Using Regression Analysis to Address Methodological and Theoretical Issues in IT Cost Benchmarking

Abstract

Abstract: The practice of IT cost benchmarking using IT managerial control ratios suffers from a number of methodological and theoretical issues. These issues arise from the following three assumptions: (1) the functional form of the relationship between a numerator and denominator used in an IT managerial control ratio is strictly proportional; (2) the nature of the underlying probability distribution of industry samples of IT managerial control ratios is normal and (3) position of an organization in relation to an industry norm can be unambiguously interpreted. If these assumptions are not met, then determining and interpreting a company’s position with respect to IT costs in relation to industry averages and other companies within the industry is subject to some ambiguity. This paper uses empirical tests and theoretical arguments to show that these three assumptions may not hold true in practice. It is then argued that regression‑based analysis of IT costs can be used to address these issues. Further theoretical and empirical work is needed to develop these regression models so that practitioners can have a reliable and valid method for estimating and interpreting their company’s position with respect to IT costs in relation to an industry norm. At the minimum, practitioners should not rely on IT cost benchmarking for setting their IT budgets without taking into account the methodological and theoretical issues.