reply
TRANSCRIPT
Reply By
Peter Wiles
(I) Occam's Razor says tha t "ent ia non sunt mult ipl icanda praeter necessitatem:" not tha t "ent ia non sunt mult ipl icanda." When we regress Soviet on U.S. mili tary expenditures, I submit tha t the existence or not of the Korean War is an ens of such necessitas tha t Occam himself would have insisted on its inclusion. When we regress all European mil i tary expenditures on each other, I submit tha t geographical location (Balkans versus Scan- dinavia) or imperial involvement are infinitely more impor tant than GNP. The simple fact is tha t with so many competing claimants for the role of strategic variable, and so few data, mult ip le regression is impossible; yet only multiple regression would be at all convincing. This is the excuse for my " l i te rary" approach.
I am prompted to make a general observation about Occam's Razor, which is so glibly wielded by statisticians anxious for results. I t does not advocate the simplest explanation, but the simplest good explanat ion; t ha t "praeter necessitatem" is the loophole through which human judgement re-enters every statistical analysis. In other words Occam's Razor tells us almost nothing: it is hardly a principle of scientific conduct at all.
(2) I must apologize to Mr. Pryor here. The result, tha t mil i tary personnel are more positively correlated with population in 1958 than in 1858, is indeed of interest. But I suspect tha t a correct politico-social interpretat ion of this fact remains to be carried out.
(3) Agree.