For democracy to function, its citizens must hold attitudes and then vote in ways that advance their preferences and hold elected officials accountable for their actions (or inactions, in some cases). Political scientists have long debated whether people actually hold attitudes. The most vocal proponent of this perspective is Phil Converse, co-author of The American Voter. In this seminal text, the authors examined national election study data from 1948 to 1956 and found that the respondents were inconsistent in their policy preferences from year-to-year and these preferences did not predict voting behavior. These findings led Converse to conclude that people did not have attitudes or coherent clusters of attitudes (also known as ideology).
Social psychologists assumed people held attitudes, but often found that people's attitudes did not predict their behavior. If people's evaluations of activities, beliefs, policies did not predict whether they acted upon them, then the prospect for the existence of attitudes was bleak. Attitude researchers, fearing the absolution of their topic, sought out to squelch this fear by demonstrating that attitudes just weren't being measured right.
Over time, social psychologists found evidence suggesting two measurement-related explanations for the poor behavior prediction of attitudes. First, they found that attitudes can predict behavior, if the measured attitude is very specific to the behavior to be performed. For example, if I ask you how much you like dessert foods, I might be able to predict that you're going to eat chocolate cake when reading my blog, but I am more likely to be wrong in predicting that you're eating cake now than if I had asked you how much you like to eat chocolate cake when reading about political psychology on the internet. Second, they found that social norms and the motivation to appear to hold socially-desirable attitudes (not being racist, for example) can also create the appearance of attitudes being unrelated to behavior. In an attempt to get around biased responding on sensitive subjects, scholars created "implicit" attitude measures that make it difficult (if not impossible) for participants to respond in a socially-desirable way (see some examples of these measures on the Project Implicit website).
Brian Nosek wrote an excellent summary of the disconnect between implicit and explicit attitudes, showing that the more socially-acceptable it is to hold an attitude, the more implicit and explicit measures converge. For example, people are plenty comfortable reporting their attitudes toward Democrats and Republicans, but less comfortable reporting their attitudes toward thin people and fat people. There are other reasons why implicit and explicit attitudes diverge (e.g., affect, familiarity with attitude objects, etc.), but that is beyond the scope of this post. The key point is that people have some attitudes, but are sometimes unable or unwilling to state what those attitudes are.
So what attitudes do people hold? Which attitudes are they willing to express?
Jon Krosnick presented on his research at the Stanford Summer Institute in Political Psychology suggesting that people tend to hold a few attitudes about issues important to them, and have fickle non-attitudes about issues that are less important. The attitudes people deem important to them usually correspond to values that they learned in their youth (and, potentially inherit via their genetics, although this is still debated).
As people age into young adulthood they form selective social networks which reinforce their policy preferences and further strengthen the policy attitudes. As attitudes become stronger, they grow increasingly resistant to persuasive attempts and stable over time (that is, less likely to appear as the non-attitudes that Converse described). Attitude strength also predicts the way we process information. When we hold strong attitudes, we are much more likely to process information in a biased and "motivated" way, where we seek out evidence that supports our belief and may not even see disconfirming evidence that is in front of us (Lord, Ross, and Lepper have some very cool research on this phenomenon called "biased assimilation"). And, lastly, attitude strength affects behavior. People with strong attitudes are much more likely to engage in behaviors that will further their cause. Think, for example, of someone with strong attitudes on abortion. They are much more likely to participate in a protest than someone who doesn't have as strong of an attitude.
This theory of attitude strength responds to the threats to democracy highlighted in The American Voter by showing that people do hold attitudes on some issues (even if they couldn't care less about others), important attitudes lead to a more responsible citizenry, and the government does receive feedback on the issues of the day from many different constituents (although this feedback comes from different people on different issues).
Furthermore, these findings have implications for political practice. Specifically, when candidates are speaking about a policy issue, they are really speaking to the small segment of the electorate that deems that policy issue as important while those who view that issue as unimportant tune out. Further, when the majority of voters favor the candidate's policy position, the candidate should openly state the position. However, when the majority of voters with strong attitudes oppose the candidate's position, the candidate should remain mute. Lastly, politically-minded folks should recognize that trying to change the policy position of someone who already has a strong attitude on that policy is futile. The better tact, according to Krosnick, is trying to persuade potential voters to enhance the importance they attach to issues where the voters and candidates agree.
So, are we a nation of flip-floppers? Of course not. Well, at least not on attitudes that we deem to be important to us. On other matters, though, a coin-flip may be good enough.