A problem with runmplus_load_savedata.

[Min Zhang]

Dear Statalist,

I am trying to use the runmplus to generate a confirmatory factor analysis and encountered an error with runmplus_load_savedata which I did not understand at all.

The codes ran perfectly. But after some respondents were dropped, an error occurred after runmplus_load_savedata.
The codes remain the same; the only change was that I deleted some respondents. The variables that I used to drop those respondents were not involved in runmplus codes.

. use temp1, clear 

. qui runmplus s007   d018 d022 d057 d059 , idvariable(s007)  ///
>   categorical(d018 d022 d057 d059) /// 
>   ANALYSIS(TYPE = general; ) model(f1 by d018 d022 d057 d059) ///
>   savedata(save=fscores; file=c:\trash\trash.dat) savelogfile(c:\trash\trash)

. runmplus_load_savedata , out(c:/trash/trash.out) clear

The case in which some respondents were dropped

. use temp1, clear 

. keep if s003==50  | x048==356003 | x048==356004 | x048==356008 | x048==356011 | ///
>  x048==356017 
(1303 observations deleted)

. qui runmplus s007   d018 d022 d057 d059 , idvariable(s007)  ///
>   categorical(d018 d022 d057 d059) /// 
>   ANALYSIS(TYPE = general; ) model(f1 by d018 d022 d057 d059) ///
>   savedata(save=fscores; file=c:\trash\trash.dat) savelogfile(c:\trash\trash)

. runmplus_load_savedata , out(c:/trash/trash.out) clear
invalid syntax
r(198);

end of do-file

I have read the help files many times and still have not got a clue.
I appreciate any suggestions on why the error happened and how to fix it.

PS: I am using Stata 13 MP, and the package runmplus is up to date.

[William Lisowski]

I'd guess this is a programming problem in the runmplus package: somehow your data has a combination of characteristics runmplus did not expect, and that caused runmplus to create and execute an incorrect Stata command that had a syntax error. That is, I don't think the "syntax error" is on your part, but rather on the part of the author of runmplus.

The help file for the various runmplus commands gives the author's name and email address; you might try contacting him directly.

[Min Zhang]

Hi Lisowski,

Thank you for your reply. I have reported this issue to the author.

Regards,
Min

[wbuchanan]

Why not fit the model in Stata directly? Also, remember that file paths in Stata are case sensitive, so you may want to make sure the file path is correctly entered including the casing of the letters.

[Wendy Olsen]

Thanks Min Zhang, I advise that MPLUS does have some special characteristics for Categorical variables but if the vars were yes/no, then those would not have much added value over the STATA version. If the variables are Likert Scales then RUNMPLUS will have a special value. Secondly, re failure on the 2nd try, it is like that the Log file of Mplus, which is sitting inside the STATA log file, will show you the 2nd Run didn't work. As such, (it didn't work because of too few cases, did not converge, too few bits of data, or a zero-cell problem), there is no scores file for the 2nd command to seek. If you can widen the range of Manifest Vars to 6 or 7, it will be a stronger factor, converge, and then the 2nd program which has fewer cases will not fail.

[Nathan E. Fosse]

Hi, Min Zhang:

Several thoughts:

(1) Check an earlier version of Stata:

I'd go back to an earlier version of Stata to see if that's the problem; I know the authors have more recently relied on R, so the updates may not be compatible with Stata 13. Runmplus came out in 2010, so that may be part of the issue.

(2) Check if this is a problem within Mplus.

I'd go back to the Mplus interface and the model there. Are you encountering problems there? Also, are you using the latest version of Mplus?

(3) As others have pointed out, why not use Stata?

Alan Acock's book (link) has a good example of CFA in the first chapter, if I recall correctly.

(4) Re: the authors:

I won't speak on behalf the authors formally. However, I've spoken with Rich Jones and know they're using the lavaan package in R as a wrapper for Mplus. If you go to the Mplus website you'll find the R wrapper advertised there. Therefore, some of these issues may be related to compatibility. Also, just FYI, the correspondence of the authors may not be up-to-date.

(5) Re: CFA

Having worked with several stat packages, including Mplus and Latent GOLD, I've found that latent variable modeling using the Latent GOLD software is far faster, more reliable, and more intuitive (I'm a bit biased, however, but this was borne out in the crucible of research!). This url provides information on Latent GOLD 5.1, which just came out; again, depending on other analyses, this may work for you. That said, for more simple analyses, Stata may well suffice, and it's worth it (in my view) to check out updating to Stata 14.

