ATLASResearch
methods
Quantitative/ Analysis

Structural equation modelling

Connect measurement models with structural hypotheses

Structural equation modelling combines equations among observed or latent variables, often linking a measurement model to structural paths. Covariance-based SEM evaluates restrictions implied by the whole model. Path analysis is the observed-variable special case. Model identification, estimator choice and temporal or causal assumptions must be separated from the act of fitting arrows.

WHEN IT FITS

Use to evaluate a theoretically specified system of relationships where measurement error or several connected equations matter. Ensure identification and data quality, and justify any causal interpretation through design and assumptions.

Strengths

  • Models measurement and structural relationships together
  • Enables explicit tests of constrained theoretical models

Limitations

  • Equivalent models may fit the same covariance matrix
  • Complexity can exceed the information available

Know the boundary

An arrow in SEM does not establish causality; cross-sectional fit rarely identifies temporal mechanisms.

USED ACROSS
Banking & financeBusiness & MBAPsychologyComputer scienceEducation