Understanding the basic concepts of structural equation modelling is essential before researchers use SEM-based software for analysis. Even with a solid background in statistics or mathematics, it remains useful to understand a few fundamentals directly as a researcher. I will probably spend more time writing about structural equation modelling in the future. The promise has already been kept through this article.
Full text here and https://doi.org/10.1016/j.chbr.2023.100291
What is a correlation coefficient?
These are the bivariate correlations between a construct and its indicators. They determine the absolute contribution of an item to its latent variable. Correlation coefficients are especially important when evaluating reflective measurement models, but they are also interpreted when formative measures are involved. They are also called outer loadings. The recommended value is 0.70.
What is indicator reliability?
Indicator reliability is the square of the correlation coefficient of a standardized indicator. It represents the share of an indicator’s variance explained by the construct and is often described as the item’s extracted variance. Correlation coefficients above 0.708 are recommended because they indicate that the construct explains more than 50% of the indicator’s variance, which provides acceptable indicator reliability.
The beta coefficient?
These are directional coefficients estimated in the structural model, that is, between constructs in the model. They correspond to standardized betas in a regression analysis.
See Hair Jr, J. F., Hult, G. T. M., Ringle, C. M., Sarstedt, M., Danks, N. P., & Ray, S. (2021). Partial least squares structural equation modeling (PLS-SEM) using R: A workbook.