CFA in R: The variance-covariance matrix of the estimated parameters (vcov) does not appear to be positive definite. What is the problem?

Asked Mar 29 '23 at 09:14

Active Mar 29 '23 at 09:14

Viewed 25 times

I am trying to calculate a CFA for the Theory of Planned Behaviour (TPB). I have split the 4 factors (Attitude, SubNorm, PBC, Intention) into inverted and non-inverted items. Enclosed is the measurement model.

N = 75. The items were rated on a scale of 1-7.

enter image description here

However, after I run the second part (categorical part), I receive a warning message that says:

enter image description here

Do any of you know the problem and what can be done?

asked Mar 29 '23 at 09:14

Paul

That message means there is multicollinearity among the parameter estimates, which can happen when you try to estimate too many parameters relative to the sample size. I already wouldn't recommend fitting such a large model to continuous indicators. Treating them as categorical would only compound the problem (there are 6 thresholds per item). – Terrence Mar 30 '23 at 11:26

CFA in R: The variance-covariance matrix of the estimated parameters (vcov) does not appear to be positive definite. What is the problem?

0 Answers0