Troubleshooting
Problem
AMOS uses FIML (Full information maximum likelihood) to impute for missing values. Is this the same as the EM (expectation maximization) method in the Missing Values module?
Resolving The Problem
The implementation of FIML (Full information maximum likelihood) in AMOS is not always equivalent to an EM (expectation maximization) based method, though they will usually give very similar results.
Since SPSS Missing Value Analysis only estimates a mean vector and covariance matrix and doesn't fit a further model, doing the same thing in AMOS should match it.
However, comparing AMOS FIML to EM in general won't always give identical results.
Please refer to the link : http://www.smallwaters.com/amos/faq/faqa-missdat.html#t1q1
Related Information
Historical Number
61379
Was this topic helpful?
Document Information
Modified date:
16 June 2018
UID
swg21478046