When trying to argue about methods of election, it helps to be able to have some objective examples. I've used a particular model before, assuming a single dimension and a uniformly-distributed electorate, and it's that model that I'm using here. What I've done additionally here (the whole point of this page) is to be able to compute a palatability index for a given segment of the electorate and a given candidate.
The basic assumptions here are that every candidate is representable by a single number on the closed interval [-5,5], and that the electorate is uniformly distributed on either that interval or some single subinterval. The calculation then attempts to measure a weighted average of the distance of the candidate from each member of the electorate. As the electorate is a continuous distribution, the calculation requires a definite integral, but it wasn't a hard one. (I did run into a removable discontinuity as the width of the interval reached zero, but the limit existed (did I mention it was removable?), so that's okay.)
The palatability scores are scaled such that a perfect candidate (equal to the entire electorate) has a palatability score of 100, and the opposite of that (electorate at one extreme, candidate at the other) has a palatability score of 0.