This research is focused on the construction of composite indicators: a complex process involving various steps that have significant impact on the results. One of the main problems in constructing composite indicators is its reliance on multiple subjective judgments (Cherchye et al., 2008). This was clearly demonstrated in the case of Website Excellence Model (WEM) scores, whose main purpose is to assess and compare the performance of Dubai Government departments’ website. Many subjective judgments were being made by different parties in each of the three main stages of the WEM process: pre-assessment, assessment and post-assessment stage. This level of subjectivity led to a problem where many departments end up being unsatisfied with the overall scores and the general process of deriving the results.This research indicates that at each stage of the WEM process, the reliability, validity and fairness of the results were affected. To construct a more accurate, flexible, equitable and transparent WEM scoring methodology, we proposed the use of geometric data envelopment analysis model (G-DEA) along with some general guidelines to be followed during different stages of the process. G-DEA methodology combines positive characteristics of geometric aggregation, Analytical Hierarchy Process (AHP) and DEA. Geometric aggregation makes improvements on two different levels. First, it is better suited for constructing WEM scores than the “standard” additive aggregation, for much the same reasons as for why the switch from additive to geometric aggregation took place for Human Development Index back in 2010. Second, it allows for DEA-like models to be easily extended and applied to a composite indicator irrespective of how complex its hierarchy structure may be. The elements of AHP and DEA contribute through their own well-known properties, such as the reduction of decision bias (AHP and DEA) and an equitable evaluation of departments relative to the observed best practices (DEA).In short, this thesis proposes the use of G-DEA model and discusses the most relevant theoretical and practical aspects and features of that method when applying it to WEM scores. G-DEA methodology is well suited for the WEM scoring framework but there are certainly many other applications, relating to the construction of composite indicators that could benefit from the same methodology. Overall, this study aims to provide both practitioners and academics in the field of composite indicators with a clear application focus on using G-DEA to assess website performance, penetrating the area which so far has never been used in the context of composite indictors. In addition, this study clearly illustrates how G-DEA can combine many good qualities of different well-known techniques for constructing composite indicators.
|Date of Award
|4 Jul 2019
|Ozren Despić (Supervisor) & Ali Emrouznejad (Supervisor)
- composite indicators
- data envelopment analysis (DEA)
- weighting aggregation
- performance measurement