Hi,
I am planning to run a same structural equation model for three survey datasets, and the sample size for each is 1000, 200, and 50. The survey questions used for the measurement model are 12 questions.
any issues you can think of and how I can avoid ?
Thank you very much!
Yuting
Hi Chen
your Q is general So, try to determine what are you looking for or narrow your Q?
Hi,
Concerning the measurement model, I recommend you to start the analyses (Exploratory and Confirmatory Factorial Analyses) on the largest sample. After, to ensure factorial structure stability across the two samples (i.e., the same observed score is measured across samples for the same true score), you can test the measurement invariance of the Confirmatory Factorial Analysis model across the other samples.
You may apply the same approach for the structural model. But, some issues may appear on fit with the smaller sample. It may be too small to run an SEM analysis.
Nicolas
Regarding the small sample, i.e. 40, just found a paper talking about the range for SEM, "Sample Size Requirements for Structural Equation Models: An Evaluation of Power, Bias, and Solution Propriety" by Wolf et. al. 2013, maybe this paper can help us determine whether the small sample works or not in SEM.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling