Does Industry Classification Matter in IT Business Value Research?

Abstract

Studies which examine the effect of IT on firm performance make the implicit assumption that the choice of industry classification (e.g., SIC instead of NAICS) and the level of aggregation (e.g., 2-digit SIC instead of 4-digit SIC) is unlikely to affect the results. Given that practically none of the studies in our literature review has tested the robustness of their results to an alternative industry classification method, we do not know whether this choice matters (i.e., we don’t know whether the evidence generated from this stream of research is robust). To answer this question, we replicated three studies (Bharadwaj 2000; Chae et al. 2014; Santhanam and Hartono 2003) and found that results are likely to differ if we change the classification method and/or the level of aggregation within a given industry classification method. We encourage IT researchers to replicate prior studies to assess the sensitivity of their results to such choices as well as encouraging future studies to consider implementing robustness checks. The need to consider such robustness checks spans all fields (e.g., accounting, strategic management, and marketing) that use statistical analysis which requires control for role of industry.