Создание условий взаимодействия вручную - PullRequest
Новая
       7

Создание условий взаимодействия вручную

1 голос
/ 04 июня 2019

Я хотел бы включить термин взаимодействия в мою модель: 

gen state_dom = 0 
replace state_dom=1 if state_ownership>=25 
gen state_min = 0 
replace state_min=1 if state_ownership<25 & state_ownership>0

egen voc = group(vis) //CME=1, Emergent LME=2, Hierarchically Coordinated=3, State Led=4.

gen statedomCME = 0
gen statedomELME = 0
gen statedomHC = 0
gen statedomSL = 0
replace statedomCME = state_dom*voc if voc==1
replace statedomELME = state_dom*voc if voc==2
replace statedomHC = state_dom*voc if voc==3
replace statedomSL = state_dom*voc if voc==4

xtset id_company year //definisco un binomial panel NxT

xtreg foreign_revenues state_dom##voc log_age log_asset log_gdp_capita i.sector i.year, robust 
xtreg foreign_revenues state_dom i.voc statedomCME statedomELME statedomHC statedomSL log_age log_asset log_gdp_capita i.sector i.year, robust

Почему я получаю разные результаты для двух команд xtreg?

1 Ответ

3 голосов
/ 04 июня 2019

Во второй модели вы не настраиваете взаимодействия соответствующим образом вручную.

Рассмотрите следующий пример, используя набор игрушек Stata nlswork: 

webuse nlswork, clear
xtset idcode

generate wks = 1 if wks_work <= 30 
replace wks = 2 if wks_work > 30 & wks_work < 60
replace wks = 3 if wks_work > 59

xtreg ln_w age wks##i.race south, robust

Random-effects GLS regression                   Number of obs     =     28,502
Group variable: idcode                          Number of groups  =      4,710

R-sq:                                           Obs per group:
     within  = 0.1135                                         min =          1
     between = 0.2051                                         avg =        6.1
     overall = 0.1621                                         max =         15

                                                Wald chi2(10)     =    1786.90
corr(u_i, X)   = 0 (assumed)                    Prob > chi2       =     0.0000

                             (Std. Err. adjusted for 4,710 clusters in idcode)
------------------------------------------------------------------------------
             |               Robust
     ln_wage |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         age |   .0155126   .0006254    24.80   0.000     .0142868    .0167383
             |
         wks |
          2  |   .1220641   .0073903    16.52   0.000     .1075793    .1365488
          3  |   .1525508   .0093953    16.24   0.000     .1341364    .1709652
             |
        race |
      black  |  -.0720868   .0141309    -5.10   0.000    -.0997828   -.0443908
      other  |   .1073435   .0682385     1.57   0.116    -.0264015    .2410885
             |
    wks#race |
    2#black  |  -.0158733   .0136542    -1.16   0.245    -.0426351    .0108885
    2#other  |  -.0419947   .0518261    -0.81   0.418    -.1435719    .0595826
    3#black  |  -.0179945   .0167701    -1.07   0.283    -.0508633    .0148743
    3#other  |  -.0588866   .0681432    -0.86   0.388    -.1924448    .0746716
             |
       south |  -.1231714   .0107808   -11.43   0.000    -.1443013   -.1020415
       _cons |   1.179882   .0180816    65.25   0.000     1.144443    1.215321
-------------+----------------------------------------------------------------
     sigma_u |  .32371944
     sigma_e |  .30145969
         rho |  .53556032   (fraction of variance due to u_i)
------------------------------------------------------------------------------

Вы можетевручную создайте взаимодействия, используя устаревший префикс xi: 

xi: xtreg ln_w age i.wks*i.race south, robust
i.wks             _Iwks_1-3           (naturally coded; _Iwks_1 omitted)
i.race            _Irace_1-3          (naturally coded; _Irace_1 omitted)
i.wks*i.race      _IwksXrac_#_#       (coded as above)

Random-effects GLS regression                   Number of obs     =     28,502
Group variable: idcode                          Number of groups  =      4,710

R-sq:                                           Obs per group:
     within  = 0.1135                                         min =          1
     between = 0.2051                                         avg =        6.1
     overall = 0.1621                                         max =         15

                                                Wald chi2(10)     =    1786.90
corr(u_i, X)   = 0 (assumed)                    Prob > chi2       =     0.0000

                              (Std. Err. adjusted for 4,710 clusters in idcode)
-------------------------------------------------------------------------------
              |               Robust
      ln_wage |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------+----------------------------------------------------------------
          age |   .0155126   .0006254    24.80   0.000     .0142868    .0167383
      _Iwks_2 |   .1220641   .0073903    16.52   0.000     .1075793    .1365488
      _Iwks_3 |   .1525508   .0093953    16.24   0.000     .1341364    .1709652
     _Irace_2 |  -.0720868   .0141309    -5.10   0.000    -.0997828   -.0443908
     _Irace_3 |   .1073435   .0682385     1.57   0.116    -.0264015    .2410885
_IwksXrac_2_2 |  -.0158733   .0136542    -1.16   0.245    -.0426351    .0108885
_IwksXrac_2_3 |  -.0419947   .0518261    -0.81   0.418    -.1435719    .0595826
_IwksXrac_3_2 |  -.0179945   .0167701    -1.07   0.283    -.0508633    .0148743
_IwksXrac_3_3 |  -.0588866   .0681432    -0.86   0.388    -.1924448    .0746716
        south |  -.1231714   .0107808   -11.43   0.000    -.1443013   -.1020415
        _cons |   1.179882   .0180816    65.25   0.000     1.144443    1.215321
--------------+----------------------------------------------------------------
      sigma_u |  .32371944
      sigma_e |  .30145969
          rho |  .53556032   (fraction of variance due to u_i)
-------------------------------------------------------------------------------
...