Year of Publication: Jul - 2013
Page Numbers: 68-73
Authors: Chia Yean Lim, Vincent K.T. Khoo
Conference Name: The Third International Conference on Digital Information and Communication Technology and its Applications (DICTAP2013)
- Czech Republic


In conventional online questionnaires, the respondents are always provided with a complete set of static questions, for determining both the relative importance and the degree of relative importance between two criteria in one single question showing a symmetric agree-disagree Likert scale. It can be tiring to perform two different tasks at the same time. As the number of questions increases, it takes a toll on the concentration span of the respondents. In addition, it defeats the purpose of computing a consistency ratio of the responses if it is only performed when all questions have been responded, as in most online questionnaires. This research attempts to explore a novel way for generating the questions dynamically, where only the relative importance of each pair of criteria is required. The inconsistent responses are highlighted as and when they arise, as opposed to the computation of consistency ratio at the end. Effectively, an attempt is made to explore a possible inconsistency detection mechanism with the intention of studying the trends and patterns of detected inconsistencies. Various issues have been investigated including question generation, formation of triads, and identification of logical rules, as well as deliberation on whether a detected inconsistent response should be rectified. If ever an inconsistent response is rectified, what are the reasons behind the detected inconsistency? Can these reasons be meaningfully elicited, stored, analyzed and reused in similar multi-criterion decision-making (MCDM) processes?