Wednesday, May 20, 2009

Observation Selection Effects

Selection effects are biases introduced by limitations in one’s data collection process. A notable example is a 1936 presidential poll that predicted FDR's opponent would win the upcoming election in a landslide; yet, FDR won handily. Why was the poll so wrong? Because its main source of data was relatively wealthy people. Depression-era poor overwhelmingly tended to support Roosevelt, but these people were not accounted for in the poll.

Observation selection effects* (OSEs) constitute special kind of selection effect that may be introduced not because of inadequate data collection techniques or tools but because all the evidence depends on the existence of an observer to have the data in the first place. It's related to the old zen question: if a tree falls in a forest and no one is around to hear it, does it make a sound? Admittedly, this is an imperfect example, for we would have several ways to deduce with very reasonable probability that this tree did make a sound when it fell. But suppose there are such things as trees that fall without making any sound - only when no one is around to hear. The OSE would preclude our knowledge of any such trees because in all our experience trees make sound when they fall in the forest.

OSE have become a topic of philosophical interest, and they have important applications in a diverse range of scientific areas, from cosmology to evolutionary theory to questions about possible multiple universes.

I believe that OSEs also have relevance to text and textuality. After all, texts are indeed preconditioned on the existence of an observer - and yes, an author himself or herself may qualify as an observer - to have the text and to know it and use it as a text. And one more key thing here is that we need not only for the observer to "read" the text but also to record or report the existence of the text. So I think almost anytime we discuss texts, OSEs get introduced and need to be accounted for.

Within the domain of textuality, how do we account for OSEs and what does it matter? I don't have clear answers to these questions at the moment, but I am interested in the issues and I intend to continue thinking on the subject. I should also say that my hunch is that OSEs will have applications in Anglo-Saxon Studies as well.

Stay tuned.


* I credit Nick Bostrom for discussing OSEs and observation selection theory.

No comments: