Can you please help me with some information? I am only in the primary stage of learning EFA, CFA and SEM. Can you please tell me what should be the sequence of performing these? I have survey data at hand. Should I perform both CFA and SEM?
Paroma Arefin First of all, clean your data. Second, you may run EFA, but this is optional if all your constructs have been validated in previous studies. Third, you may sun CFA to check the model fit of the measurement model, construct validity, and reliability. Last, run SEM (structural model).
If the factor structure of the scale was examined in a similar sample, you do not need to do EFA. It would be appropriate to perform SEM analysis after applying DFA for each scale. Do not ignore choosing the appropriate correlation matrix for your scale.
Best wishes.
İbrahim.
Paroma Arefin You may find the following article by Anderson and Gerbing helpful. They explain a 2-step approach that examines properties of the measurement (CFA) model first before examining the full structural (latent variable) SEM model:
https://psycnet.apa.org/buy/1989-14190-001
http://www.personal.psu.edu/jxb14/M554/articles/Anderson&Gerbing1988.pdf
In very simplified terms, their point is that it is often useful to identify potential problems (misfit) of the measurement (i.e., CFA) model first, so that those issues can be addressed before examining the (often more complex) structural (latent variable) model in detail. If you started out with a full, complex SEM, and if it didn't fit, then you would have a much harder time finding out about the sources/reasons/causes of the misfit, which could be related to constraints/restrictions in either the measurement (CFA) part of the model, the structural part of the model, or both.
Paroma Arefin I totally agree with Long She you can follow the steps mentioned. However, just for your knowledge, Measurement Model Analysis or CFA is a necessary part of SEM before moving forward to Structural Model Analysis. Exploratory Factor Analysis (EFA) usually serves the purpose of factor analysis during the development of a new scale. However, the measurement model process and steps are different for CB-SEM and PLS-SEM. YouTube will help you get quality stuff as per your choice. Tq
Paroma Arefin try the following best quality YouTube tutorials for AMOS (a CB-SEM tool) by Mike Crowson. These will provide you a detailed understanding of the SEM analysis in AMOS.
https://www.youtube.com/watch?v=PkHt_pOjXbM&list=WL&index=49
https://www.youtube.com/watch?v=DSHa0WreIYc&list=WL&index=48
First of all, You shall have some basic knowledge of the three methodogies. Then compare the difference between EFA and CFA, and choose which one is suitable for your study. The last, if you want to confirm the relationships between the factors explored by EFA or CFA, you can try to use SEM (PLS-SEM or CB-SEM). Hope helpful to you.
Yes, you can do CFA before PLS-SEM, or do EFA before PLS-SEM. CFA or EFA can help us prelimnarily determine the mesure model of the PLS-SEM.
Start by EFA (especially if you translated your survey to your language), then CFA then SEM the below youtube playlist will help you to understand the steps
https://www.youtube.com/playlist?list=PLnMJlbz3sefLdYnz7TysHKrMBby7igpV9
- Badges
- Science topic
- Similar topics
- Statistics
- Survey