My research sample size is 383. I want to analysis following structural equation model. So I want to advice which software is better smart PLS or Amos.
For SEM, the adequacy of sample size is relative to the number of parameters in the model. A very general rule of thumb is 10 cases per parameter. In CB-SEM, of course, a parametric multivariate analysis will provide you with a stronger model compared to PLS-SEM that is non-parametric, distribution-free. However, most data violates assumption of CB-SEM and forcing your data by data manipulation and transformation would generate questionable results.
Generally, IBM Amos runs CB-SEM and SmartPLS (one of the popular PLS software) runs PLS-SEM. However, there are also non-commercial software such as Python, R that can run these analysis as well. In the end, it is not about the software, rather which model is possible for best fitting your data.
As Hanif said these two software run different SEM methods. In AMOS we can apply only a CB-SEM analysis, while SmartPLS supports both methods (CB-SEM and PLS-SEM). It all depends on the objective of your research and the population's size. If the objective of your research is the confirmation of theory, than you have to apply CB-SEM. However, this method doesn't perform very well with small sample sizes as yours. PLS-SEM is used when the objective of research is prediction, or the research is exploratory, and the constructs are composites of their indicators. Thus, it is about the objective of your research as well as the size and distribution of your data.
- Badges
- Science topic
- Similar topics
- Filing