Eugene Volokh thinks that ballots should be accessible to even below average intelligence voters. That's fine, as far as it goes. I happen to agree that if you're a little slow, you shouldn't basically have your vote error rate so high that members of your IQ level are basically voting white noise. How far down the intelligence scale does that go? You've got to test ballots, like you calibrate standardized tests. So what's the appropriate minimum IQ that the ballot should be tested for?
The nasty thing is that you can't really answer because the entire concept is incredibly politically incorrect. There's some real life history behind being gunshy about such questions because lots of states used to have literacy tests and those tests were applied discriminatorily against racial minorities as a disenfranchisement tool, mostly against Republicans. Those sorts of literacy tests have been banned decades ago, and rightly so. But I wouldn't be surprised if the fear of being associated with such a base practice out of America's darker past has stopped election board officials from doing the easy thing, simply picking an intelligence standard, let's say an IQ of 75, finding a bunch of people with an IQ of 75 and asking them to fill out the ballot. If the error rate is below a certain amount, the ballot design passes and is used for the election.
At that point, "partisan ballot design" charges go out the window. If you tested the ballot, it fell within the error rate specs, you've got yourself a nice, scientifically based safe harbor to protect yourself from litigation. At a certain point of low intelligence, any ballot is going to be too complicated to fill out. You have to draw the line somewhere. Why not set a standard, test the things before they go into use, and where you still get error rates higher than normal, you can at least eliminate poor ballot design as a culprit.