[Carlo Lazzaro]

Paolo:
the two codes are different:

. use https://www.stata-press.com/data/r17/union. . use https://www.stata-press.com/data/r17/union
(NLS Women 14-24 in 1968)

. xtlogit union age grade not_smsa i.south##c.year, fe vce(bootstrap, reps(200) seed(12345) dots(1))
(running xtlogit on estimation sample)

Bootstrap replications (200)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
..................................................    50
..................................................   100
..................................................   150
..................................................   200

Conditional fixed-effects logistic regression        Number of obs    = 12,035
                                                     Replications     =    200
Group variable: idcode                               Number of groups =  1,690

                                                     Obs per group:
                                                                  min =      2
                                                                  avg =    7.1
                                                                  max =     12

                                                     Wald chi2(6)     =  38.72
Log likelihood = -4510.888                           Prob > chi2      = 0.0000

                              (Replications based on 1,690 clusters in idcode)
------------------------------------------------------------------------------
             |   Observed   Bootstrap                         Normal-based
       union | coefficient  std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         age |   .0710973   .0945357     0.75   0.452    -.1141894    .2563839
       grade |   .0816111     .05568     1.47   0.143    -.0275198    .1907419
    not_smsa |   .0224809   .1536343     0.15   0.884    -.2786368    .3235986
     1.south |  -2.856488    .879619    -3.25   0.001    -4.580509   -1.132466
        year |  -.0636853   .0952784    -0.67   0.504    -.2504275    .1230568
             |
south#c.year |
          1  |   .0264136   .0109133     2.42   0.016     .0050239    .0478032
------------------------------------------------------------------------------

. bootstrap, reps(200) seed(12345): xtlogit union age grade not_smsa i.south##c.year, fe
(running xtlogit on estimation sample)

Bootstrap replications (200)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
..................................................    50
..................................................   100
..................................................   150
..................................................   200

Conditional fixed-effects logistic regression        Number of obs    = 12,035
                                                     Replications     =    200
Group variable: idcode                               Number of groups =  1,690

                                                     Obs per group:
                                                                  min =      2
                                                                  avg =    7.1
                                                                  max =     12

                                                     Wald chi2(6)     =  44.14
Log likelihood = -4510.888                           Prob > chi2      = 0.0000

                                  (Replications based on clustering on idcode)
------------------------------------------------------------------------------
             |   Observed   Bootstrap                         Normal-based
       union | coefficient  std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         age |   .0710973   .1149636     0.62   0.536    -.1542272    .2964217
       grade |   .0816111    .057185     1.43   0.154    -.0304694    .1936915
    not_smsa |   .0224809   .1352539     0.17   0.868    -.2426119    .2875738
     1.south |  -2.856488   .8285044    -3.45   0.001    -4.480326   -1.232649
        year |  -.0636853   .1161776    -0.55   0.584    -.2913893    .1640186
             |
south#c.year |
          1  |   .0264136   .0101533     2.60   0.009     .0065135    .0463136
------------------------------------------------------------------------------

.

That sais, I do not get why Stata complains about your first code.

[Paolo Maldini]

Paolo:
the two codes are different:

. use https://www.stata-press.com/data/r17/union. . use https://www.stata-press.com/data/r17/union
(NLS Women 14-24 in 1968)

. xtlogit union age grade not_smsa i.south##c.year, fe vce(bootstrap, reps(200) seed(12345) dots(1))
(running xtlogit on estimation sample)

Bootstrap replications (200)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5
.................................................. 50
.................................................. 100
.................................................. 150
.................................................. 200

Conditional fixed-effects logistic regression Number of obs = 12,035
Replications = 200
Group variable: idcode Number of groups = 1,690

Obs per group:
min = 2
avg = 7.1
max = 12

Wald chi2(6) = 38.72
Log likelihood = -4510.888 Prob > chi2 = 0.0000

(Replications based on 1,690 clusters in idcode)
------------------------------------------------------------------------------
| Observed Bootstrap Normal-based
union | coefficient std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
age | .0710973 .0945357 0.75 0.452 -.1141894 .2563839
grade | .0816111 .05568 1.47 0.143 -.0275198 .1907419
not_smsa | .0224809 .1536343 0.15 0.884 -.2786368 .3235986
1.south | -2.856488 .879619 -3.25 0.001 -4.580509 -1.132466
year | -.0636853 .0952784 -0.67 0.504 -.2504275 .1230568
|
south#c.year |
1 | .0264136 .0109133 2.42 0.016 .0050239 .0478032
------------------------------------------------------------------------------

