st: xtmixed - "Missing Values Encoutered"

[Christian Wei� <C.Weiss@plusser-portal.de>]

Hello!


I am currently trying to "bug fix" a mixed model which unfortunately did not converge. Having eliminated some relatively highly correlated variables the following message was displayed : could not calculate numerical derivatives
missing values encountered (complete output see below)

What is ment by "missing values" in this case? Do you have an suggestion how to tackle this particular problem?

Thank you very much in advance!
Best regards,
Christian




Performing EM optimization:

Performing gradient-based optimization:

Iteration 0:   log restricted-likelihood = -8380.4495  (not concave)
numerical derivatives are approximate
flat or discontinuous region encountered
Iteration 1:   log restricted-likelihood = -8380.1053  (not concave)
numerical derivatives are approximate
flat or discontinuous region encountered
Iteration 2:   log restricted-likelihood = -8380.1053  (not concave)
numerical derivatives are approximate
flat or discontinuous region encountered
Iteration 3:   log restricted-likelihood = -8379.8833  (not concave)
numerical derivatives are approximate
flat or discontinuous region encountered
Iteration 4:   log restricted-likelihood = -8379.8833  (not concave)
numerical derivatives are approximate
flat or discontinuous region encountered
Iteration 5:   log restricted-likelihood = -8379.7179  (not concave)
numerical derivatives are approximate
flat or discontinuous region encountered
Iteration 6:   log restricted-likelihood = -8379.5492  (not concave)
numerical derivatives are approximate
flat or discontinuous region encountered
Iteration 7:   log restricted-likelihood = -8379.5402  (not concave)
numerical derivatives are approximate
flat or discontinuous region encountered
Iteration 8:   log restricted-likelihood = -8379.5392  (not concave)
numerical derivatives are approximate
flat or discontinuous region encountered
Iteration 9:   log restricted-likelihood = -8379.5178  (not concave)
numerical derivatives are approximate
flat or discontinuous region encountered
Iteration 10:  log restricted-likelihood = -8379.5178  (not concave)
numerical derivatives are approximate
flat or discontinuous region encountered
Iteration 11:  log restricted-likelihood = -8379.4144  (not concave)
numerical derivatives are approximate
flat or discontinuous region encountered
Iteration 12:  log restricted-likelihood = -8379.4144  (not concave)
numerical derivatives are approximate
flat or discontinuous region encountered
Iteration 13:  log restricted-likelihood = -8379.3935  (not concave)
numerical derivatives are approximate
nearby values are missing
Iteration 14:  log restricted-likelihood = -8379.3908  (not concave)
numerical derivatives are approximate
nearby values are missing
Iteration 15:  log restricted-likelihood = -8379.3895  (not concave)
numerical derivatives are approximate
nearby values are missing
Iteration 16:  log restricted-likelihood =  -8379.389  (not concave)
numerical derivatives are approximate
nearby values are missing
Iteration 17:  log restricted-likelihood =  -8379.389  (not concave)
could not calculate numerical derivatives
missing values encountered

Mixed-effects REML regression                   Number of obs      =      1829

-----------------------------------------------------------
                |   No. of       Observations per Group
 Group Variable |   Groups    Minimum    Average    Maximum
----------------+------------------------------------------
           _all |        1       1829     1829.0       1829
        country |       11         82      166.3        265
   prima~1digit |       88          1       20.8        142
-----------------------------------------------------------

                                                Wald chi2(0)       =         .
Log restricted-likelihood = -8380.4495          Prob > chi2        =         .

------------------------------------------------------------------------------
         bh5 |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
       _cons |   41.70724   8.873858     4.70   0.000     24.31479    59.09968
------------------------------------------------------------------------------

------------------------------------------------------------------------------
  Random-effects Parameters  |   Estimate   Std. Err.     [95% Conf. Interval]
-----------------------------+------------------------------------------------
_all: Independent            |
                    sd(beta) |   9.273805          .             .           .
                sd(logtot~g) |   2.883647          .             .           .
                sd(indust~n) |   1.348894          .             .           .
                sd(divers~n) |     .88744          .             .           .
                sd(netinc~r) |    .001017          .             .           .
                sd(countr~n) |   .0195022          .             .           .
                     sd(uai) |   .4257892          .             .           .
                sd(stockm~v) |   .0993575          .             .           .
                   sd(_cons) |   1.443472          .             .           .
-----------------------------+------------------------------------------------
country: Identity            |
                   sd(_cons) |   7.083008          .             .           .
-----------------------------+------------------------------------------------
prima~1digit: Identity       |
                   sd(_cons) |    4.61776          .             .           .
-----------------------------+------------------------------------------------
                sd(Residual) |   23.14977          .             .           .
------------------------------------------------------------------------------
LR test vs. linear regression:      chi2(11) =   533.23   Prob > chi2 = 0.0000

Note: LR test is conservative and provided only for reference.
Warning: convergence not achieved; estimates are based on iterated EM

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/