[Min Zhang]

I have received the author's reply email and, as he asked, posted the runmplus results of the first successful run.

. 
.  runmplus s007 d018 d022 d057 d059 , idvariable(s007)  ///
>   categorical(d018 d022 d057 d059) /// 
>   ANALYSIS(TYPE = general; ) model(f1 by d018 d022 d057 d059;) ///
>   savedata(save=fscores; file=c:\trash\temp2.dat) savelogfile(c:\trash\trash)
THE MODEL ESTIMATION TERMINATED NORMALLY
Mplus VERSION 6.12
MUTHEN & MUTHEN
09/08/2015   1:45 PM

INPUT INSTRUCTIONS

  TITLE:
    Variable List -

    s007 :
    d018 :
    d022 :
    d057 :
    d059 :

  DATA:
    FILE = __000001.dat ;
  VARIABLE:
    NAMES =
      s007 d018 d022 d057 d059 ;
    MISSING ARE ALL (-9999) ;
    CATEGORICAL =
      d018
      d022
      d057
      d059
      ;
    IDVARIABLE = s007 ;
  ANALYSIS:
  OUTPUT:
  SAVEDATA:
  save=fscores ;
  file=c:\trash\temp2.dat  ;

  MODEL:
  f1 by d018 d022 d057 d059 ;




INPUT READING TERMINATED NORMALLY




Variable List -

s007 :
d018 :
d022 :
d057 :
d059 :

SUMMARY OF ANALYSIS

Number of groups                                                 1
Number of observations                                        3461

Number of dependent variables                                    4
Number of independent variables                                  0
Number of continuous latent variables                            1

Observed dependent variables

  Binary and ordered categorical (ordinal)
   D018        D022        D057        D059

Continuous latent variables
   F1

Variables with special functions

  ID variable           S007

Estimator                                                    WLSMV
Maximum number of iterations                                  1000
Convergence criterion                                    0.500D-04
Maximum number of steepest descent iterations                   20
Maximum number of iterations for H1                           2000
Convergence criterion for H1                             0.100D-03
Parameterization                                             DELTA

Input data file(s)
  __000001.dat

Input data format  FREE


SUMMARY OF DATA

     Number of missing data patterns            15


COVARIANCE COVERAGE OF DATA

Minimum covariance coverage value   0.100


     PROPORTION OF DATA PRESENT


           Covariance Coverage
              D018          D022          D057          D059
              ________      ________      ________      ________
 D018           0.984
 D022           0.905         0.917
 D057           0.851         0.812         0.863
 D059           0.901         0.858         0.834         0.914


UNIVARIATE PROPORTIONS AND COUNTS FOR CATEGORICAL VARIABLES

    D018
      Category 1    0.062      210.000
      Category 2    0.938     3194.000
    D022
      Category 1    0.866     2750.000
      Category 2    0.134      425.000
    D057
      Category 1    0.183      546.000
      Category 2    0.288      862.000
      Category 3    0.359     1074.000
      Category 4    0.169      506.000
    D059
      Category 1    0.229      723.000
      Category 2    0.422     1333.000
      Category 3    0.275      869.000
      Category 4    0.075      237.000



THE MODEL ESTIMATION TERMINATED NORMALLY



MODEL FIT INFORMATION

Number of Free Parameters                       12

Chi-Square Test of Model Fit

          Value                             38.422*
          Degrees of Freedom                     2
          P-Value                           0.0000

*   The chi-square value for MLM, MLMV, MLR, ULSMV, WLSM and WLSMV cannot be used
    for chi-square difference testing in the regular way.  MLM, MLR and WLSM
    chi-square difference testing is described on the Mplus website.  MLMV, WLSMV,
    and ULSMV difference testing is done using the DIFFTEST option.

RMSEA (Root Mean Square Error Of Approximation)

          Estimate                           0.073
          90 Percent C.I.                    0.054  0.093
          Probability RMSEA <= .05           0.026

CFI/TLI

          CFI                                0.761
          TLI                                0.282

Chi-Square Test of Model Fit for the Baseline Model

          Value                            158.215
          Degrees of Freedom                     6
          P-Value                           0.0000

WRMR (Weighted Root Mean Square Residual)

          Value                              1.629



MODEL RESULTS

                                                    Two-Tailed
                    Estimate       S.E.  Est./S.E.    P-Value

 F1       BY
    D018               1.000      0.000    999.000    999.000
    D022              -0.659      0.151     -4.361      0.000
    D057               2.166      0.857      2.528      0.011
    D059               0.471      0.101      4.657      0.000

