You are using an outdated browser.
Please upgrade your browser
and improve your visit to our site.
Skip Navigation

Sober Lemmings

The Social Norms Approach to Preventing School and College Age Substance Abuse: A Handbook for Educators, Counselors, and Clinicians

By H. Wesley Perkins
(Jossey-Bass, 320 pp., $45)

In the 1950s, Solomon Asch, an enterprising psychologist at Swarthmore College, engaged in some remarkable studies of conformity. Asch wanted to find out whether group pressures would lead people to reject the unambiguous evidence of their own senses. In Asch's experiments, the subject was placed in a group of seven to nine people who seemed to be other subjects in the experiment but who were actually Asch's confederates. The subject's ridiculously simple task was to "match" a particular line, shown on a large white card, to one of three "comparison lines" that was identical to it in length. The two non-matching lines were substantially and visibly different. In the first two rounds of the experiments, everyone agrees about the right answer. But in the third round all the other group members make what is obviously--to the subject and to any reasonable person--a big error, matching the line at issue to one that is conspicuously longer or shorter. In these circumstances, the subject has a choice: he can maintain his independent judgment or accept the view of the unanimous majority. What happens? Remarkably, most people end up yielding to the group, at least some of the time. No less than 70 percent of people went along with Asch's confederates, and defied the evidence of their own senses, once or more in a series of twelve trials.

Asch's findings have been replicated across cultures and nations. Conformity experiments of this kind have now been undertaken in seventeen countries, including Zaire, Germany, France, Japan, Norway, Lebanon, and Kuwait, and there are no less than 133 sets of results. People in nations such as Japan, with "conformist" cultures, do conform somewhat more than people in more "individualist" cultures, such as the United States. Norwegians are more likely to conform to group pressures than are people from France. But the overall pattern of errors does not show large differences across nations.

Asch's experiments involved the length of lines. But do group pressures also affect people's publicly expressed judgments about morality and politics? Consider the following statement: "Free speech being a privilege rather than a right, it is proper for a society to suspend free speech when it feels threatened." In one experiment, only 19 percent of people agreed with this statement when polled individually. But when confronted with the unanimous opinion of others in a small group, 58 percent of people agreed publicly! In a similar finding, people were asked, "Which one of the following do you feel is the most important problem facing our country today?" Five possibilities were offered: economic recession, educational facilities, subversive activities, mental health, and crime and corruption. Asked privately, only 12 percent chose subversive activities. But when exposed to a spurious group consensus unanimously selecting that option, 48 percent of people made the same choice in public.

These are mere laboratory experiments, but countless studies have shown that peer pressure matters in real life. Teenage girls who see that other teenagers are having children are more likely to become pregnant themselves. The level of violent crime is greatly influenced by the perceived behavior of others in the community. Employees are far more likely to file suit against their employers if members of the same workgroup have also done so. The academic effort of college students is affected by their peers, so much so that random assignments of first-year students to dormitories have significant consequences. Richard Revesz has demonstrated that federal judges on three-judge panels are conformists, too. Revesz shows that while a judge's ideology is a pretty good predictor of likely votes, "the ideology of one's colleagues is a better predictor of one's vote than one's own ideology." When environmental regulations are challenged by industry, both Republican-appointed and Democratic-appointed judges become much more likely to accept the challenge if they are sitting on a panel with two other Republican appointees. And my own studies show that when a Republican-appointee judge sits with two Democratic appointees, he often tends to vote as Democrats do; when a Democratic-appointed judge sits with two Republican appointees, he often tends to vote as Republicans do. Federal judges are strongly affected by their peers in numerous areas, including sex discrimination, race discrimination, campaign finance regulation, and affirmative action.

In short, an understanding of conformity explains a lot about human behavior--some of it harmless, some of it disturbing, some of it dangerous. Studies have shown that political violence and even terrorism are often not a product of social deprivation. Much of the time, they are fueled by social pressures. But an important question remains. Is it possible for well-motivated reformers, or for government, to use an understanding of conformity to move people in helpful directions? This book, mostly a collection of essays on substance abuse in college, suggests a fascinatingly positive answer to that question. Unfortunately, most of the essays are not reader-friendly. They are written in an unusually inelegant academic style, and there is too much repetition. And yet the book deserves a lot of attention. It offers an intriguing message, one that is beginning to be used in many policymaking domains, and whose implications go well beyond the problem of substance abuse.

