what does estimates store store?

[Sehajroop Bath]

I've checked resources online but I'm not sure what all things estimates store stores after running a logistic regression. I need to be able to extract the coef, lci, uci, pval from the stored regression output and I'm not sure how to access those. Thanks in advance!

[Clyde Schechter]

The specific statistics you mention are found in r(table) after an estimation command. But those are not stored with -estimates store-. However, you can still find the coefficients in the pseudo-matrix _b[]. Combining that with the standard errors that are easily accessed in pseudo-matrix _se[], from there you can just calculate the confidence bounds and p-values yourself in the usual way.

To answer your question more directly, whatever shows up in e() after an estimation command is what -estimates store- stores. If you aren't sure what -e()- has after a certain command, run the command and then run -ereturn list- to see.

[Sehajroop Bath]

How do I store the regression output as pseudo-matrix and then retrieve _b[] and _se[] later?

[Wouter Wakker]

You can do that with estimates store and estimates restore, see the following example.

. sysuse auto
(1978 Automobile Data)

. qui reg price mpg weight

. di _b[mpg]
-49.512221

. di _se[mpg]
86.156039

. estimates store M

. 
. qui reg price headroom i.foreign

. capture noisily di _b[mpg]
[mpg] not found

. capture noisily di _se[mpg]
[mpg] not found

. 
. estimates restore M
(results M are active now)

. di _b[mpg]
-49.512221

. di _se[mpg]
86.156039

[William Lisowski]

The model results in r(table) are, as Clyde noted, not included among the "estimation results" that estimates store saves. (The active estimation results it saves are described at the end of the output of, for example, the help logistic command.)

The community contributed estadd command, part of the estout package written byBen Jann and available from SSC (see the output of ssc describe estout) allows you to add a copy of r(table) to the estimates, so it is then stored by estimates store and retrieved by estimates restore. Here's an example.

sysuse auto, clear
regress mpg weight i.foreign
ereturn list
return list
matrix list r(table)
estadd matrix table = r(table)
estimates store gnxl
ereturn clear
estimates restore gnxl
ereturn list
matrix list e(table)

. sysuse auto, clear
(1978 Automobile Data)

. regress mpg weight i.foreign

      Source |       SS           df       MS      Number of obs   =        74
-------------+----------------------------------   F(2, 71)        =     69.75
       Model |   1619.2877         2  809.643849   Prob > F        =    0.0000
    Residual |  824.171761        71   11.608053   R-squared       =    0.6627
-------------+----------------------------------   Adj R-squared   =    0.6532
       Total |  2443.45946        73  33.4720474   Root MSE        =    3.4071

------------------------------------------------------------------------------
         mpg |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      weight |  -.0065879   .0006371   -10.34   0.000    -.0078583   -.0053175
             |
     foreign |
    Foreign  |  -1.650029   1.075994    -1.53   0.130      -3.7955    .4954422
       _cons |    41.6797   2.165547    19.25   0.000     37.36172    45.99768
------------------------------------------------------------------------------

. ereturn list

scalars:
                  e(N) =  74
               e(df_m) =  2
               e(df_r) =  71
                  e(F) =  69.74846262000308
                 e(r2) =  .6627029116028815
               e(rmse) =  3.407059285651584
                e(mss) =  1619.287698167387
                e(rss) =  824.1717612920727
               e(r2_a) =  .6532015851691599
                 e(ll) =  -194.1830643938065
               e(ll_0) =  -234.3943376482347
               e(rank) =  3

macros:
            e(cmdline) : "regress mpg weight i.foreign"
              e(title) : "Linear regression"
          e(marginsok) : "XB default"
                e(vce) : "ols"
             e(depvar) : "mpg"
                e(cmd) : "regress"
         e(properties) : "b V"
            e(predict) : "regres_p"
              e(model) : "ols"
          e(estat_cmd) : "regress_estat"

matrices:
                  e(b) :  1 x 4
                  e(V) :  4 x 4

functions:
             e(sample)   

. return list

scalars:
              r(level) =  95

matrices:
              r(table) :  9 x 4

. matrix list r(table)

r(table)[9,4]
                            0b.          1.            
            weight     foreign     foreign       _cons
     b  -.00658789           0  -1.6500291   41.679702
    se   .00063711           .   1.0759941   2.1655472
     t  -10.340218           .  -1.5334927    19.24673
pvalue   8.283e-16           .    .1295987   6.896e-30
    ll  -.00785825           .  -3.7955004   37.361724
    ul  -.00531752           .   .49544223   45.997681
    df          71          71          71          71
  crit   1.9939434   1.9939434   1.9939434   1.9939434
 eform           0           0           0           0

. estadd matrix table = r(table)

added matrix:
              e(table) :  9 x 4

. estimates store gnxl

. ereturn clear

. estimates restore gnxl
(results gnxl are active now)

. matrix list e(table)

e(table)[9,4]
                            0b.          1.            
            weight     foreign     foreign       _cons
     b  -.00658789           0  -1.6500291   41.679702
    se   .00063711           .   1.0759941   2.1655472
     t  -10.340218           .  -1.5334927    19.24673
pvalue   8.283e-16           .    .1295987   6.896e-30
    ll  -.00785825           .  -3.7955004   37.361724
    ul  -.00531752           .   .49544223   45.997681
    df          71          71          71          71
  crit   1.9939434   1.9939434   1.9939434   1.9939434
 eform           0           0           0           0

.

[daniel klein]

The community contributed estadd command, part of the estout package [...] allows you to add a copy of r(table) to the estimates, so it is then stored by estimates store and retrieved by estimates restore.

Not to take away from estadd but this seems unnecessarily complicated. You can just

estimates replay name_of_stored_results

to bring back* r(table).


* Technically, the p-values and CIs are re-calculated from the coefficients and covariance matrix, but that only takes a couple of milliseconds.