I enjoyed The Myth of the Rational Voter: Why Democracies Choose Bad Policies by Bryan Caplan (ISBN-13: 978-0691129426)
He starts with an analysis that is a fair starting point for his thesis, although he debunks it in favor of a more complex and subtle thesis toward the end. A "rational" voter will realize that their vote means nearly nothing. It doesn't really matter. Their one vote will not determine the outcome of anything. So, to deal with the cognitive dissonance raised by the effort of voting, they choose their vote based on psychological satisfaction. The "rational" voter does not attempt to consider their self-interest in the policy outcome. They pick a vote that lets them leave the voting booth feeling happy.
This is followed by a fairly extensive analysis of data regarding public perceptions vs informed perceptions of various economic issues. From there, various voting strategies and political strategies are considered against observations. The "rational" voter described above is not a fully accurate predictor of voting behaviors, but it captures the general theory. There is a kind of cascading of specificity both by politicians and voters. For example, voters may select the "feel good" vote for the politician who does something for "jobs". They don't inquire closely into the later decision that "jobs" means a regulatory deal to favor a particular industry and particular unions. The politicians, industry, and unions understand that they cannot break the feel good illusion, but it will be safe to employ corruption to ensure that these jobs go to the highest bidder.
The book is primarily expository and tries to explain the voting thesis, explain the supporting material, and discusses some of the weaknesses of the theory. It's not trying to push a "solution". It's describing and explaining observed behavior, not selling any particular policy.