> It's empirically observable that a proportion of the electorate doesn't vote in its own best interests.

I'm not sure that's true (that it's empirically observable without bias). The trouble comes from an assumption that by giving up X (that you value only a little) to achieve Y (that you value a lot), that those other people would be better off, when they simply value X a lot more than you do (or Y a lot less). When they consistently vote to prioritize X over Y, you empirically observe some outcomes, apply your own weights to each factor, and conclude they're acting against their own best interests and cite your evidence as empirical proof, ignoring the judgment you applied when determining the weights.

I don't think people exhibiting this bias (unconscious or conscious) are intending to be condescending, but it comes across that way quite often.

Some people prefer to live in a society where the government has minimal power/influence over citizens' daily life (X) while other people prefer to live in a society where government power and influence is strong and consistently exercised to protect the most vulnerable (Y). They're different choices.