Technology forecasting using structural equation modeling based data fusion: Analysis of strengths and weaknesses using a national research and education network example

Abstract

This paper considers strengths and weaknesses of the framework for Technology Forecasting using Structural Equation Modeling based context sensitive Data Fusion, which was first presented by Staphorst et al. in 2013. The framework is an exploratory Technology Forecasting technique that employs a Partial Least Squares based Structural Equation Modeling implementation of context sensitive Data Fusion in order to model complex multi-layered interrelationships between technology inputs, outputs and context related exogenous factors. Strengths and weaknesses considered for this framework, emanating from the extensive bodies of knowledge on Data Fusion and Structural Equation Modeling, include its ability to incorporate contextual information in its forecasting calculations and high sensitivity to structural model misspecification, respectively. An example model instantiation of the framework for the National Research and Education Network technology domain is used to quantitatively analyze the impact of these strengths and weaknesses. This example model instantiation, which is a significantly improved version of the one originally presented by Staphorst et al. in 2014, was constructed using knowledge gained through action research in the South African National Research Network, hypotheses from peer-reviewed literature and insights from the Trans-European Research and Education Network Association's annual compendiums for National Research and Education Network infrastructure and services trends.