A sociologist at hobart and William Smith Colleges, H. Wesley Perkins has pioneered the "social norms approach," which is motivated by a belief that alcohol and drug abuse are serious problems on college campuses, and by a great deal of skepticism about current efforts to address these problems. A recent survey by the Harvard School of Public Health, for example, finds that no fewer than 44 percent of college students engaged in binge drinking in the two-week period preceding the survey. (Binge drinking is defined as five drinks or more in a row for men and four or more in a row for women.) On campus, binge drinking is in turn associated with serious problems, including physical injury, trouble with the police, and unprotected or unplanned sexual activity. Perkins urges that efforts at rehabilitation "are labor-intensive and expensive" and "do not reduce the overall prevalence of the problem among high-risk youth." Educational efforts, emphasizing health risks, have not been shown to have much effect, especially among young people, who do not care greatly about possible long-term dangers, dismissing their own chance of facing serious harm. Some schools have tried the comic and clueless-sounding strategy of "alternative social events," but these are costly and have not been shown to work. Punitive approaches have had some good results among adolescents, but they have been less successful for college students.

The "social norms approach" is intended as a substitute for these failed methods. Perkins's opening claim is simple. It is that students are likely to overestimate, by a significant margin, the level of substance abuse among their peers. More particularly, Perkins shows that at a wide range of institutions students think that "the norm for the frequency and amount of drinking among peers was much higher than the actual norm or average level of consumption." In addition, students believed that "their peers were much more permissive in personal attitude about substance abuse than was the true pattern of attitudes." This pattern was found at such diverse places as Princeton University, the University of Virginia, the University of Washington, the University of California at Los Angeles, and Northern Illinois University--and at every one of the hundred institutions that participated in a nationwide study. Importantly, the pattern of overestimation can be found even in places where the level of substance abuse is high. Wherever you look, students exaggerate the level of substance abuse, and they think that their peers are more willing to approve of such abuse than they actually are.

Why do such misperceptions exist? As Perkins suggests, there are several possibilities. The first has to do with human cognition--with what we are likely to recall. The "behavior of an individual or a few people under the influence of alcohol or other drugs is easily noticed and remembered," he writes, "whether it is a funny scene of uninhibited action, the disgusting circumstance of someone sick from inebriation, or a frightening encounter with a belligerent or violent individual." Without using the term, Perkins is capturing what psychologists call the "availability heuristic," by which people answer questions of probability by seeing whether it is easy to recall examples. If incidents of substance abuse come readily to mind (and for college students they are likely to do so), then the scene is set for an exaggerated sense of reality.

The second source of misperceptions has to do with the role of the media, which re-affirm and amplify those exaggerations. Instances of substance abuse receive a great deal of attention both in entertainment and in community forums. The result is an inflated sense of the problem. And Perkins does not mention another factor, the "above average" effect, by which most people think that they are better or luckier than the average person. About 90 percent of drivers, for example, think that they are superior to the average driver and more likely to avoid a serious accident. Most people think that they are less likely than the average person to get cancer, to get divorced, to have a heart attack, or otherwise to face misfortune. In this light, it seems reasonable that students would overstate the average level of substance abuse on campus and thus present themselves as doing better, or far better, than the norm.

Whatever the source of the error, Perkins concludes, plausibly enough, that an exaggerated sense of alcohol abuse is likely to have substantial consequences for student behavior. If students are affected by what other students do, the overestimation will inevitably increase alcohol and drug abuse. Hence Perkins urges that excessive drinking is spurred by a "reign of error," with "students following `imaginary peers' as they attempt to conform to erroneously perceived group patterns." If students behave like the subjects in Asch's experiments, then Perkins is on firm ground here. "Students who are ambivalent about drinking or using other drugs and prefer to abstain feel pressure to indulge because they erroneously perceive that `everyone' expects it of them."

