Abstract
Despite increases in the number of articles published in higher education journals using structural equation modelling (SEM), research addressing their statistical sufficiency, methodological appropriateness and quantitative rigour is sparse. In response, this article provides a census of all covariance-based SEM articles published up until 2013 (N = 143) across eight leading higher education journals: The Review of Higher Education, Journal of Higher Education, Research in Higher Education, Higher Education, Studies in Higher Education, Assessment & Evaluation in Higher Education, Higher Education Research & Development and Journal of College Student Development. Several areas for improvement are found regarding the statistical application of SEM in higher education research. Recommendations are to focus on: building theoretically supported models, data screening, missing data, estimation methods, sample size and power, fit indices, validity and reliability and the testing of alternative models. Best-practice statistical guidelines for higher education researchers wishing to apply SEM to their research are provided.
Acknowledgements
A sincere thank you to Associate Professor Ravi Pappu for introducing me to SEM, inspiring me to learn more advanced SEM techniques and for providing helpful comments on the final version of this article.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1 Thank you to an anonymous reviewer for this suggestion, the interested reader is referred to lavaan.ugent.be/.
2 The search was manually completed and cross-checked on 28 August 2014. Any omission of CBM SEM articles from this review up to and including 2013 in the aforementioned journals is unintentional.