Copyright Sociological Research Online, 2000


Howard Becker (2000) 'Predictions'
Sociological Research Online, vol. 4, no. 4, <>

To cite articles published in Sociological Research Online, please reference the above information and include paragraph numbers if necessary

Received: 23/2/2000      Accepted: 23/2/2000      Published: 29/2/2000


My realisation that, in an unguarded and unwise moment, I agreed to write something for this special issue of Sociological Research Online on prediction, made me anxious. I had committed myself to doing something I don't believe can be done and so is not worth attempting.

I know that most sociologists think that predicting what will happen next is the hallmark of a real science. If you have real knowledge of social life you know the conditions under which X leads to Y, however complex and multiple the Xs and Ys may be. If you don't know that, then you are merely describing things, not producing analytic propositions that specify causal relations. And so forth. The argument is well known.

I will call on three equally well known arguments to defend my refusal to predict against the charge that this proves that I don't actually know anything worth knowing about social life. The first is that prediction requires not only the possession of such laws or propositions, but also requires that you connect them to phenomena in the world which embody the elements your theory specifies as the essential components of such propositions. You must, that is, connect the Xs and Ys of the propositions to identifiable components of that world. Not so easy to do, since in any such world we study these components seldom appear in pure form and are even less often connected to easily observed social markers that would permit us to make such a connection.

Thus, I might, following Lindesmith, predict that anyone who goes through the three steps of taking enough opiates to experience withdrawal sickness when they stop taking them, recognizing that the sickness comes from not taking the drug, and then takes a further dose to alleviate the pain, will become an opiate addict. But I wouldn't be able to connect that sound scientific proposition to people who might actually do those things in such a way that I could, before they took the three crucial steps, know or guess who would actually take them and thus become addicted. I'd know that if they did (take the steps) then they would (become addicts).

That's my first line of defence: we can't connect our specalised knowledge to the people and groups who might be examples of them. At least not until after the fact. Because it's clear that so-called retrodiction --working backwards from observed effects to causes-- can be done. That's how Lindesmith got to his proposition about the process that leads to addiction.

But many people would not consider that sort of retrodiction a 'real' prediction. No. They want us to tell them what will happen before it happens, not tell them how or why it happened after it already did. They want us to say that the rate of drug use will go up or down this year or this decade, or that people of this or that social type will become, or will more likely become, addicts.

They would like to know, before the fact, who is likely to take those fateful steps: young people or old, men or women, members of which ethnic groups or inhabitants of which areas of a city. Those social markers are relatively easily ascertained and if you could connect them to the three steps, you could make the predictions those people want, But there's no reason to think that you can create an X-leads-to-Y kind of law connecting those easily ascertained social markers to the causal process involved in addiction.

The second line of defence is less practical, more abstract, more 'theoretical'. This is a lesson I learned from answering a question in an examination set by Herbert Blumer: "Discuss the possibility of experimentation in social psychology." The lesson, like most of Blumer's lessons, applied well beyond the limits of whatever social psychology might consist of. It was simply that people and groups build their lines of action in interaction with one another, responding to what others do on the basis of the interpretations they make of others' actions. There are two elements that can't be known in advance in that formulation. One is what the act or will make of what others do, what sort of interpretation he or she will make of someone else's gesture. But even supposing that we had a law relating the X of interpretation to the Y of what others do, we still wouldn't know what the other was going to do that might provoke that 'predictable' response. Might as well ask me to predict the fifth move I will make in a chess game, which I can't do because I don't know what the other player will do to which I will be responding.

My third reason for not predicting is the dismal record of sociological predictions and, even worse, failures to predict. Everett Hughes noted, in his Presidential Address to the American Sociological Association, that all the sociologists studying race relations in the United States in the 1950s and 1960s had thought that those relations were getting 'better' all the time, that the measurements of racial tensions made by experts had indicated that there was no danger of trouble ahead. And then the great disturbances in American cities happened, troubles that no one had predicted before the fact, though lots of us had explanations afterward. And, in the more recent past (much noted at the time but since pretty well ignored), none of the experts had predicted the fall of the USSR. In both cases, I suppose, it could be argued that we hadn't yet achieved the law-like propositions that would make such predictions possible. Not an impressive argument, since few areas of social life had been studied more intensively and for such a long time.