So far, so good. In fact, there is nothing terribly original here. But Perkins's fresh idea is that these points can be enlisted in efforts to control substance abuse. Give students accurate information about what other students are doing. If the accurate information shows significantly less substance abuse than students expect, and if students care about what other students do, then the result should be to decrease substance abuse. Remarkably, the approach seems to work. It has been used in many places, and apparently it produces significant decreases. To understand why, imagine that many people now engage in some behavior of which they are not entirely proud--but at least they are able to say, with some confidence, that "everybody else does it." Now suppose that they learn that everyone else does not do it--that they have greatly overestimated how many others do what they do, and how many other people approve of what they do. Perkins's hunch is that some of them will change their behavior.

Much of this book consists of detailed accounts of how Perkins's method has been used at various institutions. The contributors claim to have found a consistent, positive result across a wide range of schools in every region of the United States. The first experiment, and a representative one, involved Northern Illinois University. The central goal was to get all students to understand that on campus heavy drinking was the exception rather than the rule. Hence the university's previously unsuccessful informational campaign, focusing on the harms associated with heavy drinking, was scuttled. It was replaced by a new and quite different campaign, emphasizing the value of moderate drinking and the fact that most students were not heavy drinkers. The new approach appeared to succeed. Between 1989 and 1998, the campus showed significant declines in self-reported heavy drinking, in misperceptions of what other students do, and in self-reported injuries to self or others as a result of drinking. (This study, like the others here, relies overwhelmingly on self-reports--an issue to which I will return.)

A much more aggressive approach was taken at Perkins's home institution, the Hobart and William Smith Colleges. Before the social norms campaign began in 1995, the two campuses had quite serious drinking problems, with 41 percent of students qualifying as binge drinkers. But surveys showed that students thought that the level of drinking was even higher than it was. The experimenters' main goal was to change the misimpression. They began by displaying several sets of posters around campus and in local newspapers. The first set was mysterious, consisting solely of an equation: "2/3 = 1/4." After a set of hints, further posters provided an explanation, to the effect that two-thirds of all students drink only one-quarter of all the alcohol consumed on campus. In this way the experimenters sought to show that heavy drinkers were a minority on campuses, emphasizing on posters that the "large majority of students do only a small portion of the drinking that takes place at HWS throughout the academic year."

A second set of posters, called "Reality Check," began with a description of widely believed myths and then offered corrections. A third set, called "Healthy Choices Are on the Rise," described recent increases in the number of students who didn't miss class or engage in risky sexual practices as a result of drinking. The poster series was complemented by "Campus Factoids" in a newspaper column, containing information on a range of topics, including drinking, where false impressions were corrected. Campus computers were also enlisted, with library and administrative computers displaying relevant messages whenever they remained idle for ten minutes. Online discussions of substance abuse were made available for interested students. The curriculum was affected as well, with a team-taught course on alcohol use and abuse, and with some general discussion of the social norms campaign in the classroom.

Does all this sound silly or doomed? In the authors' account, the experiment did exactly what it was supposed to do. Student perceptions of substance abuse were significantly affected, with reductions in the perceived percentage of heavy drinkers and the perceived average number of drinks at parties. Personal behavior apparently changed as well. The percentage of self-reported abstainers doubled, rising from 5 percent to 10 percent. The percentage of those drinking more than five drinks in a row in the last two weeks dropped from 41 percent to 28 percent. Along every indicator, and among men and women alike, self-reported alcohol abuse decreased. And many other institutions have tried similar approaches with similar success.

Perkins and his colleagues are well aware that the same approach might be used for countless other problems--cigarette smoking, use of unlawful drugs, sexual assault, even failure to attend class. Virginia Commonwealth University tried a social norms strategy to reduce tobacco smoking. A campaign poster emphasized that "7 of 10 college students don't smoke." Before the campaign, 57.1 percent of college students smoked, a percentage that dropped to 46.9 percent ten weeks after the campaign began. Montana has adopted an aggressive "Most of Us" educational campaign, emphasizing that 70 percent of Montana teens do not smoke. The campaign is reported to have produced a substantial decrease (41 percent) in the number of teens who begin smoking. Alan Berkowitz, a contributor to this volume, suggests that male students greatly overestimate the extent to which other male students are comfortable with forced sex and with behavior that objectifies or degrades women. He contends that if male students had an accurate understanding of what other male students thought, sexual assault on campus would be significantly reduced.

