Why Science Requires Theories | Hacker News Reader

Why Science Requires Theories

psychsciencenotes.blogspot.com

29 pointssaulrh8 comments

Economics has the exact opposite issue. We have this very strong theory about optimizing homogenous agents that create a pareto efficient outcome. Any empirical fact that doesn't fit (which happen to be almost all of them) is simply thrown out. Considering this it is quite logical why the main policy recommendation for the last 30 years was deregulate everything, remove all government intervention. The hope was to be finally able to observe some facts that could actually be reasoned about with the existing theories.

Education and its trying-to-be-legitimate partner Educational Psychology is even worse. It is essentially a smal set of theories based on very small sample quantitative and sketchy qualitative research. There's an added bonus of philosophical and political policy initiatives that go against accepted theory "just cause."

Yeah there's a 'hard science' crowd among educational (psychology) researchers, that only accepts quantitative, experimental, randomized studies, but even that form of research has serious flaws when applied to education. They ignore biases that can affect results (most often experimenter expectancy http://en.wikipedia.org/wiki/Observer-expectancy_effect ), and they ignore ecological validity (many findings 'from the lab' end up not working at all, being weakened or even reversed when tested in a classroom). On top of all this, they never open source their software (if any), and they don't share the data.

As one person put it, these studies, usually done with college student volunteers as participants, "include participants who have no specific interest in learning the domain involved and who are also given a very short study time" (http://edtechdev.wordpress.com/2009/11/16/cognitive-load-the...)

Most research is driven by the constraints of tenure-track jobs. Theory doesn't get you tenure. Journal articles with empirical p values < .05 do, regardless of whether they are never replicated, never applied to the field, and never influence changes in practice.

Doing theory takes more time and more space (word wise) than a typical journal affords - it's more compatible with books than journal articles, and most journals don't publish theoretical articles. Books are often not counted for tenure, actually: http://edtechdev.wordpress.com/2010/05/03/would-dewey-piaget...

On top of all this, educational research is barely funded at all. Engineering and medical organizations usually spend 5-15 percent on research & development. In education it is more like 0.01 percent - the majority of which is spent on research, not development. http://edtechdev.wordpress.com/2007/05/18/the-state-of-educa... http://edtechdev.wordpress.com/2010/03/17/need-more-d-in-edu...

It's an interesting argument, but (coming from a physics perspective) I don't know if psychology should let Physics Envy drive it away from being about exploring phenomenology.

It seems to me like any "theory" that psychologists come up with is going to be necessarily less ironclad than what physicists come up with - more likely to be a guideline (a mnemonic, almost) for understanding how phenomena relate to one another. I sort of think their time is more profitably spent in just obtaining encyclopedic knowledge of phenomena.

This could all just be my obsession with first principles talking, though.

I've observed this same issue among medical doctors. They see little coherent theory of the world, just a slew of results validated by clinical trials. Until you have a clinical trial to prove it, a fact falls squarely in the "unknown" category.

I once argued with a doctor about the possibility of cell phone and wifi radiation causing cancer. She couldn't even conceive of the argument I was making: infants are made of organic molecules at room temperature and we know from a large number of experiments how such molecules react to radiation. "But how do you know infants aren't more susceptible?"

For a long time, I thought of medical doctors as scientists. Then one day I realized they were technicians. I still have an enormous respect for them, but it did change my expectations.

Very good points. This is also why traditional economists (while having their own reasons to struggle with being called scientists, at least have theories) complain about behavioral economists. They say, "You point out exceptions to our theories, but what's your counter-theory?" In general the behavioralists are content to point out that the emperor has no clothes.

It's not anybody's fault that there's no theory. It's just too hard to "uncover what's actually going on". Maybe wait 200 years when we have the tools (sensors, processing power, etc) to make sense of things.

Reply on news.ycombinator.com