Message/Author 


Hello, I am running a multiple group two factor CFA with four and 24 continuous indicators for each factor (many variables are quite nonnormal). When I ran the CFA, I got the following error message: THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES COULD NOT BE COMPUTED. THE MODEL MAY NOT BE IDENTIFIED. CHECK YOUR MODEL. PROBLEM INVOLVING PARAMETER 62. Parameter 62 is the factor correlation. In order to deal with this identification problem, I tried the bootstrapping option with 1000 iterations. The output looked fine, giving me model fit indices, etc. However, the standard errors estimated were all given as 999.000. Are these standard errors due to the nonnormality of the data, or do they indicate some other problem? Are the model fit indices still trustworthy so that I can use them for group invariance tests? Is there perhaps a better way to deal with this identification issue? Thanks a lot for your reply. Here is the code I have used: Analysis: estimator=ml; bootstrap=100; MODEL: F1 BY DW1 DW2 WCt1  WCt6 Rt1  Rt13 Lt1  Lt5 ; F2 BY St1 St2 St3 S14; F1 WITH F2 ; Model new: F1 BY DW1 DW2 WCt1  WCt6 Rt1  Rt13 Lt1  Lt5 ; F2 BY St1 St2 St3 S14; 


The identification problem is caused by the fact that you are freeing the first factor loading in the group NEW. Change MODEL NEW to Model new: F1 BY DW2 WCt1  WCt6 Rt1  Rt13 Lt1  Lt5 ; F2 BY St2 St3 S14; 


Thanks so much for the quick response. 

yezi posted on Wednesday, June 15, 2011  12:35 am



Hello, As is well known, when we use bootstrap method to generate bootstrap samples, the model parameter estimates of some samples may be improper. Are the standard errors of the model parameter estimates which mplus gives based on all bootstrap samples or the bootstrap samples that have the proper solution? For example, if I set bootstrap=1000(generating 1000 bootstrap samples) and the model parameter estimates of 20 bootstrap samples are improper, are the standard errors of the model parameter estimates that mplus gives are based on 1000 bootstrap samples or 980 bootstrap samples£¨only having proper bootstrap samples) £¿ Thank you for your time¡£ 


Mplus will use all replications that converge. The number of converging replications can be found in the output: Number of bootstrap draws Requested 1000 Completed 980 

yezi posted on Wednesday, June 15, 2011  10:12 pm



Hello, Thank you. May I undestand that the standard errors of the model parameter estimates that mplus gives are based on 1000 bootstrap samples and not 980 bootstrap samples? 

yezi posted on Wednesday, June 15, 2011  10:29 pm



Hello, In other word, standard errors of the model parameter estimates that mplus gives are based on 1000 bootstrap samples and not 980 bootstrap samples. Is it(that I understand above) right? Thank you. 


No, they are based on the 980 samples that converged. 

yezi posted on Thursday, June 16, 2011  12:08 am



Thank you. DO the model of bootstrap samples that converged all fit well when doing CFA? 

yezi posted on Thursday, June 16, 2011  12:15 am



In other word, whether mplus checks the fit index of the model of bootstrap samples that converged well when doing CFA or SEM. Thank you. 


Fit is not checked during bootstrapping. 

yezi posted on Thursday, June 16, 2011  10:08 pm



Thank you. Can the bootstrap samples that mplus generates save as external file? 

yezi posted on Thursday, June 16, 2011  10:14 pm



Can the bootstrap samples that mplus generates save as other file during bootstrapping? Or Mplus only gives the standard errors of the model parameter estimates during bootstrapping? Thank you. 


The bootstrapped samples cannot be saved. They are used to create the bootstrapped standard errors. 

yezi posted on Thursday, June 16, 2011  11:37 pm



Thank you very much. 

Guillermo posted on Saturday, December 21, 2013  5:37 pm



Dear Linda, Is there any way to bootstrap the standard error of a deltaRsquare when fitting a latent variable interaction model? Thank you very much. 


Using the BOOTSTRAP option, if you express a deltaRsquare in MODEL CONSTRAINT, you will get Delta method standard errors using the bootstrap standard errors of the parameters estimates. 

Guillermo posted on Sunday, December 22, 2013  11:47 am



Excuse me, Linda. I'm a little bit lost. I've fitted a latent variable interaction model. Then, given that the STANDARDIZED option of the OUTPUT command is not available with TYPE=RANDOM, I've computed by hand the corresponding Rsquare following the steps indicated in Bengt's paper. Next, I obtain the deltaRsquare by subtracting the obtained Rsquare to the Rsquare previously obtained with the model without the interaction term. Now, how can I express this deltaRsquare in MODEL CONSTRAINT? And how do I obtain the bootstrap standard error? Thank you for your time, Linda. I really appreciate your help. 


It sounds like your deltaRsquare is obtained from two different runs, that is, two different models. In that case, you can't express it in Model Constraint. I don't see a way to get SEs for it in an Mplus run. 

Guillermo posted on Sunday, December 22, 2013  8:41 pm



Thank you so much anyway. 

Back to top 