 Thresholds
    D018$1            -1.541      0.034    -45.485      0.000
    D022$1             1.108      0.028     39.590      0.000
    D057$1            -0.905      0.027    -33.909      0.000
    D057$2            -0.072      0.023     -3.147      0.002
    D057$3             0.957      0.027     35.199      0.000
    D059$1            -0.743      0.025    -30.121      0.000
    D059$2             0.386      0.023     16.851      0.000
    D059$3             1.440      0.033     43.505      0.000

 Variances
    F1                 0.120      0.056      2.142      0.032


R-SQUARE

    Observed                   Residual
    Variable        Estimate   Variance

    D018               0.120      0.880
    D022               0.052      0.948
    D057               0.564      0.436
    D059               0.027      0.973


QUALITY OF NUMERICAL RESULTS

     Condition Number for the Information Matrix              0.159E-03
       (ratio of smallest to largest eigenvalue)


SAMPLE STATISTICS FOR ESTIMATED FACTOR SCORES


     SAMPLE STATISTICS


           Means
              F1
              ________
 1             -0.001


           Covariances
              F1
              ________
 F1             0.051


           Correlations
              F1
              ________
 F1             1.000


SAVEDATA INFORMATION

  Order and format of variables

    D018           F10.3
    D022           F10.3
    D057           F10.3
    D059           F10.3
    S007           I7
    F1             F10.3

  Save file
    c:\trash\temp2.dat

  Save file format
    4F10.3 I7 F10.3

  Save file record length    5000


     Beginning Time:  13:45:22
        Ending Time:  13:45:22
       Elapsed Time:  00:00:00



MUTHEN & MUTHEN
3463 Stoner Ave.
Los Angeles, CA  90066

Tel: (310) 391-9971
Fax: (310) 391-8971
Web: www.StatModel.com
Support: [email protected]

Copyright (c) 1998-2011 Muthen & Muthen

. runmplus_load_savedata , out(c:/trash/trash.out) clear

[Rich Jones]

Hi Min,

Your issue is a mystery to me. Based on your MPlus log I can runmplus_load_savedata no problem with a dummy data set (log below).

I would like to help you and figure out what is going wrong so we can put a fix in runmplus_load_savedata if it is needed.

Maybe you could "log using foo.log, text replace" and "set trace on" and the run your "runmplus_load_savedata" and we could see where it's going wrong.

Some other things:

I see you are using MPlus 6.12, which is out-of-date. I keep runmplus current with MPlus. And when Mplus changes the structure and content of output files, the post-processing that is baked into runmplus and associated files may bomb out. So, you should keep your MPlus current. I don't think that's the issue, since I was able to read your OUT and read the associated DAT file. But...

Also, I keep the version of runmplus and associated files on my AWS file server more up-to-date than the SSC, which although is a little easier for me to keep updated. Instructions are in the runmplus help file for getting your update from there.


Rich

PS Hi Nathan.

PPS Runmplus and associated files should work with current versions of associated software (version 14 Stata, Version 7.31 Mplus), and on MAC/OS, other *NIX, and PC/Windows environments. If it does not please let me know.

. clear

. type c:\trash\temp2.dat
12345678 22345678 32345678 42345678 1 5345678

. type c:\trash\trash.out
THE MODEL ESTIMATION TERMINATED NORMALLY
Mplus VERSION 6.12
MUTHEN & MUTHEN
09/08/2015 1:45 PM

INPUT INSTRUCTIONS

TITLE:
Variable List -

s007 :
d018 :
d022 :
d057 :
d059 :

DATA:
FILE = __000001.dat ;
VARIABLE:
NAMES =
s007 d018 d022 d057 d059 ;
MISSING ARE ALL (-9999) ;
CATEGORICAL =
d018
d022
d057
d059
;
IDVARIABLE = s007 ;
ANALYSIS:
OUTPUT:
SAVEDATA:
save=fscores ;
file=c:\trash\temp2.dat ;

MODEL:
f1 by d018 d022 d057 d059 ;




INPUT READING TERMINATED NORMALLY




Variable List -

s007 :
d018 :
d022 :
d057 :
d059 :

SUMMARY OF ANALYSIS

Number of groups 1
Number of observations 3461

Number of dependent variables 4
Number of independent variables 0
Number of continuous latent variables 1

Observed dependent variables

Binary and ordered categorical (ordinal)
D018 D022 D057 D059

Continuous latent variables
F1

Variables with special functions

ID variable S007

Estimator WLSMV
Maximum number of iterations 1000
Convergence criterion 0.500D-04
Maximum number of steepest descent iterations 20
Maximum number of iterations for H1 2000
Convergence criterion for H1 0.100D-03
Parameterization DELTA