However, even those of us who avoid letting ourselves in for being called positivists or similar bad names succumb now and then to the temptation to predict and not always with such bad results. I think when that happens we are just lucky. I will immodestly cite one of my one predictive successes, which history allowed to come out right, as it might well not have. Sometime in the 1960s I predicted (unfortunately not in print and not even in one of those sealed envelopes deposited with a scientific society so prominent in histories of 17th and 18th century science, but to almost anyone who would listen) that as the amount of LSD use increased, the amount of psychological pathology associated with its use would decrease. This is a good prediction because it is counterintuitive--more use should produce more trouble, not less.

There were several elements in the theory that produced this prediction, all based on my understanding of a similar wave of problems associated with the increase in marijuana use in the United States from the 1920s to the 1950s and beyond. The first element was that pathological reactions to a drug (reactions which come to the attention of physicians and so get such a medical tag) occurred when people who used the drug did not have available interpretations of the ensuing experience which told them they had nothing to worry about. On the contrary, what they thought they knew about the drug's effects gave them reason to worry when they had those untoward experiences. The second was that, after a period in which drug use occurred a lot, users developed understandings of the effects which made them seem less frightening and, further, developed folk remedies which alleviated the frightening symptoms. This would happen if (a very large if and the weakest link in the chain of reasoning supporting the prediction) users were so organized that a body of knowledge of this kind could be developed and communicated to all or most users. My guess was that use would grow through organized networks and that word about individual experiences and experiments would pass from group to group and person to person and eventually reach most, if not all, users.

With these interpretations and remedies available, the larger number of users would also now have a way to deal with them such that they wouldn't need to take people experiencing troubling experiences to a doctor (and thus to be reported as an instance of 'pathology'). Pathology rates would go down.

I thought that something like that had happened when marijuana use began to spread beyond the limited ethnic and professional circles it had previously been limited to. As use spread, physicians reported (in medical journals) cases of 'morbid reactions', people being frightened by experiences for which they had no reassuring interpretation. As use spread further, a culture developed which included those reassuring interpretations and the associated folk remedies.

If things happened that way again, then as more people found LSD available there would be a spike of 'pathological reactions' but more or less simultaneously a 'drug culture' containing the appropriate reassurances and remedies would spread and cause the eventual decrease of reported problems. Note that I did not predict that people would not occasionally freak out as a result of taking LSD, but rather that those incidents would not come to the attention of physicians and others who would treat them as medical problems and report them as such.

I never did the research to test this prediction, let alone its component links, but the amount of use surely grew over the years following and the number of reports of clinical problems dropped precipitously. For what it's worth, most people close to the drug-using scene thought that was exactly what happened.

Well, you might ask, if you had such success then, why not give it another shot? The reason is what I have already mentioned, that I think this one came out 'right' because I was lucky. Think of the hundreds of things that could have happened otherwise. What if communication links between LSD users had not been so redundant that the news of what had happened to this or that person, the report of what had worked in talking this or that person through the kind of frightening experience that came to be called a 'bad trip' had not spread rapidly and evenly through the ranks of users? In that case, the reported instances of morbidity associated with dropping acid would presumably have had a positive correlation with the amount of use instead of the negative one I forecast. It's easy to dream up other conditions that would have scuttled my bold prediction. Which is probably why I wasn't so bold as to publish it and instead just told friends and other professionals interested in the phenomenon.

Mind you, this was a kind of situation about which I knew, relatively, a great deal. Making open-ended predictions about all sorts of other things of which I know very little takes even more nerve that I don't have. Fortunately, some people are less cowardly and bravely make predictions about coming decades and centuries. I still am awestruck by the bravery of my former colleague at Northwestern University, Bernard Beck, never one to be frightened by a journalist with a notepad or a TV camera. Asked by one such what trends could be expected in the then upcoming 80s, Bernie unequivocally predicted that underarm hair would stage a comeback and once more be fashionable. I don't know if he was right, but I admire his chutzpah.

Copyright Sociological Research Online, 2000