Measurement Models in Social Work Research: A Data-Based Illustration of Four Confirmatory Factor Models and Their Conceptual Application

Andrea N. Cimino, Michael O. Killian, Adam K. Von Ende, Elizabeth A. Segal

Research output: Contribution to journalArticle

Abstract

Confirmatory factor analysis (CFA) is a valuable tool for social work researchers to examine validity of measurements and other latent constructs. Best practice recommendations are to specify and evaluate the fit of multiple models, balancing plausibility, parsimony and quantitative indices. However, little attention has been given to the conceptual and theoretical implications of CFA model variations. This article offers a brief report on the state of CFA modelling published in social work research and presents a data-based illustration of four CFA models of a measure of empathy including a single-factor, correlated factors, higher order and bifactor models. We present results from each model and describe the models' conceptual application with substantive explanation and theoretical application to the measurement of empathy. Syntax for all models in Mplus, R, Stata and EQS programmes are provided for reference. As familiarity with CFA and latent variable modelling methods grows, researchers must understand the theory-based implications of varying measurement models and test which model best represents their data and explain their conceptual application.

Original languageEnglish (US)
Pages (from-to)282-301
Number of pages20
JournalBritish Journal of Social Work
Volume50
Issue number1
DOIs
StatePublished - Jan 1 2020

Keywords

  • bifactor
  • confirmatory factor analysis
  • empathy
  • instrument validation
  • structural equation modelling
  • theory

ASJC Scopus subject areas

  • Health(social science)
  • Social Sciences (miscellaneous)

Fingerprint Dive into the research topics of 'Measurement Models in Social Work Research: A Data-Based Illustration of Four Confirmatory Factor Models and Their Conceptual Application'. Together they form a unique fingerprint.

  • Cite this