Perhaps, allow me to offer a counter argument. I claim that if the only criteria a voter uses in voting is "I like how your part of the government is working", vote the incumbent. If they don't like how its going, vote someone else. Very simple, no research, no time wasted at all. And absolutely effective.
The reason its effective is because elections in the US are very bias sensitive, the very small margins of victory for one candidate or another means that a small change in voter sympathy can change the outcome of the election. Further, by creating a selection criteria of 'like' / 'don't like' in the voter input, you set up a genetic algorithm for evolving government.
There a probably a couple of dozen 'evolution simulators' available on the web. They all work in the same way, they take a population of organisms, they exist for a while in an environment, and then they 'reproduce' with a selection bias. Over a few generations the selection bias results in a population that demonstrates the characteristic that the selection bias is targeted toward. The basic mechanism is that organisms that align with the selection bias 'win' (and thus reproduce) and those that don't align 'die.'
That algorithm doesn't care what the 'lifeform' is, and it always works. That is part of the reason it annoys folks who favor an intelligent design theory of evolution.
Elections are the 'reproduction' cycle of politicians. Their 'survival' depends on them managing the bias in their favor which they have a limited amount of power to do. If the current non-voting block of citizens jumped into the game and voted 'like'/'do not like' it would force government to 'evolve' a set of politicians that were doing a good job as defined by the people who vote.
Clearly this doesn't work when the government is rigging the elections (as Russia was accused of doing to counter unexpected voter turnout) but it certainly does in a fair election.