problem with collect, unable to grab number of observations

[Jean-Michel Galarneau]

Hello all, I am having difficulties collecting the number of observations for my models. Any help would be greatly appreciated.

. collect clear

. collect N=_N OR=_r_b SE=_r_se Lower=_r_lb Upper=_r_ub Pvalue=_r_p,:xtgee injury age,fam(bin) link(logit) eform corr(exc) i(teamclu
> ster) exp(exphours) robust

Iteration 1: tolerance = .00925955
Iteration 2: tolerance = .00010796
Iteration 3: tolerance = 1.629e-06
Iteration 4: tolerance = 2.645e-08

GEE population-averaged model                        Number of obs    =  1,775
Group variable: teamcluster                          Number of groups =    129
Family: Binomial                                     Obs per group:  
Link:   Logit                                                     min =      1
Correlation: exchangeable                                         avg =   13.8
                                                                  max =     43
                                                     Wald chi2(1)     =  45.82
Scale parameter = 1                                  Prob > chi2      = 0.0000

                            (Std. err. adjusted for clustering on teamcluster)
------------------------------------------------------------------------------
             |               Robust
      injury | Odds ratio   std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         age |   1.420585    .073675     6.77   0.000     1.283282     1.57258
       _cons |   .0000174   .0000131   -14.48   0.000     3.93e-06    .0000765
ln(exphours) |          1  (exposure)
------------------------------------------------------------------------------
Note: _cons estimates baseline odds (conditional on zero random effects).

. 
. collect dims

Collection dimensions
Collection: default
-----------------------------------------
                   Dimension   No. levels
-----------------------------------------
Layout, style, header, label
                      cmdset   1         
                       coleq   1         
                     colname   45        
               program_class   2         
                      result   50        
                 result_type   3         
                     rowname   43        

Style only
                border_block   4         
                   cell_type   4         
-----------------------------------------

. collect levelsof result,

Collection: default
 Dimension: result
    Levels: Lower N N_g OR Pvalue R SE Upper _r_b _r_ci _r_df _r_lb _r_p _r_se _r_ub _r_z _r_z_abs chi2 chi2_dev chi2_dis chi2type
            cmd cmdline corr depvar deviance df_m df_pear dif dispers estat_cmd family g_avg g_max g_min ivar link marginsnotok
            model offset p phi predict properties rank rc scale tol vce vcetype

. collect label list result, all

  Collection: default
   Dimension: result
       Label: Result
Level labels:
       Lower  
           N  Number of observations
         N_g  Number of groups
          OR  
      Pvalue  
           R  Estimated working correlation matrix
          SE  
       Upper  
        _r_b  Coefficient
       _r_ci  __LEVEL__% CI
       _r_df  df
       _r_lb  __LEVEL__% lower bound
        _r_p  p-value
       _r_se  Std. error
       _r_ub  __LEVEL__% upper bound
        _r_z  z
    _r_z_abs  |z|
        chi2  χ²
    chi2_dev  χ² test of deviance
    chi2_dis  χ² test of deviance dispersion
    chi2type  Type of model χ² test
         cmd  Command
     cmdline  Command line as typed
        corr  Correlation structure
      depvar  Dependent variable
    deviance  Deviance
        df_m  Model DF
     df_pear  DF for Pearson χ²
         dif  Achieved tolerance
     dispers  Deviance dispersion
   estat_cmd  Program used to implement estat
      family  Distribution family
       g_avg  Average group size
       g_max  Largest group size
       g_min  Smallest group size
        ivar  Group variable
        link  Link function
marginsnotok  Predictions disallowed by margins
       model  Model
      offset  Linear offset variable
           p  Model test p-value
         phi  Scale parameter
     predict  Program used to implement predict
  properties  Command properties
        rank  Rank of VCE
          rc  Return code
       scale  Scale parameter
         tol  Target tolerance
         vce  SE method
     vcetype  SE method

. collect layout (colname) (result[N OR SE Lower Upper Pvalue])

Collection: default
      Rows: colname
   Columns: result[N OR SE Lower Upper Pvalue]
   Table 1: 2 x 5

--------------------------------------------------------
          |       OR       SE    Lower    Upper   Pvalue
----------+---------------------------------------------
age       | 1.420585  .073675 1.283282  1.57258 1.29e-11
Intercept | .0000174 .0000131 3.93e-06 .0000765 1.70e-47
--------------------------------------------------------

. collect style cell result, nformat(%6.3f)

. collect preview

------------------------------------------
          |    OR    SE Lower Upper Pvalue
----------+-------------------------------
age       | 1.421 0.074 1.283 1.573  0.000
Intercept | 0.000 0.000 0.000 0.000  0.000
------------------------------------------

.

I don't know what I am doing wrong. I've tried removing the _ before the N but with no success and instead I get the error

N not found

...


Regards,

Jean-Michel Galarneau

[William Lisowski]

I think you want

collect N=e(N) ...

to access the number of observations from the returned estimates e() described by the output of xtgee.

In any event, _N would be the number of observations in your dataset, which need not be the number of observations on which the estimation was run.

[Jean-Michel Galarneau]

William, I'm not seeing how I can access these results and place them in the same table as the one produced by:

collect layout (colname) (result[N OR SE Lower Upper Pvalue])

The code above has worked well for me to gather the OR,SE,CI's from a whole bunch of bivariate regressions into 1 table, I want to do the same thing but also collect the N (estimation sample) used in each estimation and so on. I've spent countless hours on this, read the manual from start to finish and I have to admit that the suite of collect commands is far from intuitive.



J-M