. bootstrap, reps(200) seed(12345): xtlogit union age grade not_smsa i.south##c.year, fe
(running xtlogit on estimation sample)

Bootstrap replications (200)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5
.................................................. 50
.................................................. 100
.................................................. 150
.................................................. 200

Conditional fixed-effects logistic regression Number of obs = 12,035
Replications = 200
Group variable: idcode Number of groups = 1,690

Obs per group:
min = 2
avg = 7.1
max = 12

Wald chi2(6) = 44.14
Log likelihood = -4510.888 Prob > chi2 = 0.0000

(Replications based on clustering on idcode)
------------------------------------------------------------------------------
| Observed Bootstrap Normal-based
union | coefficient std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
age | .0710973 .1149636 0.62 0.536 -.1542272 .2964217
grade | .0816111 .057185 1.43 0.154 -.0304694 .1936915
not_smsa | .0224809 .1352539 0.17 0.868 -.2426119 .2875738
1.south | -2.856488 .8285044 -3.45 0.001 -4.480326 -1.232649
year | -.0636853 .1161776 -0.55 0.584 -.2913893 .1640186
|
south#c.year |
1 | .0264136 .0101533 2.60 0.009 .0065135 .0463136
------------------------------------------------------------------------------

.

That sais, I do not get why Stata complains about your first code.

Thanks for the clarification that the two models are different. Yes, I am puzzled as well as to why the STATA is struggling with the first code but not the second.

[Carlo Lazzaro]

Paolo:
the usual trivial question: is your copy of Stata fully updated?

[Paolo Maldini]

Paolo:
the usual trivial question: is your copy of Stata fully updated?

Good point and indeed, I have checked as follows and it seems up to date:

update query
(contacting http://www.stata.com)

Update status
    Last check for updates:  21 Feb 2023
    New update available:    none         (as of 21 Feb 2023)
    Current update level:    10 Jan 2023  (what's new)

Possible actions

    Do nothing; all files are up to date.

[Carlo Lazzaro]

Paolo:
are you copying and pasting the instructions for your first code from a .do file or are you coding them directly in the Stata Command window?

[Paolo Maldini]

Paolo:
are you copying and pasting the instructions for your first code from a .do file or are you coding them directly in the Stata Command window?

Thanks Carlo for the follow up!
I always code in a do.file, is it generally recommended to code in the Stata command window?

[Carlo Lazzaro]

Paolo:
do you let the .do file run the whole code or you just copy each chunck of code form the .do file and then past it in the Stata command window?

[Paolo Maldini]

Paolo:
do you let the .do file run the whole code or you just copy each chunck of code form the .do file and then past it in the Stata command window?

I both write & run the entire code in the do.file, I never paste it on the Stata command window. However, I have just tried the Stata command window and I receive the same error message,

an error occurred when bootstrap executed xtlogit

[Carlo Lazzaro]

Paolo:
I'd bring this issue up to https://www.stata.com/support/tech-support/contact.

[Paolo Maldini]

Thanks Carlo Lazzaro for the support!
Indeed, I have also checked several times that my Stata is up to date and all looks good.

[Paolo Maldini]

Paolo:
I'd bring this issue up to https://www.stata.com/support/tech-support/contact.

Hello Professor Carlo Lazzaro, I am still facing this technical error unfortunately despite running the same code with Stata 17 on different devices, do you happen to have any updates from the tech support team?

[Carlo Lazzaro]

Paolo;
no news, unfortunately.
The usual request to call me Carlo still applies 😀!

[Joerg Luedicke (StataCorp)]

The two specifications

use https://www.stata-press.com/data/r17/union

xtlogit union age grade not_smsa i.south##c.year, fe      ///
  vce(bootstrap, reps(200) seed(12345) dots(1))
  
bootstrap, reps(200) seed(12345):                         ///
  xtlogit union age grade not_smsa i.south##c.year, fe

yield different standard errors because they apply different bootstraps. xtlogit with vce(bootstrap) correctly resamples the panel units, that is, performs a clustered bootstrap, whereas the bootstrap prefix specification resamples individual observations. The following bootstrap specification yields the same results as xtlogit, vce(bootstrap):

bootstrap, reps(200) seed(12345)                         ///
  cluster(idcode) idcluster(newid) group(idcode):        ///
  xtlogit union age grade not_smsa i.south##c.year, fe

Please see help bootstrap for further information.