Handling the reflective-formative measurement conundrum: a practical illustration based on sustainable employability

J Clin Epidemiol. 2018 Nov:103:71-81. doi: 10.1016/j.jclinepi.2018.07.007. Epub 2018 Jul 19.

Abstract

Objectives: Constructs capturing health or functioning can have reflective and/or formative measurement models. Although a construct's measurement model has extensive implications on the construction, validation, and use of a measurement instrument, measurement models are frequently wrongly or not explicitly specified. As this is likely due to a lack of guidelines, this study uses sustainable employability as an example to demonstrate a) the applicability of an adapted checklist for establishing a construct's measurement model; and b) the use of structural equation modeling to handle formative constructs.

Study design and setting: First, the checklist is applied to sustainable employability to establish its measurement model. Second, using observational self-report data from 2,544 employees, structural equation models are estimated to evaluate the structural and criterion validity of sustainable employability as a formative construct.

Results: The checklist demonstrates strong applicability, identifying sustainable employability as a formative construct. Model fit indices (Comparative fit indices ≥ 0.932, Tucker-Lewis indices ≥ 0.925, root mean square errors of approximation ≤ 0.034) suggest the formative measurement model for sustainable employability is valid.

Conclusion: The checklist and structural equation modeling facilitate handling formative constructs. By establishing sustainable employability as a formative construct, individuals' long term ability to function at work can be more adequately studied and intervened upon.

Keywords: Complex constructs; Formative measurement model; Health; Reflective measurement model; Sustainable employability; Validation.

MeSH terms

  • Employee Performance Appraisal
  • Employment* / methods
  • Employment* / statistics & numerical data
  • Health Status*
  • Humans
  • Models, Theoretical
  • Physical Functional Performance*
  • Reproducibility of Results
  • Self Report