Yesterday, I began an analysis of John McCain's presidential prospects. In my attempts to explain why these prospects have dimmed in recent weeks, I argued that we cannot understand his troubles simply by recourse to his relatively moderate issue positions. In a multi-candidate race such as the Republican nomination, reflecting the ideological preference of the median voter - in this case, a conservative Republican - is not necessarily the winning position. A moderate Republican could, in theory, win the GOP nomination.

So, why has he had so much trouble? The following is an attempt to answer this question. I don't think this is the only answer - but I do think it explains much of McCain's dilemma.

My intuition about McCain's problem is that it is due, in part, to his cultivation of the maverick label. A moderate Republican is different from a maverick Republican. If I say to you that I am a "moderate Republican," you can probably estimate my views with some precision - you take the conservative positions and dilute them by a little bit. But a "maverick Republican" implies a kind of irregularity that the moderate label does not. McCain's status as a "maverick" means not that he always acts like a moderate Republican, but that his actions are unpredictable based upon his partisan identification. Sometimes he acts like a staunch Republican, sometimes like a Democrat. In other words, the "maverick" adjective qualifies the noun "Republican" so much that the latter offers relatively little information about McCain's issue positions.

Think about the function that partisan identification serves for the average voter. If a candidate tells a voter, "I'm a Republican" - what does that mean to the voter? It offers the voter what might be called a cognitive shortcut for understanding exactly what that candidate would do once he assumes office. Thus, that voter can get a sense of whether he can support that candidate - even if, unlike political elites who possess a great deal of information, he has neither the time nor the inclination to review and parse that candidate's issue positions to determine whether they cohere with his own ideals. He knows that he is a Republican - so he can estimate his position on taxes, abortion, government regulation, etc.

And so, partisan identification is a way to acquire information at a low cost. When it comes to making decisions, information of this kind is an unqualified good. In other words, no matter how well off a political actor is, he will be better off if he knows more about the environment. Why? It is because information diminishes uncertainty, which in turn diminishes risk. And all of us are risk averse to one extent or another. We would all like to be able to minimize it as much as we are able. While it is true that, in some instances, actors have an interest in keeping information from other actors, in no instance is an actor less well off if he knows less.

On that note, return to the median voter theorem, which I discussed yesterday. Implicit in the theorem is the assumption that voters have perfect information. That is, if a candidate takes a certain issue position, we have implicitly assumed that all of the voters know that he has taken this position. And so, the risks involved in voting for one candidate over another are zero. You know exactly what you will get if you vote for that candidate.

What happens when we assume that voters have imperfect information? Their desire is to vote for the candidate who will mirror their issue positions as closely as possible upon election - but they do not know enough about every candidate's issue positions to determine with perfect certainty whether they will. That is the risk that they face - they might support somebody who, as it turns out, does not agree with them. In the face of such risk, a candidate's partisan identification is a useful tool. It can help to reduce this risk. It is a low-cost way for voters to get a sense of exactly what the candidate will do in office.

This means that a candidate who does not offer clear partisan identification may be harmed, especially when there are many candidates in the race who do offer such clarity. In an environment filled with uncertainty - what is the rational decision for an information-starved voter? Should he support a candidate about whom he has no information regarding partisanship - or should he support a candidate whose partisanship is, by the voter's estimate, roughly close to his ideal? Clearly, the latter is the better selection. The candidate about whom nothing is known is too risky. The voter cannot be sure exactly what he will do when the candidate enters office - his best move is thus to vote for the person who is a known quantity. Importantly, this is the case even if it turns out that the unknown candidate is closer to his ideal than the known candidate. Because the voter does not know this in advance, he cannot rationally support the unknown candidate. This would be like justifying the irrational purchase of a lottery ticket with the fact that you won the lottery. Sure, you won the lottery - but this event could not have been predicted before you bought the ticket, so the purchase is still irrational!

I think that this is part of McCain's problem. Of course, partisanship is not as much help in primaries - because everybody, after all, is of the same party. However, McCain has intentionally cultivated this "maverick" adjective - so that, he is not "a Republican," but "a maverick Republican." This has earned him kudos with the national press, which does not like partisanship (more on that next week), but what it does is signal to voters who lack information that he is not as firmly affixed to the party - and all of the issue positions which that affixation implies - as other candidates. In other words, in a primary where all candidates swear equal fealty to the party - the party label is not much help as an information shortcut. But when there are differences in partisan loyalties - e.g. when one has a reputation as a "maverick" - the party label becomes more helpful to voters, and problematic for the "maverick."

This is what I think has turned Republican voters from John McCain. They know that he is a "maverick Republican" - and therefore they cannot use their knowledge of what it means to be a "Republican" to estimate what he will do. The word "maverick" qualifies "Republican" so much that the latter ceases to offer much of a cue. They have trouble estimating what kind of president he would make - and so, even if objectively speaking his presidency would satisfy their policy preferences, they do not know this, and are thus now finding another candidate.

This logic applies to elites as well - and so it might also explain McCain's fundraising woes. Elites - who are characterized by the fact that they possess a large amount of relevant information - know enough about politics that they can estimate many of a candidate's positions without relying upon party labels. McCain has undermined the utility of the label for estimating his opinions, but elites do not need the label as much because they have other sources of information that they may use. However, the label can still be valuable to elites. Because McCain is a "maverick" Republican, it is difficult for anybody, elites or average voters, to estimate exactly what he will do on issues about which he has not made some kind of public pronouncement. On issues that are already in the public forum, political elites can do what average voters cannot do - draw upon their knowledge of his stated issue positions to estimate confidently what they can expect of him in office. But what about issues that are not yet salient enough for McCain to have stated his opinion, but that might become salient during a McCain presidency? If McCain was not a "maverick" Republican, informational elites could estimate what McCain would do by imagining what the average Republican would do. However, because he has eschewed the party label, none of us, not even political elites, can estimate what McCain's response would be to a surprising issue development. He might do what a Republican would do; he might do what a Democrat would do; he might do what neither would do. One might say that, even for political elites, a vote for McCain carries with it no warranty. They thus face enhanced risk in support him.

The average voter has this problem on both the salient and not-yet-salient issues of the day - because of his informational deficiency, he is not sure exactly what McCain thinks about policy matters. What is the rational response of the risk averse voter in this instance? It is to avoid McCain. He should select the candidate closest to his ideal about whom he has sufficient information. This is the best decision even if, objectively speaking, McCain would be closer to the voter than the other candidate.

Thus, McCain's problem is not simply that he is too far to the left on the ideological scale. My intuition is that, thanks to his cultivation of this "maverick" qualifier, voters cannot place him on that scale. Risk averse voters thus support the candidate closest to their ideal ideology and about whom they know more.

As I indicated yesterday, I am sure that this does not explain all of the reasons that the McCain campaign has faltered. However, I think this explanation gives us some purchase on understanding McCain's decline. To put it intuitively, I would say that voting for McCain is like buying a used car "as is" without taking it to a mechanic for inspection. You have no idea whether the car is what you are looking for until after it is already yours. It is thus a highly risky purchase - too risky for most of us.