No observation error in Stata

[Guest]

Hello, I am having an issue with running my panel regression. STATA gives a 'no observations' error. I have checked that all values are numeric, not string. I have also checked the data very carefully, ensuring there are no mistakes. There are, however, some missing values in the dataset, but I was advised not to convert them to zero as it has been distorted by results before. Please help, I am lost - why am I seeing this error message and how can I fix it?

[Andrew Musau]

If a variable has all missing values, then you cannot include it in the regression as Stata implements listwise deletion of missing values. Consider this:

webuse grunfeld
gen allmissing=.
xtreg invest mvalue kstock allmissing, fe
*DROP VARIABLE WITH ALL MISSING VALUES
xtreg invest mvalue kstock, fe

Res.:

. xtreg invest mvalue kstock allmissing, fe
no observations
r(2000);

. *DROP VARIABLE WITH ALL MISSING VALUES

. xtreg invest mvalue kstock, fe

Fixed-effects (within) regression               Number of obs     =        200
Group variable: company                         Number of groups  =         10

R-sq:                                           Obs per group:
     within  = 0.7668                                         min =         20
     between = 0.8194                                         avg =       20.0
     overall = 0.8060                                         max =         20

                                                F(2,188)          =     309.01
corr(u_i, Xb)  = -0.1517                        Prob > F          =     0.0000

------------------------------------------------------------------------------
      invest |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      mvalue |   .1101238   .0118567     9.29   0.000     .0867345    .1335131
      kstock |   .3100653   .0173545    17.87   0.000     .2758308    .3442999
       _cons |  -58.74393   12.45369    -4.72   0.000    -83.31086     -34.177
-------------+----------------------------------------------------------------
     sigma_u |  85.732501
     sigma_e |  52.767964
         rho |  .72525012   (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(9, 188) = 49.18                     Prob > F = 0.0000

.

[William Lisowski]

Another common way of losing all your observations is if you use a categorical variable to create a set of indicator (dummy) variables and don't do it correctly.

. * Example generated by -dataex-. For more info, type help dataex

. clear

. input float group

         group
  1. 1
  2. 2
  3. 3
  4. end

. * create indicator variables: this is wrong

. generate a = 1 if group==1
(2 missing values generated)

. generate b = 1 if group==2
(2 missing values generated)

. generate c = 1 if group==3
(2 missing values generated)

. * create indicator variables: this is right

. generate A = group==1

. generate B = group==2

. generate C = group==3

. * see that in every observation 2 of a, b, c are missing

. * see that in every observation none of A, B, C are missing

. list, clean

       group   a   b   c   A   B   C  
  1.       1   1   .   .   1   0   0  
  2.       2   .   1   .   0   1   0  
  3.       3   .   .   1   0   0   1  

. * see what happens

. table a b c
no observations
r(2000);

. table A B C

----------------------------
            |        C      
            |  0   1   Total
------------+---------------
A           |              
  0         |              
    B       |              
      0     |      1       1
      1     |  1           1
      Total |  1   1       2
  1         |              
    B       |              
      0     |  1           1
      Total |  1           1
  Total     |              
    B       |              
      0     |  1   1       2
      1     |  1           1
      Total |  2   1       3
----------------------------

. * this is the best:

. * don't generate indicator variables

. * use factor variable notation in your model

. * see help factor variables

.

[Guest]

Thank you very much for this!

[Nariman Sayed]

what if I have successfully run fixed effect model, however SEM-ML model using XTPDML command gives me error of no observation ?

what can I do in this regard ? I have missing values in some of the dependent and independent variables

thanks in advance !!