Authors:
Michele Lalla
1
;
Davide Ferrari
2
and
Tommaso Pirotti
1
Affiliations:
1
University of Modena and Reggio Emilia, Italy
;
2
University of Melbourne, Australia
Keyword(s):
Ordinal Scales, Likert Scale, Student Evaluation, Fuzzification, Defuzzification.
Related
Ontology
Subjects/Areas/Topics:
Approximate Reasoning and Fuzzy Inference
;
Artificial Intelligence
;
Computational Intelligence
;
Fuzzy Systems
;
Fuzzy Systems Design, Modeling and Control
;
Soft Computing
Abstract:
The handling of ordinal variables presents many difficulties in both the measurements phase and the statistical data analysis. Many efforts have been made to overcome them. An alternative approach to traditional methods used to process ordinal data has been developed over the last two decades. It is based on a fuzzy inference system and is presented, here, applied to the student evaluations of teaching data collected via Internet in Modena, during the academic year 2009/10, by a questionnaire containing items with a four-point Likert scale. The scores emerging from the proposed fuzzy inference system proved to be approximately comparable to scores obtained through the practical, but questionable, procedure based on the average of the item value labels. The fuzzification using a number of membership functions smaller than the number of modalities of input variables yielded outputs that were closer to the average of the item value labels. The Center-of-Area defuzzification method showe
d good performances and lower dispersion around the mean of the value labels.
(More)