Applying neural networks analysis to assess digital government evolution

Abstract

There are diverse measurement systems to assess the advance of digital government, but all are based on the evolutionary perspective. This view consists on a linear, progressive and add-on evolution of digital government, emphasizes the critical role of technology, and the learning sharing or imitation among governments. Official websites and portals have been subject of various studies using multiple metrics and indicators. The existing measurement approaches and traditional estimations present different limitations for assessing the complex characterization of digital government evolution overtime. Up-to-day there is no agreement of what constitute the proper approach to assess digital government evolution and more sophisticate techniques need to be developed to capture a more realistic metric of digital government advance. This article challenges the assumptions of the evolutionary perspective and argues in favor of the potential of the computational tools for the evaluation of these assessment tools of digital government performance. In particular, the technique of neural networks analysis and self-organized maps have the potential to describe the multi-parametric characterization of multiple metrics and indicators of this phenomenon and its evolution overtime. A database of a digital government ranking of Mexican states during the period 2009–2015 is used as a case study. This computational technique was useful data mining and visualization tool of patterns and profiles of digital government performance overtime. The procedure automatically arranges the available data into clusters of characteristics that subsequently are illustrated using visualizations through bi-dimensional maps to analyze the evolution of digital government advance. The results indicate that the evolutionary assumptions do not hold across states in Mexico and the dimensions of information, participation and transaction are relevant for improving digital government evolution overtime. Several theoretical and practical implications are discussed.