The error term in formative measurement models: interpretation and modeling implications

Abstract

Purpose: To clarify the nature of the error term in formative measurement models, as it had been misinterpreted in prior research. Design/methodology/approach: The error term in formative measurement models is analytically contrasted with the measurement errors typically found in reflective measurement models. Findings: It is demonstrated that, unlike in reflective measurement, the error term in formative models is not measurement error but rather a disturbance representing non‐modeled causes. It is also shown that, under certain circumstances, the inclusion of an error term is not necessary/appropriate. Research limitations/implications: Focus is only on first‐order measurement models; higher‐order specifications are not considered. Originality/value: The paper helps researchers in their initial specification of formative measurement models as well as their evaluation of the subsequent model estimates, leading to better specifications for formative constructs.