Fuzzy sets are an extension of classical sets, used to mathematically model indefinite concepts, such as that of customer satisfaction. This is obtained by introducing a membership function expressing the degree of membership of the elements to a set. Intuitionistic fuzzy sets represent an extension of the theory of fuzzy sets, in which also a suitable non-membership function is defined. In this paper we aim at quantifying a latent construct, namely satisfaction, using fuzzy sets and intuitionistic fuzzy sets. We put forth a general evaluation method: first, we introduce a fuzzy satisfaction index to obtain membership values. Second, inferential confidence intervals (ICI), calculated through Bootstrap-t and percentile procedures, are used to assess the uncertainty underpinning membership and non-membership estimates. Third, we address the problem of optimal and multiple ICI, as well as their generalization through p values and q-values. In particular, we consider the problem of analyzing the responses to evaluation questionnaires. We apply this new method to a national program of evaluation of University courses and we discuss our framework in comparison with other evaluation techniques.
Inferential confidence intervals for fuzzy analysis of teaching satisfaction
MARASINI, DONATA;RIPAMONTI, ENRICO
2017-01-01
Abstract
Fuzzy sets are an extension of classical sets, used to mathematically model indefinite concepts, such as that of customer satisfaction. This is obtained by introducing a membership function expressing the degree of membership of the elements to a set. Intuitionistic fuzzy sets represent an extension of the theory of fuzzy sets, in which also a suitable non-membership function is defined. In this paper we aim at quantifying a latent construct, namely satisfaction, using fuzzy sets and intuitionistic fuzzy sets. We put forth a general evaluation method: first, we introduce a fuzzy satisfaction index to obtain membership values. Second, inferential confidence intervals (ICI), calculated through Bootstrap-t and percentile procedures, are used to assess the uncertainty underpinning membership and non-membership estimates. Third, we address the problem of optimal and multiple ICI, as well as their generalization through p values and q-values. In particular, we consider the problem of analyzing the responses to evaluation questionnaires. We apply this new method to a national program of evaluation of University courses and we discuss our framework in comparison with other evaluation techniques.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
