At 11:51 AM 9/23/99 -0400, Max Sawicky wrote: >WS: . . . You also dismiss my argument that you may not have sufficient >empirical evidence to sort out effects of different variables by simply >calling it "babble." Well, my friend, if you ran a multiple regression with >twelve variables plus interaction effects and six cases - you would be >laughed out of the stage. What makes you think that a case-based approach >is any different, from a methodological point of view. . . . >> > >Tho I agree JB has been a little too big for his britches, >I wonder what the above means for historical analysis. >More often than not there are not sufficient cases to >use statistical tests of hypotheses; or the question >is too broad to admit of analysis via a data set. So >where does that leave historians, both economic and >otherwise? Max, imho the problem is how you construct your unit of analysis - the bigger the unit (e.g. nation-state), the fewer cases you get while the picture becomes more complicated and difficult to analyze. My suggestion would be constructing a unit of analysis at a relatively low leve of aggregation, e.g. a firm/organization instead of the nation-state or, goddess forbid, 'the world system.' This way you can: - effectively address the problem of human agency versus environmental influences - get enough emprical material (cases) to run meaninful comparisons, both within and between nation-states; - get enough cases to meet the 'ceteris paribus' and provide counterfactual - whi8ch is necessary to analytically separate and demonstrate the claimed effects of individual variables. For example to adress the question of 'what made capitalism work and reproduce itself' - it would be more fruitful to analyze the basic unit of production under capitalism and, say, fedualism and see what they share in common and how they differ - rather than addressing issue at the nation-state level and trying to guess th efactors that brough about a capitalist 'system.' To my knowledge, Russian historical economist A.V. Chayanov used that approach quite effectively. To summarize, i'd say keep your cases (units of analysis) simple, multiple - to ascertain comparisons and analytical separation of effects, and empirically verifiable (is there a counterfactual to your case?), do not loose human agency from sight, and stay clear of nation-states and world systems. wojtek
