Groups keyboard shortcuts have been updated
Dismiss
See shortcuts

error message in multigroup BCFA

15 views
Skip to first unread message

Nikola Ćirović

unread,
Feb 19, 2024, 1:01:45 PM2/19/24
to blavaan
Dear all, 
I am running mg bcfa with a small model (with 5 groups) and I got the following error message. However, I am not sure how to increase the number of iterations.

hitopvisired <- '
cog.emo.ara.dys=~physio_arousal+fear+TD+Dysthimia+gen_anx+att_dys
externalizing=~Disruptive_mood+antisoc+Self_destruct+substance
social=~antagonistic+Detach
'
fit1 <- bcfa ( hitopvisired  , data = P_study_items_prepared_for_analysisBezrecod,
                            group = "Country")

Warning messages:
1: The largest R-hat is NA, indicating chains have not mixed.
Running the chains for more iterations may help. See
https://mc-stan.org/misc/warnings.html#r-hat
2: Bulk Effective Samples Size (ESS) is too low, indicating posterior means and medians may be unreliable.
Running the chains for more iterations may help. See
https://mc-stan.org/misc/warnings.html#bulk-ess
3: Tail Effective Samples Size (ESS) is too low, indicating posterior variances and tail quantiles may be unreliable.
Running the chains for more iterations may help. See
https://mc-stan.org/misc/warnings.html#tail-ess 

Ed Merkle

unread,
Feb 19, 2024, 3:32:25 PM2/19/24
to Nikola Ćirović, blavaan
Nikola,

Those messages seem to indicate that the model didn't converge, so the results are not to be trusted. The "increase number of iterations" thing can be set by using blavaan's "sample" argument. But this also often indicates that the prior distributions may conflict with your data, especially if your dataset involves large values (say, 100s or larger). In this case, it is possible to specify better prior distributions. The following links provide more info:




Ed
--
You received this message because you are subscribed to the Google Groups "blavaan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blavaan+u...@googlegroups.com.

Reply all
Reply to author
Forward
0 new messages