April-sy 发表于 2012-11-13 11:21
非常感激蓝色分享的资料~
我将ivqrstata.zip里的文件全部拷到plus目录下,确实可以做ivqreg的命令了。
不 ...
我没有遇到你说的q()取值与结果无关的问题。
我的自有数据计算:
ivqreg stdGPA stdgk SES (mdepgpa= mdepgk) ,q(.25)
(0 observations deleted)
Initial Estimation: .25th Two Stage Quantile Regression Number of obs = 1318
------------------------------------------------------------------------------
stdGPA | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
mdepgpa | 1.37577 .8494784 1.62 0.105 -.2891766 3.040717
stdgk | .0478301 .0109073 4.39 0.000 .0264522 .0692079
SES | .0811438 .0351794 2.31 0.021 .0121935 .150094
_cons | -4.618749 .9519425 -4.85 0.000 -6.484522 -2.752976
------------------------------------------------------------------------------
Grid search is in progress (200)
.................................................. 50
.................................................. 100
.................................................. 150
.................................................. 200
.25th Instrumental Variable Quantile Regression Number of obs = 1318
------------------------------------------------------------------------------
stdGPA | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
mdepgpa | .4275757 .0012511 341.75 0.000 .4251235 .4300278
stdgk | .0394529 .010224 3.86 0.000 .0194143 .0594915
SES | .0758194 .0360322 2.10 0.035 .0051974 .1464413
_cons | -3.775353 1.131534 -3.34 0.001 -5.99312 -1.557587
------------------------------------------------------------------------------
. ivqreg stdGPA stdgk SES (mdepgpa= mdepgk) ,q(.5)
(0 observations deleted)
Initial Estimation: .5th Two Stage Quantile Regression Number of obs = 1318
------------------------------------------------------------------------------
stdGPA | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
mdepgpa | .2708511 .9382584 0.29 0.773 -1.568102 2.109804
stdgk | .0323538 .0121231 2.67 0.008 .0085929 .0561147
SES | .0679616 .0391546 1.74 0.083 -.0087801 .1447033
_cons | -2.636895 1.053859 -2.50 0.012 -4.70242 -.5713705
------------------------------------------------------------------------------
Grid search is in progress (200)
.................................................. 50
.................................................. 100
.................................................. 150
.................................................. 200
.5th Instrumental Variable Quantile Regression Number of obs = 1318
------------------------------------------------------------------------------
stdGPA | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
mdepgpa | .1463022 .0011603 126.09 0.000 .144028 .1485764
stdgk | .0350564 .0094821 3.70 0.000 .0164719 .053641
SES | .0510803 .0334177 1.53 0.126 -.0144172 .1165777
_cons | -2.790949 1.049428 -2.66 0.008 -4.84779 -.734108
------------------------------------------------------------------------------
.
关于您出问题的原因,我对这个命令也很不了解,所以无法回答您。
此外,这个ivqreg 的 结果与h3327156 所介绍的control funtion approach应用于qreg 的结果相去甚远,我也不清楚其中的机理所在。
cfa的结果如下,可与ivqreg比较可见不同,不知采信哪个较好??
. reg mdepgpa mdepgk stdgk stdGPA
Source | SS df MS Number of obs = 1318
-------------+------------------------------ F( 3, 1314) = 27.65
Model | 32.3686447 3 10.7895482 Prob > F = 0.0000
Residual | 512.74902 1314 .390219954 R-squared = 0.0594
-------------+------------------------------ Adj R-squared = 0.0572
Total | 545.117664 1317 .413908629 Root MSE = .62468
------------------------------------------------------------------------------
mdepgpa | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
mdepgk | .0013635 .0006082 2.24 0.025 .0001703 .0025567
stdgk | -.0130254 .0050306 -2.59 0.010 -.0228943 -.0031565
stdGPA | .1554516 .0178551 8.71 0.000 .1204241 .1904792
_cons | .3336425 .5564462 0.60 0.549 -.7579776 1.425263
------------------------------------------------------------------------------
. predict rho2 , r
. sqreg stdGPA SES stdgk mdepgpa rho2 ,q(.25 .5 )
(fitting base model)
(bootstrapping ....................)
Simultaneous quantile regression Number of obs = 1318
bootstrap(20) SEs .25 Pseudo R2 = 0.7990
.50 Pseudo R2 = 0.7847
------------------------------------------------------------------------------
| Bootstrap
stdGPA | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
q25 |
SES | .0175594 .0044586 3.94 0.000 .0088127 .026306
stdgk | .0824257 .0013524 60.95 0.000 .0797726 .0850788
mdepgpa | 6.25428 .0479095 130.54 0.000 6.160292 6.348267
rho2 | -6.248641 .0441583 -141.51 0.000 -6.33527 -6.162013
_cons | -7.524477 .114393 -65.78 0.000 -7.74889 -7.300064
-------------+----------------------------------------------------------------
q50 |
SES | .0160368 .0052653 3.05 0.002 .0057074 .0263662
stdgk | .081846 .0025737 31.80 0.000 .076797 .0868951
mdepgpa | 6.2776 .0463652 135.39 0.000 6.186642 6.368558
rho2 | -6.288204 .0461893 -136.14 0.000 -6.378817 -6.197591
_cons | -7.380184 .2180596 -33.84 0.000 -7.807967 -6.9524
------------------------------------------------------------------------------
.
end of do-file
.