Can someone states the steps to be followed when performing a CFA with dichotomous and ordinal data in LISREL student edition 9.30 please.
Deepani,
For dichotomous variables, many would consider using tetrachoric (polychoric) correlations instead of ordinary Pearson product-moment correlations. Tetrachoric correlations assume that each dichotomous variable has a continuous underlying distribution (which is presumed for factor analysis). Likewise, polychoric correlations extend that reasoning to discrete variables having more score values.
Here is a reference that shows how to set up PRELIS for polychoric estimates: http://www.john-uebersax.com/stat/sem.htm
For ordinal scores, the PRELIS processor can normalize the scores prior to analysis. The SIMPLIS command is NScores varlist (or NS), though that's probably not best practice any more.
Here is a reference that is publicly accessible and does touch on the matter of using polychoric correlations: https://link.springer.com/article/10.3758/BF03192753
and their program available at: http://psico.fcep.urv.es/utilitats/factor/Download.html
Good luck with your work.
Dear David, Thank you so much for sharing this useful information.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS