Sunday, 18 December 2016

Methodology: Logic first, then evidence

The confusion of correlation and causation is a problem that plagues all sciences: natural, as well as social. We observe two things changing over the course of a given time period. The question is: does the change in the one thing cause the change in the other thing, or is it merely correlated with it.

For example, we know that, in the vast majority of cases, eating food causes us to feel less hungry. We’re hungry, we eat, we feel less hungry. If we didn’t eat, we’d still feel hungry. Without the cause (eating), the effect (feeling less hungry) does not materialize.

Now, imagine that a person only eats their meals in front of the television, and they never watch television otherwise. This person would observe a strong correlation (or association) between watching television and feeling less hungry. They might even believe that watching television is a joint cause of them feeling less hungry: after all, in this person’s experience, it is only by eating and watching television that their feelings of hunger abate.

Luckily, this being a problem of natural science, there is an easy way to determine whether watching television is a cause or not: isolate the variables. This person should do an experiment: get hungry, and then try watching television without eating at the same time. Then, try eating without watching television. They will find that the television alone did nothing for their hunger, while the food alone did. Eating food is revealed as the cause; watching television as a mere correlation. Problem solved.

For many problems in the natural sciences, this method of variable-isolating, laboratory experimentation can help determine whether something is a cause of, or merely correlated with, a given effect.

In the social sciences though, this method is very rarely applicable. Virtually every aspect of human society is constantly in flux. We can use history and statistics to, imperfectly, observe individual aspects of society at various points in time; but we cannot observe what effects the changing of just one aspect has on the totality. This is because it is never just that one aspect that is changing; everything else is changing along with it. Without the ability to time travel, there is no laboratory experiment we could set up that would be otherwise.

Alright, so the social scientist cannot isolate his variables. Does that mean that correlation and causation will be perpetually confused; rendering the entirety of social science a hopeless muddle? We need not leap to that conclusion just yet. Human reason has another tool at its disposal for dealing with these issues: deductive logic.

Deductive logic starts with basic, well-known truths about reality (what we call ‘common sense’), traces out the various implications of these, and then uses them to shed light on more complex problems. Here is an example:


Premise 1: Whenever someone performs an action, they expect that that action will make them more satisfied.

Premise 2: It is very likely that the individual himself will know what will satisfy his own desires better than anyone else will.

Implication of Premise 1: When two individuals undertake a voluntary exchange of goods or services, both individuals expect that they will be more satisfied after the exchange is complete.

Implication of Premise 2: It is very likely that the individual himself will know what voluntary exchanges will satisfy his own desires the best; better than anyone else will.

Conclusion: It is very likely that allowing individuals to decide upon and execute most of the voluntary exchanges that they want will result in the maximization of their desire satisfaction. This should be taken into due consideration before proposing to coercively restrict any voluntary exchanges.


By using deductive reasoning in this way, we can first come up with a theory that makes sense, and then look for empirical evidence (via history or statistics) that might support this theory. Because it is often the case that two social commentators will agree on the facts themselves, but disagree on the proper interpretation of those facts. In order to determine whose interpretation we should accept, we have to determine, by examining the logic on which it is based, whose theory makes more sense.

For example, both Keynesian (left-wing/bigger government) economists, and Austrian (free-market/libertarian) economists, agree that in the 1930s, the United States was in an economic depression, and that after World War II, the United States was no longer in a depression. But they have very different explanations for how this came about.

The Keynesians argue that the war gave the government an excuse to go on a massive spending spree, a spree which boosted aggregate economic demand and stimulated the economy greatly, thereby putting an end to the depression. The implication is that, in the case of any future depressions, the government need not wait for a major war to break out; they could just engage in a massive peacetime spending spree and achieve the same beneficial result.

The Austrians, on the other hand, believe that the fundamental cause of depression-era unemployment was a mismatch between the wage demands of the powerful labor unions, and the amount of productive capital invested in the economy. If more productive capital had been invested, the wage demands could have been met without creating unemployment. Likewise, if the wage demands had been lower, less unemployment would have resulted. Other factors too, such as the uncertainty for businesses created by the Roosevelt administration’s radical economic innovations, and the sharp turn towards protectionism/economic isolationism, also contributed towards keeping the depression going.

World War II shook things up by causing the suspension of normal consumer life in the United States via strict rationing. Private citizens were forced to consume less than they had been, in order to free up resources for the government to prosecute the war effort with. Luckily for the post-war economy, the government’s great new demand for armaments and war vehicles enabled and encouraged private companies to invest heavily in the advanced manufacturing facilities necessary to produce these things. It so happened that after the war, these facilities were relatively easily converted to peacetime uses. The post-war economy was, therefore, much more capital-rich than the depression-era one, which allowed it to meet the wage demands of the unions without causing much unemployment (there were also fewer workers to find employment for, due to wartime casualties). The fact that the government’s economic policy was more stable in the post-war, and that the United States was in a prime position to provide much-needed goods to the countries who had been pulverized by the war, also helped greatly.

The implication of all this is that future depressions can be solved by encouraging more capital investment, by assuring businesses that future policy changes will not be detrimental to their interests, and by withdrawing support from labor unions who demand wages that are too high relative to the economic situation.

Okay, so we have two very different interpretations of the same sequence of historical events, resulting in two very different implications for future policy-making. It is imperative, for the sake of our economic well-being, that we determine which theory is correct. In order to do so, the empirical facts are not enough; we must examine the logic behind the two theories.

On this front, we have an easy winner: the free-market Austrians. The theory on which the Keynesian interpretation is based doesn’t make sense. There can never be a problem of aggregate demand for goods and services being ‘insufficient’. Until we reach a state of paradisiacal super-abundance, people will always want as many (or as high quality) goods and services as they can afford. The economic problem is never one of consumption; it is always a problem of production. Economic progress consists of a society being able to produce more and more, and better and better, goods and services, with a smaller and smaller quantity of human labor.

Seen in this light, the notion that a war could possibly be an economic boon is absurd. Wars redirect human effort away from improving society’s material standard of living and towards destroying lives and economic goods. They might accidentally result in some peacetime benefits afterwards, but society would almost certainly have been better off if it had just focused its efforts on peacetime improvement directly.    

Therefore, because the Austrian theory has both logic and evidence on its side, we can accept the causation that it posits. And, since the Keynesian theory does not have logic on its side, we must dismiss the causation that it posits as mere correlation.


In sum: in the social sciences, the correct procedure is to start with a logical theory, and only then to look for historical evidence that might back it up. Failure to do so will invariably result in the confusing of correlation with causation, and the consequent suggestion of some very bad policies. 

No comments:

Post a Comment