Il Trimestrale. The Lab's Quarterly, 1, 2008
2. Theoretical contributions in social sciences
Charles Ragin’s Fuzzy Set Social Science (2000) is at this moment one of the main contributes to fuzzy set application in social sciences, even if it is mainly oriented towards comparative analysis on small-n sets. Ragin exhaustively presents a fuzzy-set based qualitative comparative analysis method, specifying a procedure about the determination of membership values; his method is articulated in several steps (Ragin 2000, pp. 165:171): 1. specify the relevant domain of the assessment (universe of reference) 2. define the fuzzy sets (in natural language terms) 3. determine the type of fuzzy sets (multiple values or infinite values) 4. determine the likely range of membership scores 5. identify empirical evidence appropriate for indexing scores 6. translate empirical evidence into scores.
Alternatively, for some specific fuzzy sets Ragin proposes the attribution of membership values by experts or “insiders”, or by way of scores self-attributions by poll respondents. He also presents the possibility of membership values attribution by automated algorithms, as in neural net approach, but he seems to dislike this kind of approach. The six steps identified by Ragin are not to be regarded as in a strict chronological order: in fact, considering their internal logical articulation, it seems difficult to follow them in order, without creating a somehow circular or recursive path. For instance, considering the third point of his articulation - the choice of the type of fuzzy set, with continuous or discrete values - the researcher decides on the basis of his definition of the set, alreadygiven at the second point, but also on the basis of the type of evidences available to him - it’s on the measurement of these "observables" that the score attribution for single cases is subsequently conducted - but the identification of the evidences is introduced in following passages, at the point 5. Beyond the passages in which the entire process of building a fuzzy set is divided, in determining the membership function - the step which I intend to analyze in this ar-