Tuesday, November 16, 2004

shards of causation


So, back in a special Halloween episode of the recurring JFW feature "Correlation or Causality?," I said that I had run over what seemed like several bottles of beer while out driving, and the next day the tire that ran over the bottles was flat. I got the tire fixed today. The tire repairman identified the problem as being not only that the tire had been punctured by glass but, using whatever forensic techniques at his disposal, he was able to deduce that it was beer bottle glass. At least by my own personal epistemology, this would seem strong evidence for the "causality" thesis.

Irresistible technical addendum: In some quarters of sociology and political science, there are debates about the inferences that can be drawn from small numbers of case studies. One position in this debate, assumed by some quantitative methodologists who have been poisoned by too much thinking exclusively in terms of variables, is logically surprisingly analogous to the idea of knowing that, in a specific case, glass was run over (the explanatory "variable") and that the tire was flat (the outcome), but not having any way of talking about how the discovery of a giant shard of glass sticking out of the tire might be brought to bear on the question of whether the explanatory variable is causally related to the occurrence of the outcome.


2 comments:

Anonymous said...

This is SO Wisconsin

Anonymous said...

Wouldn't the discovered glass be a likely candidate for the role of "mechanism" in a causal account?