A similar approach has been used outside the campus setting. Studies show that people are far more likely to give to charity if they are informed that many or most people give to charity. (There is a helpful lesson here for those who seek charitable contributions.) Or consider Minnesota's various strategies to increase tax compliance. When people were told of the risk of punishment, compliance levels were unaffected. When people were told that taxes are used for important goods and services, including education and police protection, compliance levels were unaffected. But when people were told that more than 90 percent of citizens fully comply with the tax laws, compliance increased. Apparently those who violate the law are ashamed to learn that their conduct is worse than that of the overwhelming majority of their fellow citizens.

Perkins and his colleagues appear to have discovered a policy tool that has many applications and that might actually succeed. But their work raises a number of questions. The most fundamental has to do with the reasons for conformity. Why, exactly, is people's behavior, including that of students, so responsive to perceptions of what others do? The contributors say too little about this question. Two factors seem important. First, people usually want to do what's right (or at least what's not wrong), and the actions and beliefs of other people contain information about what's right (or at least about what's not wrong). If most people drive over the speed limit or eat meat, then there is at least some reason to think that it is not morally objectionable to drive over the speed limit or to eat meat. Moreover, people usually want to be in the good graces of others. If most people think that it is acceptable to smoke cigarettes or unacceptable not to, then there is reason to go along with them so as to avoid their disapproval.

In Asch's line experiments, both explanations played a role. Some of Asch's subjects feared that their own opinions must be wrong. Others thought that their own perceptions were fine, but they did not want to look like idiots in front of a group of people who saw things otherwise. Undoubtedly some of those who engage in heavy drinking fall in the same category. And insofar as college students are peculiarly vulnerable to the cues given by the behavior of their peers, overestimates of substance abuse will greatly affect behavior. Perkins's findings seem highly plausible in this light.

But are those findings themselves overstated? Most of the authors emphasize changed student perceptions of what most students do--a fact that is not, by itself, all that interesting. What matters is not perceptions, but whether student behavior actually changes. And here the overwhelming bulk of the evidence comes from students' self-reports--from their own descriptions of their behavior. This is hardly the most reliable measure. It is easy to imagine that the social norms strategy would affect self-reports without affecting actual use. Suppose you learn that your behavior is highly unusual; you might then understate it without changing it. Perkins and his colleagues offer too little hard evidence about the objective measures of alcohol abuse: hospitalizations, arrests for drunk driving or liquor-law violations, incidents involving campus and local police, and the like.

To be sure, one of the studies, at Hobart and William Smith Colleges, does go beyond self-reports to look at arrests for liquor-law violations. And here the evidence is consistent with the self-reports: a 46 percent drop in the period of the experiments, from 84 arrests to 44 three years later. But the authors do not say whether they controlled for other factors that might have caused or contributed to the reduction in the number of arrests, such as a rise in the price of liquor, an increase in punishment for liquor violations, or a change in enforcement policy. To be sure, their findings are highly suggestive, and because they are consistent with so much that we know about human behavior, my guess is that there are genuine success stories here. But I wish that the authors had relied much more on objective evidence.

There is a related issue. While people's behavior seems to have moved in the desired direction, the changes are not huge. Recall that at the University of Arizona the level of heavy drinking dropped from 40 percent to 30 percent, and that at Virginia Commonwealth the percentage of smokers dropped from 57.1 percent to 46.9 percent ten weeks after the campaign began. That is good, but it is not great. And, contrary to the authors' claims, it is not clear that the behavioral changes are merely a result of better information about what students do. These experiments involve far more than a simple presentation of the facts. They have a clear message, endlessly repeated on posters and computer screens: heavy drinkers are violating the social norm, and it is wrong to violate the social norm. The effort to change behavior uses the jargon of youth ("Reality Check"), provokes curiosity ("1/4 = 2/3"), and even bombards the senses. If students are responding to that effort, they are not reacting to the bare facts, but instead to an artful and highly manipulative advertising campaign.

