Wednesday, February 02, 2005

Science and Error

My new question, which I really would like someone to answer for me is: What does error mean in science?

So I think it should be a two part answer. The first part consists of inaccuracy of models and equipment. Science will say "This is accurate down to 10^-10" which means that either (a) external stuffs in nature make it too difficult for our devices to measure said phenomenon [Uninteresting] or (b) the model we are using is flawed past a certain degree of accuracy [more interesting] In this case, Science knows where the barriers to its own models are, and tells you when it stops being a good tool. In theory some version of GUFT or TOE (Grand Unified Field Theory or Theory of Everything) would not have this sort of error.

The second part to the answer is error which we expect. That is to say in any test we expect between 15-30% of the data to miss the mark. So our theory somehow cannot account for these points of error. On some level I understand that statistics accounts for this, but it is still not simple. Even on a super-quantum scale (ie the scientific world not fundamentally tied into statistics) why should we expect to see this form of error? How and why is science limited by statistics in what it can measure and how often it can measure it?

This gained new momentum in my mind when during Law and Social Theory class we discussed the impact of science on social theory. During the premodern period people believed that Science was God's headlight, leading man in the proper direction. After science was severed from theology, science still took on more of a descriptive nature- what do we observe? But with the advent of particle physics in the mid 19th century incredibly precise models are being theorized which allow for us, based on mathematical laws (e.g. F=ma), to predict what we are going to see. Every exception to a rule becomes a path for scientific inquiry, a possibility to create a new rule. Social theory takes this and appears to do the following: There are no exceptions to the rule. Those who we consider abnormal are perfectly normal, rather out models and conceptions of normal are skewed. Social theory has fought the battle to normalize the abnormal for the last century.

Does the notion of error analysis apply to social theory?

3 comments:

Yehuda said...

There was a sign on the wall of my Stat professor's office. It said, "Mathematicians and Statisticians have a relationship based on trust and undertsanding. The Mathematicians do not trust the Statisticians, and the Statisticians do not understand the Mathematicians."

miriam said...

saying "Statistics accounts for it" is meaningless, i think. statistics exists because the world produces irregular results. it may be interesting that their irregularity follows predictable patterns, but the variation is still fundamentally unexplained. so it's not that science is limited by statistics. science is made posible only be the statistical pretense that if something can be accurately described/predicted it has been satisfactorily explained.
really, i tend to take the a priori position, that every theory explains some things and not others - that's just life. from a religious (or philosophical?) perspective, the world has more than we can make of it, so we force it into various boxes that fit different elements and exclude others at different times. i don't believe there will ever be a final theory of everything. there may be a theory, but it will get tweaked and modified, because we are mortal and finite and all that.
what you are saying about "error" in social theory might mean, then, that when people's lives and such are at stake we like to be more careful and self-aware wrt the fact that our theories are necessarily limited. there's no real reason to assume that someone is deviant rather than that a model is flawed. of course we do, anyway, because we have to make decisions and policies and such, but we shouldn't do it without some anxiety.
i'm done now. make fun of me (constructively, please...) if you like.

Ernest Scribbler said...

cf. karl popper