Hey all,
I am conducting a CFA on a well know theory, the technology acceptance model theory. In one of the latent variables, I am getting both the AVE and the CR equal to approximately .75.
I do believe this is due to having only two observed variable accounted to the latent variable. As this latent variable is still part of a theory and I cannot deleted, can I go ahead with my SEM and consider this as a limitation? Thank you.
@Jawad Having AVE equal to or greater than CR is not a problem at all as both of the measures wre meant for unidimensionality of the scale. CR might be as low as .75 due to having only two observed items under the latent variable which is not considered appropriate while using covariance based SEM however, in PLS approach, it's acceptable.
It is not a problem per se to have only two indicators in a CFA or SEM as long as the model is identified and leads to a proper solution (i.e., a solution with no out of bounds estimates such as negative residual variances) with reasonable parameter estimates and standard errors. There are even SEM models that use only a single indicator per latent variable such as simplex (autoregressive) models and certain latent change score models. These models can work just fine.
Dear Jawad Harb ,
It is not a problem that AVE = or higher than CR.
You need to fulfill the criteria Cronbach Alpha (CA) > 0.70, Composite reliablity (CR) > 0.70, and AVE should be > 0.50
Hey all, thank you so much for your constructive comments. It was a relief for me 👍
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling