A few years ago, the North Central Sociological Association had its annual meeting. As is common for these sociology meetings, it had a theme: “In Defense of Theory.”
If one isn’t familiar with sociology, this might seem bizarre. Why on earth does theory need defending?
In a successful science, theory is the core of accumulated knowledge. As biologists are fond of saying when they debate creationists, calling evolution by natural selection “just” a theory misunderstands the extent to which a scientific theory can be something so strongly supported that it is for practical purposes considered a fact. To learn physics is to learn physical theory – the theory of universal gravitation, theories of motion and optics. What else would you teach your intro students if not this? Theory is inseparable from the field.
In sociology, on the other hand, theory is an odd specialty. The first extensive exposure most of our undergraduates have to it is in their third year in college, when they take a course (or, sometimes, two) in “sociological theory.” Those who go on to graduate school might take another such course, with lengthier readings, in their first year. Most who go on to be sociologists will never teach a theory course, nor will they read the few specialty journals with “theory” in their titles. The content of these journals, like most of the ideas they learned in their theory courses will have little tangible impact on any research they conduct or any of the subjects they teach.
What has led to this curious state of affairs?
Part of the answer is that most sociological theory doesn’t have a clear use. And, as sociologist George Homans argued back in 1967, the reason for this is that most sociological theory doesn’t explain anything.
Explanation is one of the central tasks of any science, and in other fields much of what is called theory is a set of ideas that explain why things are as they are. In sociology, on the other hand, what students encounter in their first undergraduate “theory” course is typically a mishmash of ideas from some illustrious thinkers – philosophies of human nature, elaborate definitions, broad ideas about how sociology ought to work, and so forth — coupled with some history of these people’s lives and times. It’s usually unclear just how one should use or apply this “theory” in either research or practical problem solving, leading some may question the relevance of theory altogether. And because sociologists and our students have so little experience with explanatory theory, they often seem unclear about what even constitutes an explanation or how to distinguish a theory that explains from one that does not.
So what exactly is an explanation?
Explanation in the scientific sense of the word means saying why the facts are the way they are. Why do we have two high tides a day instead of one or zero? Why do some people commit crimes and others do not? Why does the price of a commodity rise and fall over time?
We answer these why questions by positing relationships between things. We state a relationship between the thing we’re explaining and some other aspect of reality. Why two tides a day rather than one or zero? Because there’s a relationship between the tide and the moon: We get a high tide when the moon is directly overhead or directly on the other side of the earth; given the speed the moon orbits, this happens about twice a day. Why do some people commit crime while others do not? Criminologists propose various relationships – for instance, there is evidence that people with lower levels of self-control are more likely to commit various kinds of crime, as well as other risky behaviors. Why do the prices of commodities go up and down? There’s a relationship between price and supply – all else equal, the greater the supply, the lower the price. So if supply goes up, then price goes down.
Statements that specify a relationship between two things are called propositions. They are also called hypotheses, principles, or laws (depending on their level of generality and how confident we are that they are true). Such propositions make up the core of any explanatory theory, and no body of ideas qualifies as an explanation unless it contains at least one proposition (Braithwaite 1953; Homans 1967:7-31). A theory that explains things has propositions, and a single proposition can qualify as a theory.
As sociologist Donald Black puts it, an explanation is the act of ordering facts with propositions (Black 1995:830, n.6). The proposition shows how this fact fits in relation to other parts of reality. It makes the fact predictable, and so answers our “why” question.
We can have propositions that only address as single specific incident: A happened because B happened. Why did the dinosaurs go extinct? Physicist Luis Alvarez and colleagues famously proposed a relationship between the extinction and an asteroid impact. But scientists often seek general propositions that can apply to a whole class of events. Not just why the French Revolution happened, but what in general makes a revolution more likely; not just why Bill Jenkins developed lung cancer, but what in general causes lung cancer.
Let’s look more closely at the logic involved here.
We have a general proposition like “smoking cigarettes increases the risk of lung cancer.” What facts does this explain? Well, if we compare a sample of smokers to a sample of non-smokers, and find the smokers have higher rates of lung cancer, our proposition would explain that. And if we had data on a population showing that rates of smoking were increasing, and that there was a corresponding increase in rates of lung cancer, it would explain that as well.
What, then, is the logical relationship between this proposition and the things it explains? The things being explained are logical implications of the proposition. They can be deduced from it. Logically, if the proposition is true, then the facts we’re explaining must be true as well.
This is why many philosophers of science understand explanation as a process of logical deduction – what a theory logically implies, it explains (see Popper  2002; Braithewaite 1953; Hempel 1965).
The if-then reasoning of logical deduction is simple in principle, but it can be complicated in practice. We’ll consider the complications in a later post. For now, just note that this simple logic can also be very powerful. It is what gives a general explanatory theory its value. If the theory is accurate, it can be applied to a great variety of particular situations and so explain a great variety of things. Many disparate facts become part of simple patterns. In the words of George Homans, we no longer face “just one damned finding after another” (Homans 1967:27).
Consider a sociological example. We have studies finding that married persons have lower suicide rates than the unmarried, studies finding that suicide victims and attempters have fewer friends than do non-suicidal people, that suicide victims are less likely to be involved in religious congregations, that people with children have lower suicide rates than the childless. Is there any general relationship that implies the results of all these studies? One candidate is the proposition that suicide varies inversely with one’s integration into society – that is, with the nature and number of one’s social ties and involvements.
Furthermore, general propositions can predict facts that we have not yet observed. If we know smoking rates are going up, but haven’t yet done a study to examine lung cancer rates, we can predict what the result of such a study would be. If we know social integration is declining, we can predict a rise in suicide rates.
Prediction in this sense isn’t necessarily a forecast or prophecy of the future. Our predictions might be about something that happened in the past – if it’s true that X causes revolutions, then we expect that X preceded the French Revolution. It’s just a matter of applying the same if-then deductive logic. This logic works the same regardless of whether we’ve already observed the conclusion or not. What a theory explains it predicts, and what it predicts it explains, though as a practical matter we often reserve the term “prediction” specifically for those implications we haven’t yet observed.
The greatest explanatory theories might even predict facts that have never been previously observed, or even dreamed of, but nonetheless turn out to be true. Einstein’s theory of relativity is so amazing because it predicted bizarre things like the bending of space and slowing of time –things that weren’t observed until many years after the theory was published.
It is the predictive nature of general explanatory theory also allows us to gauge the likely consequences of our actions, and so we can apply such theory to find effective interventions and solve practical problems. It is what makes engineering possible.
Accurate explanatory theories should need no defense, because they’re proven so amazingly useful. The problem is that many explanations of human behavior aren’t obviously accurate, and many sociological theories aren’t even explanations – they don’t have propositions that can be used to make predictions.
If they’re not explanations, what are they? We’ll pick up that question in a future post.