In reality, compelling political science evidence suggests that people do not decide for themselves, and that in the absence of concentrated outside interests, elections are essentially random functions based on arbitrary group identity. We're now pretty close to sure that voters around the world don't generally individually map candidates to issue profiles (it's likely that the influence goes the other way), and that the electorate in the large doesn't aggregate preferences to a median acceptable policy. In that landscape, the one thing that does move outcomes is large amounts of political spending.

(This is the core argument of Democracy for Realists).