A negative error variance typically indicates model misspecification (e.g., too many or not enough factors). I would examine model misfit diagnostics such as standardized residuals. Another potential factor is samplesize (smaller samples are more prone to improper solutions).

Is it possible to fix the residual variance to .01 in an ESEM model?

If yes, could you provide the proper syntax?

I have tried adding to my model the following statement:

[email protected];

However, such stattemet did not work. Any suggestions?

I don't think this is an appropriate fix. A negative error variance usually signals a more profound problem with the model and/or the measures. Simply constraining the parameter to zero (or to a small non-zero value) is not going to address this.

Thanks for your kindly reply and for the paper you recommended. Could it be useful to identify the confidence interval of the negative error variance to verify whether it contains zero? Christian Geiser

• Constrain or fix the variance to a small positive value.

• Check model specification—too many factors or large cross-loadings can cause issues.

• Re-express or remove problematic indicators if they have near-perfect loadings.

• In Bayesian models, use informative priors and check posterior distributions.

• Increase sample size or iterations if needed.

• Always report and justify any adjustments made.