I receive emails every so often from readers who ask me which polls I think are suspicious. Readers like these are near and dear to my heart, for theirs is a way of looking at politics that is similar to mine. Inherent to the question is the recognition that - with so much polling data out there, some of it is probably not good. We cannot just blithely use as evidence every bit of data we can get our hands on. Not all data is evidence.

My answer to the readers' question is usually, "few of them, and too many of them." My intuition is that few if any of the major polls make use of invalid methodology. Certainly, some skew in one direction, and some skew in the other direction. This problem can be ameliorated by averaging them, which is precisely what we do here at RealClearPolitics. At the same time, too many polls are over-interpreted by either the pollster who has created the poll, or the analyst who is using the data to make an argument.

What do I mean by "over-interpret?" This is what happens when you fall prey to the fallacy that all respondents in a poll know as much about politics as you do. If you are reading this essay, you are likely what I call an "informational elite," which means that there are major differences between you and the average poll respondent. It is not simply the case that he knows fewer facts about politics than you do, though this is most likely true; I'd wager that you could name at least half of the Supreme Court, while the average respondent will be lucky to get two. It is also that he does not think about politics in the same way that you do. He knows so much less than you that he must organize his political information in a way that, to you, would seem strange and inefficient.

For instance, we like to think of ourselves as an ideologically polarized nation. 33% of America is liberal, 33% conservative, 34% moderate. Not exactly! I would agree that these might be the percentages of people who identify themselves by way of these terms (when they are asked to). However, the percentage of people who make good and full use of the political ideologies we call "conservative" and "liberal" is much, much lower. I am not up-to-date on the latest research, but I have never seen a figure higher than 30%. That is, 30% of the whole public makes at least effective use of these ideologies to organize their political information (so, 30% ideological vs. 70% non-ideological).

This is what political ideologies do. They do many things, of course, but one thing they do is organize the information that we receive. Conservatives who are not experts on education policy "know" not to listen to Ted Kennedy when he argues for whatever education policy he favors. How do they, as non-experts, know that? Their ideology sets up for them a framework for looking at the political environment, and helps them make their way through uncertain situations. So, even if they are not experts in education policy, they know (a) their general philosophical position on the role of the government in something like education, (b) Ted Kennedy's philosophical position, (c) that Ted Kennedy's and their positions diverge. Thus, they can "filter" whatever Ted Kennedy is saying. Liberals can do the same with somebody like George W. Bush. Ideologies are not just a set of normative beliefs, they are also a way to structure and organize the political world.

The implication here is that because a large portion of the public is "innocent" of ideology, it therefore organizes the political world in a different manner. There are a multitude of organizational schema that you, as an informational elite, use that an average poll respondent - who has much less information than you do - cannot make use of. Ideology is one example of such a schema, but it is not the only one.

Thus, when you, as an informational elite, over-interpret a poll, you are falsely applying your organizational schema to the respondent. You interpret a question as you interpret it, and then you assume that every other person who was asked the question interprets it similarly. The consequence of over-interpreting a poll is that you improperly infer the implications of the poll's answer.

One such instance is Hillary Clinton's high negative ratings. This is once again in the news, thanks to Mason-Dixon. It recently found that 52% of respondents would not vote for Clinton. This is an instance where informational elites often fall prey to over-reading polling results.

What does this number mean for Clinton? I would agree that it means something. Obviously, she would be better off if her negatives were not as high. However, I think it is far too easy to over-read these results, i.e. to infer that this implies that she is not electable. If you inferred this from that poll, you were probably over-reading it. I counted at least four different ways to over-read the poll.

First, an informational elite knows all about the other candidates in the race. So, when asked if he would not vote for somebody, he can run a quick mental comparison between that candidate and the rest of the candidates in the race - and offer what approaches a definitive answer. The average respondent, on the other hand, is not capable of that kind of estimation. He generally lacks information about the other candidates in the race. A general election in this country is a choice between two individuals. An average respondent might be willing to say that he is not going to support Clinton, but he might not know enough about the other candidates right now to be able to warrant that his opinion will not change when given a choice between her and somebody else. Maybe he would in fact vote for Clinton if a candidate that he really does not like gets the Republican nomination.

Second, the answer given by an informational elite is probably not conditions-based. In other words, an informational elite usually has a very well-defined political ideology, and therefore has a party preference regardless of what is going on in the country. If the country is at war, a liberal wants a Democrat. If it is at peace, he wants a Democrat. If the economy is strong, Democrat. Weak, Democrat. And so on. The average respondent (at least the one who swings presidential elections) tends to lack this kind of ideological grounding, and so he might change his mind depending upon the conditions in the nation. There might be situations in which a voter who is innocent of ideology could feel that Clinton is the best choice, even if today he claims that he would never think that.

Third, an informational elite might understand the question differently than an average respondent. An informational elite hears that question, and he probably envisions a mental picture akin to the voting booth - in which he is looking at Clinton's name against any other Republican name, and asks what he would do. The average respondent might not see it in those terms. What terms might he see it in? I don't know - but consider the sharp definitions to the mental image that this question creates in the mind of the informational elite. In the mind of the average voter, the question might not evoke such clear boundaries to the mental picture, and thus the answer might not be as precise.

Fourth, an informational elite probably has a fully formed opinion about Hillary Clinton, whereas the average respondent probably does not. This means that it does not matter how much Clinton spends, she is not going to change the mind of an informational elite who does not like her. What about the average voter? Is he susceptible to a change of heart on Clinton? I would wager that the answer is yes - at least for enough of them. Granted, an average respondent knows a lot more about Clinton than he does any other candidate - so obviously, some elements of his opinion are fixed. However, his opinion of her is probably susceptible to some pro-Clinton alterations. After all, he still does not know all that much about her. He might be persuadable - especially by the hundred million or so Clinton would spend on the election. And so, his opinion now - before that hundred mil is deployed - is not nearly as fixed as the informational elite's opinion.

To fail to take these considerations to heart is to fall prey to a false analogy between you and the average respondent. This makes it much more likely that you will over-read the poll - that you will take the answer of the average respondent to be as fully-formed and clearly-defined as your own answer. Accordingly, you will be inclined to overrate the danger that these numbers pose to Clinton.

Obviously, this poll is not great news for her. It means she has some hurdles to jump. She has to change people's minds - and, as the Republican Party will be working very hard to keep minds as they currently are, she might not be successful. However, it is easy to read too much into this, to infer that this polling data is anywhere near decisive evidence against the viability of her candidacy.