I have a question regarding how the number of iterations for MI can be specified in Mplus. In other MI software, there are options to control (a) the number of iterations in the burn-in phase, (b) the number of imputations, and (c) the number of iterations between imputations.
Note: I am interested in specifying the number of iterations manually, that is, completely disregarding the PSR-criterion.
From what I have seen, Mplus offers this option as well. However, even after reading the technical appendix and performing several MI runs with Mplus, I am not entirely sure how the number of iterations are calculated. Here is how I think the respective commands work:
(a) FBITERATIONS = 50000; -> sets a fixed number of 50,000 burn-in (!) iterations, disregarding the PSR-criterion (b) NDATASTS = 100; -> draws 100 imputations afterwards (c) THIN = 1000; -> draws the imputations spread 1,000 iterations apart
Therefore, the total number of imputations would result as FBITERATIONS + THIN*NDATASETS; here, 50,000 + 1,000*100 = 150,000.
Have I understood this correctly? If not, then how are the iterations calculated using the commands above?
Thanks in advance! I am sorry if this question has been asked before. I wasn't able to find this information clearly mentioned in other topics.