Work

Scale Sum Scores Need Not be Recommendable as Dependent Variables Either: Structural Equation Modeling Can Come Again to the Rescue

Year: 2026

Type: article

Abstract: Notwithstanding the recent resurgence of interest in scale sum scores, they turn out to be associated with potentially serious drawbacks not only when employed as explanatory variables, but also if utilized as outcome measures in empirical research. We demonstrate that the use of the popular sum score as dependent variable cannot be generally recommended, due to possible marked bias and mean squared error that can result then for parameters of main concern. Employing appropriately developed structural equation models instead can offer improved parameter estimation, and is recommended to consider with large samples in behavioral and social studies.

Source: Structural Equation Modeling A Multidisciplinary Journal

DOI: 10.1080/10705511.2026.2701280

Authors: Tenko Raykov, Bingsheng Zhang, Christine DiStefano

Institutions: Michigan State University, StataCorp (United States), University of South Carolina

Language: English

Topic: Psychometric Methodologies and Testing

Subfield: Management Science and Operations Research

Field: Decision Sciences

Domain: Social Sciences

Open Access status: hybrid

OpenAlex ID: W7204256128

FWCI:

Cites: 14

Cited by:

Related to: 10