It should be no secret that most voters decide whether to support a candidate more on their perceptions of the candidate's character than on whether that candidate's positions on the issues are of benefit to them. This is not to say that issues are irrelevant; positions on some issues are seen as signs of character, and positions on those issues can seriously affect a politician's chances in elections. But issues that are not seen as indicative of character have far less effect, as seen in the situation all liberals bemoan where vast numbers of people consistently vote for politicians who advocate conservative policies that only benefit a narrow elite.
Now, I'm a highly trained philosopher. I'm pretty good at evaluating arguments, and the evaluation of evidence in general is something that I've examined closely in my work in epistemology and philosophy of science. A question I've thought about a bit (motivated somewhat by a conversation with Felicia Nimue some time ago) is whether it's even a good strategy for a more ordinary voter to be as motivated by the issues as I am. To some extent, I can tell the difference between a good argument and a scam, though even I can be fooled a disturbing amount of the time. What about somebody who has no background in statistics, hasn't been educated in how scientific studies are conducted and so has no basis for distinguishing the good ones from the bad ones, lacks my knowledge of history for looking at what the consequences of policies have actually been, etc? If someone prevents them with one of the powerful arguments that, say, single-payer health care is a good idea, should they believe it? Isn't it quite legitimate for them to suspect they're being snowed?
Now, it's easy to say that they should do the research and learn what the evidence is and how to evaluate it. That's certainly one reason education needs to be a very high priority. But not only is it utopian to expect any dramatic shift in voter behavior in this area in the near future, for many voters it is also quite impractical.
It is also important, of course, to recognize that people aren't as good of judges of character as they think they are; perhaps people should be weighing the evidence rather than judging characters, even if they're bad at weighing evidence, because they're also bad at judging character. For today's purposes, it is enough to note that people are bad at both things, so the standard voter tendency to go with character rather than evidence of the value of positions on issues is not as obviously irrational as some seem to think; looking at issues would not necessarily lead an average voter to better decisions (on the other hand, while I like to think I'm better at evaluating evidence than the average person, I certainly have no advantages when it comes to judging character; I presume I'm as bad at that as anyone. So issues are the way to go for me).
What are the consequences of this? Well, I certainly don't think that we should give up on voter education efforts; it would be nice to have more people actually thinking about the issues and voting on the basis of informed deliberation. But it would seem that focusing on emphasizing the character virtues of liberal candidates and character defects of conservative candidates should not necessarily be seen as merely strategic, merely a way of winning elections. Perhaps voting on character is not as misguided as some of us often think, and so if we really want to help ordinary voters vote well, we should be devoting more effort to helping voters accurately evaluate character, rather than trying to get them to look at issues instead.