Input data file(s)
__000001.dat

Input data format FREE


SUMMARY OF DATA

Number of missing data patterns 15


COVARIANCE COVERAGE OF DATA

Minimum covariance coverage value 0.100


PROPORTION OF DATA PRESENT


Covariance Coverage
D018 D022 D057 D059
________ ________ ________ ________
D018 0.984
D022 0.905 0.917
D057 0.851 0.812 0.863
D059 0.901 0.858 0.834 0.914


UNIVARIATE PROPORTIONS AND COUNTS FOR CATEGORICAL VARIABLES

D018
Category 1 0.062 210.000
Category 2 0.938 3194.000
D022
Category 1 0.866 2750.000
Category 2 0.134 425.000
D057
Category 1 0.183 546.000
Category 2 0.288 862.000
Category 3 0.359 1074.000
Category 4 0.169 506.000
D059
Category 1 0.229 723.000
Category 2 0.422 1333.000
Category 3 0.275 869.000
Category 4 0.075 237.000



THE MODEL ESTIMATION TERMINATED NORMALLY



MODEL FIT INFORMATION

Number of Free Parameters 12

Chi-Square Test of Model Fit

Value 38.422*
Degrees of Freedom 2
P-Value 0.0000

* The chi-square value for MLM, MLMV, MLR, ULSMV, WLSM and WLSMV cannot be used
for chi-square difference testing in the regular way. MLM, MLR and WLSM
chi-square difference testing is described on the Mplus website. MLMV, WLSMV,
and ULSMV difference testing is done using the DIFFTEST option.

RMSEA (Root Mean Square Error Of Approximation)

Estimate 0.073
90 Percent C.I. 0.054 0.093
Probability RMSEA <= .05 0.026

CFI/TLI

CFI 0.761
TLI 0.282

Chi-Square Test of Model Fit for the Baseline Model

Value 158.215
Degrees of Freedom 6
P-Value 0.0000

WRMR (Weighted Root Mean Square Residual)

Value 1.629



MODEL RESULTS

Two-Tailed
Estimate S.E. Est./S.E. P-Value

F1 BY
D018 1.000 0.000 999.000 999.000
D022 -0.659 0.151 -4.361 0.000
D057 2.166 0.857 2.528 0.011
D059 0.471 0.101 4.657 0.000

Thresholds
D018$1 -1.541 0.034 -45.485 0.000
D022$1 1.108 0.028 39.590 0.000
D057$1 -0.905 0.027 -33.909 0.000
D057$2 -0.072 0.023 -3.147 0.002
D057$3 0.957 0.027 35.199 0.000
D059$1 -0.743 0.025 -30.121 0.000
D059$2 0.386 0.023 16.851 0.000
D059$3 1.440 0.033 43.505 0.000

Variances
F1 0.120 0.056 2.142 0.032


R-SQUARE

Observed Residual
Variable Estimate Variance

D018 0.120 0.880
D022 0.052 0.948
D057 0.564 0.436
D059 0.027 0.973


QUALITY OF NUMERICAL RESULTS

Condition Number for the Information Matrix 0.159E-03
(ratio of smallest to largest eigenvalue)


SAMPLE STATISTICS FOR ESTIMATED FACTOR SCORES


SAMPLE STATISTICS


Means
F1
________
1 -0.001


Covariances
F1
________
F1 0.051


Correlations
F1
________
F1 1.000


SAVEDATA INFORMATION

Order and format of variables

D018 F10.3
D022 F10.3
D057 F10.3
D059 F10.3
S007 I7
F1 F10.3

Save file
c:\trash\temp2.dat

Save file format
4F10.3 I7 F10.3

Save file record length 5000


Beginning Time: 13:45:22
Ending Time: 13:45:22
Elapsed Time: 00:00:00



MUTHEN & MUTHEN
3463 Stoner Ave.
Los Angeles, CA 90066

Tel: (310) 391-9971
Fax: (310) 391-8971
Web: www.StatModel.com
Support: [email protected]

Copyright (c) 1998-2011 Muthen & Muthen

. runmplus_load_savedata , out(c:/trash/trash.out) clear

. list

+------------------------------------------------------------+
| d018 d022 d057 d059 s007 f1 |
|------------------------------------------------------------|
1. | 12345678 22345678 32345678 42345678 1 5345678 |
+------------------------------------------------------------+