What is the best program for MSEM? Student Request

Click For Summary
SUMMARY

The best programs for conducting multilevel structural equation modeling (MSEM) include SPSS, MPlus, SAS, LISREL, AMOS, and R2. Given the research parameters involving three levels, nested variables, and a sample size of 250-800, SPSS and MPlus are particularly recommended for their robust capabilities in handling large datasets. The student also inquired about Matlab and its statistical toolboxes, which offer reasonable pricing for students. Alternatives such as Octave, numerical Python, and Julia were also suggested for their interoperability with other programming languages.

PREREQUISITES
  • Understanding of multilevel structural equation modeling (MSEM)
  • Familiarity with statistical software such as SPSS or MPlus
  • Knowledge of data analysis techniques including EFS and PLS
  • Experience with handling large datasets and statistical tests
NEXT STEPS
  • Research the capabilities of SPSS for multilevel structural equation modeling
  • Explore MPlus for advanced MSEM analysis
  • Investigate Matlab and its statistical toolboxes for student pricing
  • Learn about Julia's interoperability with R, Python, and Fortran for data analysis
USEFUL FOR

Students and researchers in the fields of psychology, social sciences, and any discipline requiring advanced statistical modeling using multilevel structural equation models.

Knowledge Sponge
My research requires using the multilevel structural equation model (MSEM). I've read countless articles related to MSEM and I have not been able to pinpoint the best program. They vary from SPSS, MPlus, SAS, LISREL, AMOS, R2, etc. Any recommendations? I've asked around my academic circle and can't get a clear answer. I currently have R2 and SAS but I was leaning towards SPSS or MPlus. This is an out of pocket expense and I don't want to purchase an ineffective program that can't handle the enormous amount of data.

Here's the basic information for my research:
3 levels
Nested within- and between-
Level 1- Group
Level 2- Confounding Variables = quantity 3
Level 3- Latent Variables = quantity 6
Data source is 33 Likert-type questions on a 5-point scale
Sample size range 250-800

I plan to run the following tests:
EFS: Bartlett's test, KMO anti-image correlation matrix RMSEA, SRMSRb and cut-off factor loading CA Culmulative variance.
PLS: loadings, cross-loadings, Fronell-Larcker criterion, CA, CR, and AVE.
GoF: R2, CFI, and TLI.
I will also calculate R2, f2, and Q2.

Thank you in advance for your recommendations.
 
Physics news on Phys.org
Have you looked at Matlab and any of its statistical toolboxes? The student price is reasonable.

Alternatives would be Octave, numerical Python, or Julia. Julia is interesting and can interoperate with other languages like C, R, Python and Fortran.
 
No, I am not familiar with Matlab. I will definitely look it up along with the other ones you recommended. Thank you very much!