The failure of so many polls to predict the outcomes and margins of the last two presidential elections underscores the maddening challenge faced by social scientists who try to predict actual behavior from the expressed attitudes of the people that they sample.
Psychologists have been aware of this dilemma since at least the 1930s. In 1934, Stanford psychologist Richard LaPierre published a classic study in which the anti-Chinese prejudice expressed by 184 restaurants and 66 hotels across the United States resulted in only one refusal to accommodate a group of patrons that included several Chinese individuals. All of these establishments had indicated beforehand that they would not allow Chinese patrons on the premises.
Along these same lines, more recent studies have repeatedly demonstrated that attitudes toward things like attending church or donating blood are often not terribly useful when it comes to predicting who will and who will not engage in these activities.
So, should psychologists just give up trying?
The answer to this question is a clear “No.” Since the 1970s, new statistical tools fueled by the development of computers and more sophisticated research methodologies such as meta-analysis have convinced most psychologists that attitudes, at least some of the time, can be valuable predictors of behavior. The trick now is to understand under exactly what circumstances they are effective.
The information I am about to share is based upon the consensus of thousands of different studies over the past 40 years.
1. Attitudes Are Better Predictors When They Are Strong and Stable
Think for a moment about the thousands of attitudes that you hold. Some of these attitudes, perhaps about politics or religion (which is why we should avoid these topics in most social situations), are very important to you and may even be an essential part of your self-concept. Other attitudes, about different flavors of ice cream, for example, are much more tangential and less strongly held. Not surprisingly, it is strong attitudes that are more likely to predict your actions.
Needless to say, it is also essential that the attitude in question be related to the behavior at hand; trying to predict voting behavior from attitudes about ice cream would be a fool’s errand.
2. The More Specific the Attitude, the Better It Will Be as a Predictor
One of the problems with early attitude studies was that they often relied heavily upon the measurement of very broad attitudes such as whether a person was a liberal or a conservative, and then the researchers attempted to predict behavior in a very specific arena such as engaging in an anti-war protest or volunteering to work for a pro-environmental or anti-abortion activist organization.
However, the usefulness of broad attitudes for predicting specific behaviors is limited. For example, one study found that using a general attitude toward birth control to predict which women were likely to use birth control pills did not work very well, but measuring a much more specific attitude toward birth control pills did indeed provide good predictive power. Similarly, another study demonstrated that attitudes about recycling are much better predictors of recycling behavior than attitudes about environmental issues in general.
The necessity of measuring attitudes as specifically as possible has been confirmed in literally hundreds of other studies.
3. Attitudes Are More Useful as Predictors When They Are “Accessible” to the Person
Sometimes, people are just not aware of the attitudes that they hold when they are engaging in a behavior that is relevant to some of those attitudes. Dragging awareness of these attitudes into consciousness enables them to kick in as a force that drives behavior.
Several studies by University of Minnesota psychologist Mark Snyder and his colleagues illustrated just how important this can be. In one case, he found that favorable attitudes by college students toward psychology experiments did not predict actually volunteering to participate in experiments unless the students were reminded of the attitude they had expressed when a call for volunteers was made. In a separate study, students posing as jurors in a mock sex-discrimination case were more likely to render verdicts consistent with their attitudes when they were given an opportunity to reflect on their attitudes about affirmative action prior to making a judgment.
4. Attitudes Are Better Predictors of Behavior for Some Individuals Than for Others
There are a number of personality traits that may lead some individuals to be guided by their attitudes more reliably than others.
For example, some people are simply higher in something called “Private Self-Awareness.” Such individuals are more introspective and are constantly monitoring and examining their thoughts and feelings. Consequently, a person who scores high in private self-awareness will be more conscious of the attitudes that they hold and they will be more attentive to whether their attitudes and their behaviors are in synchrony with each other. We will naturally have more confidence in predicting behavior from an attitude when we are dealing with a highly self-aware person.
5. Attitudes Are Better Predictors of Behavior When the Elapsed Time Between Attitude Assessment and the Actual Behavior Is Short
A political poll taken two days before an election will be more accurate than a poll that was taken two months before an election. The longer the amount of elapsed time, the more likely it is that other forces may intervene to alter the impact of the attitude.
6. Attitudes Predict General Patterns of Behavior Over Time Better Than Single Instances of Behavior
Another problem with some of the early attitude research was that it often measured an attitude and then used it to predict a person’s behavior in a single experimental session. But we now know that attitudes are better at predicting long-term behavior patterns.
Suppose that I discovered in an attitude survey that you are an environmentally conscious person who believes in recycling. Now, imagine my surprise when I see you casually toss an empty aluminum soda can into the trash rather than saving it to recycle later.
In a single situation, there are simply too many other factors that might be at play for your attitude to be the primary determinant of what you do. Powerful other forces such as being in a hurry to get to a job interview or the time costs involved with searching for a recycling container may overpower even the most sincerely held attitudes and cause you to behave in ways that are totally out of step with your beliefs.
On the other hand, if I were to monitor your daily recycling behaviors over the next six months, I would likely see the influence of your attitudes coming through more often than not and discover that you are in fact a pretty reliable recycler.
In short, the evidence is that attitudes can be good predictors of behavior—as long as we understand the constraints that they operate under. Predicting behavior over longer periods of time and in precise situations will be best, especially if we are using very specific attitudes that are stable and held with conviction.