Re: Warning messages when trying to fit bifactor model in Lavaan. What do they mean? How to solve this?

Skip to first unread message
Message has been deleted

Terrence Jorgensen

Jun 21, 2019, 10:35:52 AM6/21/19
to lavaan
Have you tried setting orthogonal=TRUE?  Method factors are hard to interpret as method factors if they are correlated.  For identification, all the method factors at least need to be orthogonal to the general factor F.

Terrence D. Jorgensen
Assistant Professor, Methods and Statistics
Research Institute for Child Development and Education, the University of Amsterdam
Message has been deleted

Terrence Jorgensen

Jul 5, 2019, 1:46:09 PM7/5/19
to lavaan
The first warning is weird:

Warning messages:

1: In lav_partable_check(lavpartable, categorical = lavoptions$categorical,  :

  lavaan WARNING: parameter table does not contain thresholds

Why would that be?  Check your data to make sure they contain a finite number of categories:

lavTables(MyData, dim = 1)

Maybe the problem is your are passing your data.frame to the ordered= argument, instead of the character vector of names:

ordered = names(MyData)


Jul 12, 2019, 8:10:50 AM7/12/19
to lavaan
Tried several thing, but nothing works, like:

CFA_model_1_order <- cfa(CFA_model_1, data = MyData,
   ordered=c("D1.1", "D1.2", "D1.3", "D1.4", "D1.5", "D1.6",
          "D2.1", "D2.2", "D2.3", "D2.4", "D2.5",
          "D3.1", "D3.2", "D3.3", "D3.4",
          "D4.1", "D4.2", "D4.3", "D4.4", "D4.5",
          "D5.1", "D5.2", "D5.3", "D5.4", "D5.5", "D5.6", "D5.7", "D5.8",
          "D6.1", "D6.2", "D6.3", "D6.4", "D6.5", "D6.6", "D6.7", "D6.8"))

or to make sure the items are not classed as numeric but as ordered factors:
itemsfac <- items %>% mutate_if(is.numeric, as.ordered)
CFA_model_1_order <- cfa(CFA_model_1, data = MyData)

Any suggestions?
Reply all
Reply to author
Message has been deleted
0 new messages