Actually, this is a perfect example of what sociologists call "status generalization" - which denotes making inferences about qualifications and skills of a person who is hitherto unknown to the observer. According to the status generalization model, such an observer would use cognitively salient clues to make inferences. Such clues typically include conventional signifiers of social status - especially pattern of speech and demeanor - which are conventionally being associated with certain skills and qualifications. This is one of the best empirically documented areas of sociology - although some questions remain about the exact mechanisms through various types of clues work.
Stated differently, what matters is now what a political candidate says, but how he/she says it and how he/she presents him/herself while saying it. A candidate that projects a right combination of "status clues" is likely to be perceived more competent and thus win the election. The important aspect of this model that it moves above and beyond the ration/irrational models of human behavior.
I would dub it "cognitively rational." It retains the purpose-rational aspect of the rationalist model - i.e. it assumes that people would try to maximize their benefit i.e. make the best, most rational decision under the circumstances. However, it substitutes the "perfect information" assumption that often plagues that model with the "imperfect information" one and posits that people use socially constructed cognitive devices to overcome the imperfect information.
To illustrate that with an example - if one finds oneself in a strange city or country and one feels hungry, one would certainly look for suitable places to eat (the purpose-rational aspect of the model). However, in deciding which place is suitable, rather than investigating all catering places in the area and select the best (the perfect information), one would rely on shortcuts, such as known information, past experiences, stereotypes, appearances, brand names, etc. - which are cognitive mechanisms, heuristic devices if you will, facilitating a quick decision making under condition of uncertainty. Indeed, trying to get a full information to make the best possible decision would be quite wasteful under the circumstances, because the transaction cost (finding information) would likely exceed the marginal benefit (i.e. the difference between utility of the best possible place to eat and the satisfactory one found most quickly).
Applying this to elections, the transaction cost of determining the best possible candidate is enormous and beyond the reach of most, even sophisticated, people. It generally exceeds the marginal benefit of making the "right" choice or the marginal cost of making a "wrong" one - with few notable exceptions, such as 1933 Weimar Germany where the marginal cost of the "wrong" choice was simply catastrophic. But in the US, the choice is between policies slightly more or slightly less favorable to the monied class, but which make little difference to anyone else, at least in the short run.
Given that transaction-cost/marginal-benefit balance, a better way to vote is to find a quick way of determining which candidate may "satisfice" one's political expectations. In search of finding such quick ways, different people would use different cognitive devices. The smart folk would examine the candidate's voting record, read what he/she wrote on different political subjects, etc. even though it's a well known fact that past performance guarantees future success. Similarly, the Bobbitts would look at certain personality traits, such as "folksy" demeanor and "one of us" looks, cockiness, or machismo, while religious nuts would look into his "character," "values," self-righteousness etc. Neither of these groups is more or less "rational" than others - they simply use different models of cognitive rationality to make a political decision.
So the bottom line is, as I said in my other posts, that the elections are decided by which cognitive devices produced by the prospective candidates resonate with the majority of voters. This time, it is the ones that appeal to Bobbits (Bush) vs. those appealing to the smart, artsy, and urbane folk (Kerry). Methinks, the latter will prevail because Boobus Amercianus, while still plentiful, lost much of the ground in modern American society - most Americans are not Menckenian boobs anymore, but artsy-smarty types, or at least wannabes.
Wojtek