This point raises a much larger one. Return to the massive campaign at Hobart and William Smith Colleges, with the three waves of happy-sounding posters (culminating with "Healthy Choices Are on the Rise" and related messages popping up on idle computer screens). Learning about this campaign, I was reminded of my experience in Beijing about fifteen years ago, when I read the China Daily every morning. The China Daily was filled with merry news: about the week's good deeds, about the growing happiness and increased productivity of happy workers, about the wisdom and kindness of leaders. It would not have been at all surprising to see a big headline that said "Healthy Choices Are on the Rise." I don't know for sure, but my guess is that the China Daily did not have much of an effect on behavior. One reason involves trust: if a source of information is known to be attempting to manipulate people, the manipulation is unlikely to work. And if people feel infantilized by official statements, they will cease to take those statements seriously. Some of the studies here sound a little like the China Daily. But efficacy is not the only issue. It is hardly unproblematic to try to manipulate people, even if the manipulation can be made to work and even if it is in the service of desirable ends.

We can get at this issue by asking some questions of fact. Would the social norms approach work at an institution at which the incidence of substance abuse is very high? If the existing norm is really bad, won't the approach backfire? Perkins is aware of the problem, and he responds that some of these studies were done on campuses with high drinking rates, and they worked even there, simply because students believed that the rates were even higher than they actually were. In addition, the "majority of students always hold a relatively moderate attitude about drinking, even if they do not always behave in accordance with the attitude." But on this subject Perkins has something more disturbing to say. He insists that it is always possible to identify some variable where things are actually pretty good--and to publicize that one variable if we wish: "If one measure of an actual norm is not as positive as we might like, we should consider ... what other measures might also be available that give a different picture." In other words, "we" might consider highlighting, and publicizing, the particular "measure" that portrays the norm in the most favorable light. The editors of the China Daily could not have said it better.

Perkins should be a little more selfconscious about the risks. Advertising executives, and propagandists of all stripes, usually select the best measures for moving people in their preferred directions. Since there are so many measures, there is a real danger of manipulation. In fact, some such measures could easily be invoked, if we wished, to increase the level of substance abuse. Consider an example. In 1995, 95 percent of students at Hobart and William Smith Colleges said that they did not abstain from drinking. This figure could well be used to marginalize abstainers--and to decrease their numbers. Social norms strategists come in any number of stripes, and those who know about the power of conformity might well choose a "measure" that would serve their interests. (How many Europeans distrust the United States? How many Americans accept the divinity of Jesus Christ? How many Palestinian teenagers are willing to consider becoming suicide bombers? And isn't terrorism on the rise?) When they are dealing with college students or anyone else, they should treat their audience with respect, rather than as objects of marketing.

Even more fundamentally, the social norms approach offers an unfortunate, if implicit, message: you should care a great deal about what your peers are doing, and you should be inclined to do what they do. This message fits uneasily with the basic premises of liberal education, which tries to inculcate independent judgment and some skepticism about the crowd. John Stuart Mill complained that "society can and does execute its own mandates," and so he urged the importance of protecting "against the tendency of society to impose, by means other than civil penalties, its own ideas and practices as rules of conduct on those who dissent from them." Perkins and his colleagues are attempting to enlist that very tendency, and there is a real problem here. If students should not drink heavily, the reason is not that most students do not drink heavily; it is that heavy drinking is harmful to oneself and to others. Social norms strategies try to motivate behavior by emphasizing the wrong factor.

And yet substance abuse is a serious issue, and if social norms strategies work well, we probably should not object to them too strenuously, especially if administrators are correcting widespread misconceptions by telling the truth. Some of the authors here are too preachy; their research methods leave gaps; and perhaps they take excessive delight in their ability to move students in their preferred directions. But Perkins and his colleagues deserve considerable credit for developing a promising method for reducing harmful behavior, one that has the unusual virtue of improving matters without relying on punishment or even regulation.

Cass R. Sunstein is a contributing editor at The New Republic.

By Cass R. Sunstein