Dear Researchers,
I am quite new to structural equation modeling, therefore, I would like to ask you a question.
In my research, I use a questionnaire to collect data for testing the structural equation model, however, some of the latent constructs I utilize do not have a well-established measurement indicators. In this case, can I introduce indicators relying on conceptual papers and testing construct validity with CFA?
Also, I have a question regarding hypotheses development. Is it crucial that they should be developed on a basis of theory or is it possible to develop hypotheses for SEM on a basis of data obtained from a survey?
Thank you a lot for your help!
Kindly,
Anastasia
In SEM you have to build a set of equations measuring the unidirectional or the reciprocal causation between variables. I would say that it is better to have a hypothetical model based on theory and /or review of literature before start using this powerful multivariate technique. After having a theoritical model and hypotheses then, you go a head and collect data to test if the theoretical model is consistent with reality. As a result, you may accept, reject ar modify your model. This process may lead to theory construction.
1) If no model whatsoever has ever been tested on this survey, I would go with an EFA first (or maybe an ESEM). I would only run a CFA if I had an already-established model to test.
2) Hypotheses should be grounded in theory as much as possible, especially because in building a SEM model you have to decide where will the unidirectional arrows point (i.e., which is the cause and which is the effect). A data-driven approach won't help you with that.
Hello,
Whatever research question you are going to apply it would be appropriate to support it with literature. If not, it would pose a serious limitation for your research. This is something I am struggling to some extent currently, so perhaps some analogous ideas could be used instead of clear-cut examples.
Hope it helps a bit,
Regards, Witold
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling