Stata xtreg在使用i.year时省略年份伪变量
我拥有以下年份的面板数据集：Stata xtreg在使用i.year时省略年份伪变量,stata,Stata,我拥有以下年份的面板数据集： tab year year | Freq. Percent Cum. ------------+----------------------------------- 2000 | 31 12.55 12.55 2001 | 31 12.55 25.10 2002 | 30
tab year
year | Freq. Percent Cum.
------------+-----------------------------------
2000 | 31 12.55 12.55
2001 | 31 12.55 25.10
2002 | 30 12.15 37.25
2003 | 31 12.55 49.80
2004 | 31 12.55 62.35
2005 | 31 12.55 74.90
2006 | 31 12.55 87.45
2007 | 31 12.55 100.00
------------+-----------------------------------
Total | 247 100.00
xtreg dv iv I.year20002007xtreg local_gr rtxdum i.year
note: 2007.year omitted because of collinearity
Random-effects GLS regression Number of obs = 247
Group variable: province_n~e Number of groups = 31
R-sq: Obs per group:
within = 0.6194 min = 7
between = 0.0016 avg = 8.0
overall = 0.2356 max = 8
Wald chi2(7) = 341.51
corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000
------------------------------------------------------------------------------
local_gr | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
rtxdum | -753799.7 291543.7 -2.59 0.010 -1325215 -182384.5
|
year |
2001 | 388246 291543.7 1.33 0.183 -183169.2 959661.2
2002 | 745406.4 294294.5 2.53 0.011 168599.8 1322213
2003 | 1175610 291543.7 4.03 0.000 604194.4 1747025
2004 | 1773982 291543.7 6.08 0.000 1202567 2345397
2005 | 2600005 291543.7 8.92 0.000 2028589 3171420
2006 | 4425318 291543.7 15.18 0.000 3853903 4996734
2007 | 0 (omitted)
|
_cons | 1564670 447832.4 3.49 0.000 686934.1 2442405
-------------+----------------------------------------------------------------
sigma_u | 2217878.8
sigma_e | 1150064.9
rho | .78809251 (fraction of variance due to u_i)
------------------------------------------------------------------------------
消息说2007年由于共线性被省略了，但我不明白为什么2000年不会出现在结果中？因为它是基准水平。您可以使用
allbaselevelswebuse nlswork, clear
xtset idcode
xtreg ln_w grade tenure i.race not_smsa south, allbaselevels
Random-effects GLS regression Number of obs = 28,091
Group variable: idcode Number of groups = 4,697
R-sq: Obs per group:
within = 0.1005 min = 1
between = 0.4498 avg = 6.0
overall = 0.3305 max = 15
Wald chi2(6) = 6509.50
corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000
------------------------------------------------------------------------------
ln_wage | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
grade | .07605 .0018128 41.95 0.000 .0724969 .0796031
tenure | .0361319 .0006298 57.37 0.000 .0348975 .0373663
|
race |
white | 0 (base)
black | -.0530121 .0102916 -5.15 0.000 -.0731832 -.0328409
other | .0762678 .0415911 1.83 0.067 -.0052492 .1577849
|
not_smsa | -.1289554 .0074296 -17.36 0.000 -.1435172 -.1143936
south | -.0786512 .0075533 -10.41 0.000 -.0934555 -.063847
_cons | .6759773 .0244723 27.62 0.000 .6280125 .7239421
-------------+----------------------------------------------------------------
sigma_u | .26440074
sigma_e | .30295598
rho | .43235646 (fraction of variance due to u_i)
------------------------------------------------------------------------------
正如@PearlySpuncer所暗示的，您总是——使用这种语法，您看到的指标（虚拟）变量的结果要比您投入回归的指标（虚拟）变量少。为什么省略2007也是因为数据集中的共线性。这是一个统计理解的问题。很明显，你没有读过这个问题，我说的结果是省略了两年，而不是一年。2000年和2007年。所以不是少一个，少两个。我不明白为什么我要在这篇评论中再次解释我自己。我确实阅读并理解了这个问题。重复：总是少看到一个指标（那就是一个）；由于数据集中的共线性，2007也被省略（在您的案例中是两个）。