Zur Problematik der Ableitung in sozialwissenschaftlichen Aussagensystemen / On the Problem of Deducing Valid Conclusions in Systems of Social Science Propositions. Part 2

Abstract

Abstract The structure of some kinds of arguments typically to be found in theoretical-empirical social science is examined with special attention to the rules of deduction which in the majority of cases remain implicit. Four types of propositions - describing deterministic relations between attributes (statements of the form ‘if x, then y’), describing monotonous deterministic relations between serials (‘the more x, the more y’) and their probabilistic counterparts - as well as two classes of rules of deduction (based on the properties of transitivity and conjunctivity of corresponding statement forms) are treated. Of the propositions only those of the deterministic if-then kind are unproblematic. The thesis is advanced that most of those ‘arguments‘ presented in social science consisting of sentences of relatively complicated structures - using rules, however, which are exclusively valid for deterministic if-then-statements - are not correct. Deterministic propositions being discussed in Part One, Part Two treats probabilistic ones with the following results: (I) By considering the ‘statistical syllogism’ the non-conjunctivity of probabilistic implications is shown; (II) their non-transitivity follows from the fact that in complex causal structures direct and indirect effects may vary independently of each other. For the simplest case of three attributes a qualitative analogon to the Simon-Blalock-Procedure is constructed and illustrated by an example from mobility research. (Ill) If monotonous probabilistic relations are characterized by coefficients of linear correlation, in the general case no deductions are possible which would use a rule of transitivity. As a general solution to the problem of deducing valid conclusions it is suggested that the verbal language - which, though enriched by a technical vernacular, with respect to its formal apparatus is entirely based on elementary logics - should be replaced by languages of richer logical structures; this, however, means formalization. According to this function to warrant valid deductions the concept of formalization is specified and delimited from similar concepts as the construction of propositions within a linguistic framework of relatively rich logical structure which is given as a special calculus.