say that such a simplistic, binary thought process fails to achieve much. Some of us only have the capacity to know that things are wrong, but not why, and not how to fix it.