Could you critique or comment on this example which, I hope, reveals the folly of empiricism?
<<So I think that german tennis players are better than french tennis players. You disagree. We both pick our favorite/best german and french tennis player. The german beats the frenchman. They play again. Again the german beats the frenchman. I then turn to you and say “I told you german tennis players were better.” Am I right in saying so?>>
Is this an eloquent or accurate example? Do any of you have better ones? I’d like to have one on hand when discussing things with statheist-types.
Hmmm…not sure I’d go with that. Understand that the problem is not empiricism per se, but strict empiricism in the realm of the social sciences, or, more to the point, the methods of the natural sciences are insufficient in developing theory in economics and the social sciences as a whole.
The reason involves, in part, the fact that each acting person in an economy is a variable, and these variables cannot possibly be controlled for in any kind of experiment, leading to post hoc errors when relying solely on statistics.
Here’s an analogy. In auto racing, drivers want to know what will help their cars go faster on the track. Some variables might include tire pressure, stagger, rear spoiler, wedge, etc…Each one of these can influence lap time. So, when testing to see how tire pressure varies with speed, the pit crew will keep all other components constant, and vary tire pressure while noting the relationship. Then they can hold everything constant but the spoiler, and test that, and so on…it would make no sense to change all the variables at once and then take a few laps…we wouldn’t know which settings are cancelling out.
This cannot be done in economics. Variables cannot be controlled, and we are left looking for statistical correlations without knowing what variables are influencing the data in what way. Thus trying to be empirical leads us into error. It’s like the pit crew changing, at once, all the variables that affect the car’s aerodynamics and then trying to figure out how one of them affects speed. This would be silly. If variables cannot be controlled, then it’s a faulty method.
So in the case of economics, the pit crew cannot control any variables and thus cannot conduct any worthwhile experiments. Even if they could, the subjects would have to know they are involved in an experiment, which would be an additional problem.
The racing analogy seems to work in places where you can mix some Austrian economics into NASCAR discussions, like here in Indiana where I live.
Perhaps that can be cleared up. I had concluded that, by empiricism, the OP was referring to positivism as a foundation for economic theory. Perhaps I misread, then.
I’m not sure that invalidates empiricism as a theory of knowledge. That would just be a hasty generalization. Empiricists such as Hume wouldn’t deny that faulty conclusions can be reached from prior instances; in fact, that’s the basis of the empiricist criticism of rationalism as far as I know.
But if you want to criticize positivism, that’s different than empiricism.
Run the experiment a second time. And a third time. And a fourth time. It will prove absolutely nothing about whether every single German on the planet is to superior to every single Frenchman in playing tennis. That would require more experimentation, suffice it to say.
A guest lecturer at my school was talking about experiments in political science. He was convinced they could be done and was so naive it was hilarious. According to him, if we want to find out whether the prof lectures better when he’s wearing a tie than when he’s not, all we have to do is have him wear ties for some students and not others. Come the end of the term, we examine the exam results of both groups to see whether one did better than the other.
It would only be possible to infer causation from such a correlation if you could hold all other variables equal. Of course, in the real world, ceteris paribus never holds. What if the first group does better because their topics were easier, or because their slides were easier to read, or because the light was better in their lecture hall? How do you control for any of this? The guest lecturer believed that, if you did enough experiments, the influence of other factors would disappear. Of course, this is delusional. The fact is, all the students are human beings, with their varoius goals and dreams and lives and likes and dislikes, who all come into the lecture with their own prejudices and tendencies and experiences, and any one of these myriad considerations could influence their test mark and would require an infinite number of experiments to “control” for.
You cannot experiment on human beings as if they were lab rats. You cannot study man as you study the atom. The crude methods of 19th century science may be useful for studying the water cycle and other inanimate process, but tell us nothing meaningful about purposeful human action.
Of course, because of our belief in a priori analysis, Austrians are predictably accused of dogmatism. Of course, who is the real dogmatist? The person who says that there is only one way to do science, like it or lump it, or the person who says that while empirical knowledge is important, it is not “the sum of all knowledge”, as Hayek says, and there is another kind of knowledge we have access to when we attempt to study human action that we ought to exploit, namely those statements that are necessarily true and self-contradictory to attempt to deny about the category of action.
