Can I use multiple imputed data in a principle component analysis if my data are not normally distributed?

[Deborah Jehu]

Is it possible to use multiple imputed data in a principle component analysis if my data are not normally distributed? I found this helpful link: https://www.stata.com/support/faqs/s.../cmdok-option/ but I get the following errors because of my non-normally distributed data: Warning: variance matrix is nonsymmetric or highly singular and:
mi estimate: omitted terms vary. The set of omitted variables or categories is not consistent between m=1 and m=41; this is not allowed. To identify varying sets, you can use mi xeq to run the command on individual imputations or you can reissue the command with mi estimate, noisily r(498);

Here is my code:

mi estimate, cmdok noisily: pca OverallFallsRiskScore Dual_TUG NIA_TotalScore GDS shortFES VAS FCI MOCA TrailB_A Stroop3_2 DigitBackward DSST_n, vce(normal) mineigen(1) comp($ncomp) blanks(.3)

Thanks for your help!

Deb

[Nick Cox]

Principal component analysis is just a transformation method. There is no assumption of normality. Prior transformation of the original variables might be a good idea in practice, but that is a different story.

i don’t see that the error message you got arises just because of non-normality.