The Analysis of Formative Measurement in IS Research: Choosing between Component- and Covariance-based Techniques

Formative measurement is a valuable alternative to reflective measurement when developing indicators of latent variables in structural equation models (SEM). Our goal is to further guide IS research in the application of formative measures by comparing the two dominant analysis techniques: component- (e.g., partial least squares - PLS) and covariance-based SEM techniques. We demonstrate that covariance-based techniques can be appropriate for formative measurement despite a near absence of their use within IS research, which favors PLS. In addition, we discuss the advantages and disadvantages of using either technique and offer six prescriptions to consider when choosing one technique over another for formative measurement analysis. We present these and other contributions towards encouraging the continued and expanded use of formative measurement in IS research and the diversity of techniques to analyze formative measures.