And naturally, the empiricist-dogmatist is caught in a hopelessly self-contradictory position. The belief that the only statements we should take seriously are those which can be empirically verified is itself a statement which cannot possibly ever be empirically verified (because it is a value judgement, not a factual statement). So either we can chuck all the natural sciences out the window, or we can allow for some possible disagreement about what actually constitutes a “scientific method”. As Hayek writes,
That is something of an example, however it’s not a particularly good one. The problem with empiricism in human affairs is not entirely sample size (although that is a problem), such as this problem is, it is the fact that past performance is no indication of future performance and without an a priori theory the data concerning this is useless. For instance, how can you state what the better tennis team is without knowing what constitutes as “good”? What factors should we look at to determine whether or not it has done well?
Your example fails because it could be rectified through increased and very possible data collection. When all the data is compiled we could say who was better in the past, and this would allow us to fairly accurately predict (so long as there is a fair amount of discrepency and no apparent changes in factors whihc will radically alter the results) who will do better in future.
Mises was a radical empiricist - more than just about any “mainstream economist”. It is methodology which is the main difference… this scientific method = empiricism, is a cultural hold of of the days of British science hobbyists.
Whatever these labels like “empiricism” mean, it shouldn’t be too hard to keep our eyes on the ball if we remember that the final purpose of any scientific methodology is to guide our actions, that is, make reliable predictions. (Some would say the predicting part is outside the scope of science, but either way, no reliable prediction = no benefit.) The question is which methodologies allow us to make the most, broadest, and most reliable predictions, and allow us to continue refining the models that yielded such a benefit so that we may continue receiving more and more reliable guidance in the future.
Regarding the OP’s example, such a method has several flaws: it introduces bias in the choice of “favorite” or “best” players. If you want to do it statistically, you could take a random sample of French players and pit them against a random sample of German players, and if the sample size were large enough you could get a decent confidence interval out of it. Of course, truly random sampling is problematic.
The problem with statistical methods, of course, is they do nothing to explain the underlying mechanisms. If you want to get continued guidance from a scientific methodology, you have to theorize about mechanisms of action. To avoid doing this is to effectively prevent all future advancement in that science until that generation of scientists dies out or Internet rogue scientists tackle the problem. This why basic physics hasn’t advanced in much of a useful way in the past 80 years since the adoption of statistical methods and the abandonment of the search for mechanistic explanations.
Back to Human Action, guys. Mises was no empiricist because Kant was no empiricist.
Kant sought to marry the two (rationalism and empiricism), and you can clearly find this exact sentiment expressed in the first hundred or so pages of HA.
“So I think that german tennis players are better than french tennis players. You disagree. We both pick our favorite/best german and french tennis player. The german beats the frenchman. They play again. Again the german beats the frenchman. I then turn to you and say “I told you german tennis players were better.” Am I right in saying so?”
Oh well a sample size of two! The central limit theorem must hold for a sampling distribution that large!
Seriously, empiricism is just the idea that claiming your theory is “self-evident” is not sufficient, you must check that it’s, at minimum, consistent with reality. Any theory that’s not empirical is just ideology, not science; something based on a bias, not reality.
Not really. I think “inspired” rather than “derived” might make your above sentence closer to the truth. In the natural sciences when you drive down to the essence, theories are that are the composite logical structures of propositions derived from competing hypothetical premises are falsified using local (often needing to be repeatable observations unless your experiment is a “demonstration”) observations, when their predictions that are derived as conclusions deviate from these observations.
In the social sciences however, I think the more fruitful approaches to the subject have been essentialist in nature, along the lines of Menger and Mises, while often being inspired by empirical observations. I think the same could be said of whatever remains “intuitive” in Neoclassical Economics.
The problem of each individual being a variable, influence by the act of observation, is a problem. But quarks are pretty random, and influenced by the act of observation too.
Just because we have not yet come up with empirical theories of human behavior is no reason to think we won’t. I believe cognitive neuroscience is a step in the right direction… they just need to start studying monks and other people who remain peaceful in all circumstance without causing themselves mental damage (there’s being accepting and being happy. And then there’s just forcing happiness, which is a lie and will catch up).