Hi,
I am using SmartPLS. I am confused between common factor model and composite model. Can anyone please provide me the literature which can help me understanding the difference between the both? So far what I have read, I have understood that common factor model is reflective and composite model is formative. Need more clarity on it.
Hi
I think it will be helpful to read the "A Primer on Partial Least Squares Structural. Equation Modeling (PLS-SEM) by Joseph F Hair, Jr. Kennesaw State University. G. Tomas M. Hult"
Good luck
Moreover I recommend to study "Advanced Issues in Partial Least Squares Structural Equation Modeling"
in simple terms, the main difference between the two is that factor model simply tends to examine a model fit and you interpret whether or not factors in a model are fit or not. whereas in composite analysis is when you have a variable that is based on several dimensions. in other words, composite analysis is for when you have a more complicated latent variables.
also the sample size is important. while factor models require larger samples, composite analysis can be done with very small sample size. in a composite analysis, linear combinations of the observable or indicators are considered as composites.
further reading:
Kock, N. (2017). Structural equation modeling with factors and composites: A comparison of four methods. International Journal of e-Collaboration, 13(1), 1-9.
&
Composite-Based Structural Equation Models, Rainer Schlittgen (2019)
the reflective and formative part is actually correct but not solid.
while you cannot test formative measures in factor analysis (covariance based SEM), you can test both reflective and formative measures in PLS (composite)
I recommend
Article Estimation issues with PLS and CBSEM: Where the bias lies!
Article PLS-SEM or CB-SEM: updated guidelines on which method to use
I find some information and hope you find it helpful. As mentioned by Hair et al. (2017) in their PLS book, I just quote directly below
"In a nutshell, the distinction between composite and causal indi- cators relates to a difference in measurement philosophy. Causal indicators assume that a certain concept can—at least in principle— be fully measured using a set of indicators and an error term. Composite indicators make no such assumption but view measurement as an approximation of a certain theoretical concept. In social sciences research, viewing measurement as an approximation seems more realistic (e.g., Rigdon, 2014b), which, from a conceptual standpoint, favors the use of composite indicators over causal indicators. For the sake of simplicity and in line with seminal research in the field (e.g., Fornell & Bookstein, 1982), we therefore refer to formative indicators when referring to composite indicators (as used in PLS- SEM) in the remainder of this book. Similarly, we refer to formative measurement models to describe measurement models comprising composite indicators."
Also, you can find Sarstedt et al.'s (2016) paper useful, which provides a good discussion on this issue.
Reference
Hair 2017 - A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM) [2nd ed.] by Joseph F. Hair, G. Tomas M. Hult, Christian M. Ringle, Marko Sarstedt
Sarstedt, M., Hair, J. F., Ringle, C. M., Thiele, K. O., & Gudergan, S. P. (2016). Estimation issues with PLS and CBSEM: where the bias lies!. Journal of Business Research, 69(10), 3998-4010.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis