Second-order factor analysis

[Andrea Toledo]

Dear StataListers, I am working on a project on which I am trying to estimate a factor analysis using 9 dummies variables.
On a first step I estimated a first order confirmatory factor analysis, and I didn't have problems but when I tries to estimate
a second order confirmatory factor analysis, the model runs and runs and runs and gives me the message that iterations are "not concave".
Because of all the variables have two categories, I use the method adf (asympotically distribution free),
but I also tried to estimate the model using the method (ml) combined with option vce (robust) and with vce (bootstrap). I have the same problem

My question is the following: what does this "not concave" message mean? What should I do?

My code:

(1) FIRST ORDER CONFIRMATORY FACTOR ANALYSIS

sem (Ayudaeconomica -> coberturamedica ITICBT ayudaexterna ,)
(Zonaubicacion -> basural zonainundable villaemergencia,)
(Infraestructura -> baño pisos techointerior hacinamiento),
latent (Ayudaeconomica Zonaubicacion Infraestructura) method(adf)

(2) SECOND ORDER CONFIRMATORY FACTOR ANALYSIS

sem (Incapacidadeconomica -> coberturamedica ITICBT ayudaexterna ,)
(Zonaubicacion -> basural zonainundable villaemergencia,)
(Infraestructura -> baño pisos techointerior hacinamiento,)
(Pobrezamultidimensional -> Infraestructura Incapacidadeconomica Zonaubicacion ),
latent (Pobrezamultidimensional Incapacidadeconomica Zonaubicacion Infraestructura)
method(adf) iterate ( 100 )

RESULTS
Endogenous variables

Measurement: coberturamedica ITICBT ayudaexterna basural zonainundable villaemergencia baño pisos techointerior hacinamiento
Latent: Incapacidadeconomica Zonaubicacion Infraestructura

Exogenous variables
Latent: Pobrezamultidimensional

Fitting baseline model:

Iteration 0: discrepancy = .43060013
Iteration 1: discrepancy = .28040098
Iteration 2: discrepancy = .28040098

Fitting target model:
Iteration 0: discrepancy = 11.835502 (not concave)
Iteration 1: discrepancy = 1.3103552 (not concave)
Iteration 2: discrepancy = .07200014 (not concave)
Iteration 3: discrepancy = .06590574 (not concave)
Iteration 4: discrepancy = .06213892 (not concave)
Iteration 5: discrepancy = .06168234 (not concave)
Iteration 6: discrepancy = .06140415 (not concave)
Iteration 7: discrepancy = .06100357 (not concave)
Iteration 8: discrepancy = .06040023 (not concave)
Iteration 9: discrepancy = .06021822 (not concave)
Iteration 10: discrepancy = .05955194 (not concave)
Iteration 11: discrepancy = .0594336 (not concave)
Iteration 12: discrepancy = .05862482 (not concave)
Iteration 13: discrepancy = .0575207 (not concave)
Iteration 14: discrepancy = .05601082 (not concave)
Iteration 15: discrepancy = .05516756 (not concave)
Iteration 16: discrepancy = .05490172 (not concave)
Iteration 17: discrepancy = .05486508 (not concave)
Iteration 18: discrepancy = .05484651 (not concave)
Iteration 19: discrepancy = .05484535 (not concave)
Iteration 20: discrepancy = .05484476 (not concave)
Iteration 21: discrepancy = .05484462 (not concave)
.
.
.
.
Iteration 98: discrepancy = .05484455 (not concave)
Iteration 99: discrepancy = .05484455 (not concave)
Iteration 100: discrepancy = .05484455 (not concave)
convergence not achieved

Structural equation model

Number of obs = 15,254
Estimation method = adf
Discrepancy = .05484455

[William Lisowski]

Welcome to Statalist.

You have accidentally posted your topic in Statalist's Mata Forum, which is used for discussions of Stata's Mata language, which is different than Stata's command language. Your question will see a more appropriate, and much larger audience if you post it in Statalist's General Forum.

Also, if you have not already done so, take a look at the Statalist FAQ linked to at the top of this page for posting guidelines and suggestions.

[Andrea Toledo]

Thank you so much