Poverty Reduction:

Fuzzy Sets vs. Crisp Sets Compared

抄録

     This paper examines the results of economic poverty reduction modelling in selected countries 1992-2002 using the fuzzy set method (fsQCA) and the crisp set method (csQCA). The fsQCA and csQCA are the two main configurational comparative methods (CCM). This paper primarily focuses on making sensitivity assessments of the fsQCA and csQCA results. The outcomes of CCM based on the truth table algorithm are determined by the calibration of the set-relation membership score as well as the outcome variable of the interim truth table (called the consistency cutoff). Calibration of the raw data into crisp- and fuzzy-set membership scores based on theoretically and empirically grounded establishment of thresholds has been emphasised as it shapes the truth table algorithm. Thus, like previous studies of sensitivity assessment we focus on calibration. However this paper shows how to determine the balance of consistency and coverage outcomes based on various cutoff points as being highly important for a sensitivity assessment. We argue that the optimal consistency cutoff point helps us optimally determine the configurational multiple causality. The outcomes of fsQCA and csQCA are considered in relation to the balance of consistency and coverage. The robustness of the results of the truth table algorithm depends on the balance of consistency and coverage. Using poverty reduction as a dependent variate, we compare the two methods which are both useful.