Crossed random effects model-error r(3203)

[Matthew Singer]

I have been running crossed random effects mixed models and the syntax that has always worked before (but which I had not used since January) has stopped working for me. The data set involves survey data with countries and years cross-nested. The basic model specification I have used is (here without any predictors for simplicity):

mixed dem_sat || _all: R.year || country:

But where this command was working for me in January, now I get the following output:

MapRowsToIntegers(): 3203 <tmp>[1,306293] found where colvector required
GenGroupSampleNumeric(): - function returned error
<istmt>: - function returned error
r(3203);

That happens no matter what dependent variable I use. I am running stata 16.1.

[Carlo Lazzaro]

Matthew:
is your copy of Stata fully updated?

[Matthew Singer]

yes it is

[Carlo Lazzaro]

Matthew:
unfortunately, I cannot replicate your problem with the following toy-example:

. use https://www.stata-press.com/data/r16/productivity
(Public Capital Productivity)

. mixed gsp  || region: || state:

Performing EM optimization:

Performing gradient-based optimization:

Iteration 0:   log likelihood =  200.86162 
Iteration 1:   log likelihood =  200.86162 

Computing standard errors:

Mixed-effects ML regression                     Number of obs     =        816

-------------------------------------------------------------
                |     No. of       Observations per Group
 Group Variable |     Groups    Minimum    Average    Maximum
----------------+--------------------------------------------
         region |          9         51       90.7        136
          state |         48         17       17.0         17
-------------------------------------------------------------

                                                Wald chi2(0)      =          .
Log likelihood =  200.86162                     Prob > chi2       =          .

------------------------------------------------------------------------------
         gsp |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
       _cons |   10.65961   .2503806    42.57   0.000     10.16887    11.15035
------------------------------------------------------------------------------

------------------------------------------------------------------------------
  Random-effects Parameters  |   Estimate   Std. Err.     [95% Conf. Interval]
-----------------------------+------------------------------------------------
region: Identity             |
                  var(_cons) |   .4376123   .2697616      .1307299    1.464887
-----------------------------+------------------------------------------------
state: Identity              |
                  var(_cons) |   .6080626   .1382733      .3893904    .9495358
-----------------------------+------------------------------------------------
               var(Residual) |   .0246633   .0012586      .0223159    .0272577
------------------------------------------------------------------------------
LR test vs. linear model: chi2(2) = 2750.56               Prob > chi2 = 0.0000

Note: LR test is conservative and provided only for reference.

.

Therefore, I would:

-re-run your model with two-level and then three-level design and see when Stata gasps;
-contact Stata Tech support if what above does not give you any clue about what went wrong along the way.

[Amal Alfawzan]

Hi
On my phd I have an argument states that gender is a predicted variable for the relationship between innovation and religion. Men is more likely to have higher effect than women. The data showed me a significant difference between genders, however, the difference is very small regarding the coefficient only 0.05. My supervisor is not happy with this result. My questions are , 1-what is the acceptable coefficient difference between tow groups 2- are there any methods that can boost the difference between genders? I am using STATA 2015
thanks

[Carlo Lazzaro]

Duplicate post.
See https://www.statalist.org/forums/for...tow-groups-ols

[Stefano Amato]

Hi Matthew, did you solve the problem? I have the same issue after running mixed models (the last time was months ago).
Best,
Stefano

[Mads Moring]

I have the same problem when trying to run a cross-classified multilevel model.

[Mads Moring]

I have the same problem when trying to run a cross-classified multilevel model. Carlo Lazzaro, I believe that the reason you don't get the error is perhaps because you dont specify "all_: r.region"

I have longitudinal data with students nested in schools. However students change schools during the time spand, so the data is cross-classified.
I get the error when running a model like this one:

 mixed score i.gender##i.time || _all: r.school || student_id:, covar(unstruct)

[Carlo Lazzaro]

Mads:
thanks for pointing this out.
However, even specifying the -_all:- option, I cannot replicate the error:

. use https://www.stata-press.com/data/r16/productivity
(Public Capital Productivity)

.  mixed gsp  || _all: R.region|| state:

Performing EM optimization: 

Performing gradient-based optimization: 

Iteration 0:   log likelihood =  200.86162  
Iteration 1:   log likelihood =  200.86162  

Computing standard errors:

Mixed-effects ML regression                     Number of obs     =        816

-------------------------------------------------------------
                |     No. of       Observations per Group
 Group Variable |     Groups    Minimum    Average    Maximum
----------------+--------------------------------------------
           _all |          1        816      816.0        816
          state |         48         17       17.0         17
-------------------------------------------------------------

                                                Wald chi2(0)      =          .
Log likelihood =  200.86162                     Prob > chi2       =          .

------------------------------------------------------------------------------
         gsp |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
       _cons |   10.65961   .2503806    42.57   0.000     10.16887    11.15035
------------------------------------------------------------------------------

------------------------------------------------------------------------------
  Random-effects Parameters  |   Estimate   Std. Err.     [95% Conf. Interval]
-----------------------------+------------------------------------------------
_all: Identity               |
               var(R.region) |   .4376123   .2697616        .13073    1.464886
-----------------------------+------------------------------------------------
state: Identity              |
                  var(_cons) |   .6080626    .138273      .3893908    .9495349
-----------------------------+------------------------------------------------
               var(Residual) |   .0246633   .0012586      .0223159    .0272577
------------------------------------------------------------------------------
LR test vs. linear model: chi2(2) = 2750.56               Prob > chi2 = 0.0000

Note: LR test is conservative and provided only for reference.

.