I’m stuck with a CFA model in lavaan that won’t converge. It’s for the Social Dominance Orientation scale. The model has four factors: dominance, antiegalitarianism, pro-trait, and con-trait. Items are supposed to load on multiple factors.
Here’s what I’ve got:
model_SDO <- '
Hierarchy =~ Q1 + Q2 + Q3 + Q4 + Q5 + Q6 + Q7 + Q8
Equality =~ Q9 + Q10 + Q11 + Q12 + Q13 + Q14 + Q15 + Q16
Positive =~ Q1 + Q2 + Q3 + Q4 + Q9 + Q10 + Q11 + Q12
Negative =~ Q5 + Q6 + Q7 + Q8 + Q13 + Q14 + Q15 + Q16
'
fit_result <- cfa(model_SDO, data = sdo_data)
But it’s not working. The error says no solution found after tons of iterations. I’m not sure what’s wrong or how to fix it. Any ideas on getting this to work while keeping the cross-loading structure? Thanks!
Hey Nate_45Guitar, sounds like you’re diving deep into SDO modeling! That’s pretty fascinating stuff. 
Have you considered the possibility that your model might be overidentified? With all those cross-loadings, it’s like trying to solve a Rubik’s cube blindfolded! Maybe we could brainstorm some ways to simplify without losing the essence of what you’re after?
I’m curious - what made you choose this specific cross-loading structure? Is it based on previous research or theoretical grounds? Sometimes, theory and practice don’t play nice in CFA land.
Oh, and here’s a wild thought - have you tried running an exploratory factor analysis first? It might give you some clues about how your items are actually clustering. Could be eye-opening!
Anyway, don’t lose heart. CFA can be a real head-scratcher sometimes. Keep us posted on how it goes, yeah? I’m genuinely intrigued to see how you crack this nut!
I’ve encountered similar issues with complex CFA models. One approach that’s helped me is to start with a simpler model and gradually add complexity. Try running separate CFAs for each factor first, then combine them incrementally. This can help identify which specific cross-loadings are causing problems.
Another suggestion is to examine your correlation matrix. If you have very high correlations between factors, it could indicate multicollinearity issues. In that case, you might need to reconsider your factor structure or use a bifactor model instead.
Lastly, consider using a different estimator like WLSMV if you haven’t already, especially if your data is ordinal. It can sometimes handle complex models better than ML estimation. Hope this helps you troubleshoot the convergence problem!
Yo mate, convergence issues r a pain! Try tweaking ur model specs - maybe simplify it a bit? Cud try stepwise approach, start w/ basic model n add complexity. Also, check ur data 4 outliers or weirdness. Might need 2 play w/ estimation methods too. good luck!