4. GENERALIZED CRITERION METHOD
Service quality measurement problems are naturally complex. Complex problems involve multiple and mostly conflicting objectives or criteria and often no dominant alternative exists that is better than all other alternatives in terms of all objectives. In this case, the problem is one of value trade-off that requires the subjective judgment of the DM. Such problems are complex, because improving achievement with respect to one objective can be accomplished only at the expense of the other.
Service quality is naturally a multi-dimensional concept.[40,41] Five key dimensions of service quality are reliability, responsiveness, assurance, empathy and tangibles. In other words, to solve a service quality measurement problem, we need to render a Multi Attribute Decision Making (MADM) model. In reality, DM should assess each alternatives based on several dimensions or a set of multi-attribute alternatives. Defining a vector A (A1, A2, …., Am) for all possible alternatives each with k relevant attributes or dimensions, DM seeks to identify each alternative’s service quality score and rank them from one with the highest score to one with the lowest score.
As mentioned in literature, there are some difficulties with classical service quality measurement methods. Two important critiques are the simple constant additive relationships between service dimensions, and the other constraint we named it under achievement constraint. Under achievement constraint implies that all existing service quality measurement methods just try to measure the undesirable deviations from the customers’ expectations. These methods suppose that a service provider performs equal or lower than customers’ expectation (not higher than expectations) in terms of a given quality dimension. In other words, classical GAP models just determine d −i in their complex multi-dimensional service quality measurement process. Mathematically, respect to ith quality dimension and considering G as an expectation, it is possible for a firm to perform lower P1, as P1, between P1 -G, as G, or even between G-P2.