Friday, February 11, 2011

On Double Blind Experiments and “Scientism”

Robert H. is an old friend of mine from high school. We had this exchange:

Robert H. to Donald Wharton
“Has anyone designed and implemented a double-blind experiment that is capable of falsifying whether or not double-blind experiments are valid? Or would that be an exercise in circular logic? Maybe such a question is just pure nonsense; but it does challenge one to ponder the notions of axioms and undefined terms. Also: Is Scientism a religion?”

My reply:
Robert, it is not possible to prove anything about the universe based on logic alone. Our knowledge about the universe is totally contingent on the data we have. Logic is a useful tool in abstracting and verifying the knowledge we can derive from our evidence. One cannot using logic by itself

We know that double-blind experiments are better than those that are not double blind because of the overwhelming empirical evidence that biases can migrate into data if the researcher has knowledge of their subjects treatment(s). The proof of relative efficacity rests on our knowledge of how unreliable things become without this tool. Even with the use of double blind trials there are substantial problems due to selection bias. Pharmaceutical companies run double blind tests that show their experimental drugs do not work or are harmful. If they don't like the results they can just do another double-blind test and if the results are more to their liking they publish that one.

Scientism is usually a sneer word used by those who wish to undermine the confidence that people place in science. There are a number of words such as this one. The “-ism” as the end is pronounced with a hissing sound of disrespect. That disrespect is purely emotional. 'Scientism” in this usage should have no place in the vocabulary of educated people other than to respond to those who use it. That said, an educated person understands that there is no absolute mechanism that will work to totally distinguish science from literature in all cases. In philosophy this is called the boundary problem. We just need to continually evaluate the tools we use to find an increasingly more accurate understanding of the universe in which we live.


Vincent said...

There have been a few books written on the development of the double blind. It's basically founded on the realization that the person reading the results can be influencing the results with their own preconceptions. The modern double-blind test began around 200 years ago and I presume has been refined. It's more a matter of discovering an influence and trying to remove it. I wonder what the next step will be.

I think you can point out to your friend that the double blind test has been tested because if it weren't effective, the results would be the same as in single-blinded tests, which countless examples show they aren't.
Did you hear about the drug-sniffing dog tests recently? Clearly the dog is blind to the test rules, but if the handlers are not, then the dogs will find drugs where the handlers expect them to be (even if the handlers are wrong)

Don Wharton said...

Vincent, obviously you are right, the double blind methodology has been tested because it removes a source of error. I fear that my friend is a bit unhappy with "science" because people use the term to dismiss some of his more fanciful notions of how the world works. He is trying to asssert the the level of proof fails some higher test of logical certainty.
No I have not heard about the dog-sniffing tests. Can you tell me more and give me your reference?

Vincent said...

Here's an article about the test I mentioned. I think I heard about it on Science Friday.

Don Wharton said...

Vincent, thanks for the reference. It just goes to show how complex things can get when you have the cognitive complexity of a dog as part of the system.