Composite indices for the evaluation of a country's information technology development level: Extensions of the IDI of the ITU

Abstract

A valid assessment of the development status of information and telecommunication technologies (ICT) at the country level is of vital importance because a nation's ICT achievement level is a significant driver of its socio-economic change. A widely used means for such assessments is the ICT Development Index (IDI) of the International Telecommunication Union (ITU). It is a composite measure, which generates country scores from the weighted addition of 11 indicators. Unfortunately, the IDI and many other proposals for similar composite measures suffer from two shortcomings. First, they take the index scoring as an end in itself and do not construe an index in a way which maximizes its capability to predict a specific outcome criterion that translates into socio-economic achievements of a nation. Second, individual indicator and sub-index weights are frequently subjective estimates or are based on inadequate quantitative measurement models. The present work addresses both issues. It applies partial least squares (PLS) structural equation modeling (SEM) to compute aggregation weights for the 11 individual indicators and the three sub-indices of the ITU's IDI in a way that the association between the modified IDI and a chosen socio-economic target criterion – gross domestic product per capita change – is maximized. Both formative and reflective measurement models are used in calculating target-related indicator and sub-index weights and resulting total modified IDI scores for 137 countries for 2012. Whereas indicator and sub-index weights of the reflectively measured modified IDI are similar to the weights proposed by the ITU, substantial weight deviations are detected for the formatively measured modified IDI. At the sub-index level, the access subscale had a considerably lower and the use subscale a considerably higher weight in the formative specification than in the reflective model. At the indicator level, much higher formative weights are assigned to “percentage of households with Internet access” and “percentage of individuals using the Internet”. The two different measurement specifications are taken to calculate and compare modified overall IDI scores and resulting ranks for the sampled 137 countries. We conclude that weights of the individual indicators and the sub-indices in the IDI as suggested by the ITU may not be ideal if the target is to construct an index which is as closely as possible related to GDP per capita growth.