全部版块 我的主页
论坛 计量经济学与统计论坛 五区 计量经济学与统计软件 Stata专版
2752 1
2019-06-26
rdbwselect里面输出的结果有两个  一个是BW est.(h)  一个是BW bias.里面的e(h_mserd) 被解释为MSE-optimal bandwidth selector for the RD treatment effect estimator.
e(b_mserd)被解释为 MSE-optimal bandwidth selector for the bias of the RD treatment effect estimator.
请问两者的区别是什么?我会翻译但不懂其经济学含义。
最优带宽选择的时候选择前者还是后者呢?


e(h_mserd)           MSE-optimal bandwidth selector for the RD treatment effect estimator.
      e(h_msetwo_l)        MSE-optimal bandwidth selectors below the cutoff for the RD treatment effect estimator.
      e(h_msetwo_r)        MSE-optimal bandwidth selectors above the cutoff for the RD treatment effect estimator.
      e(h_msesum)          MSE-optimal bandwidth selector for the sum of regression estimates.
      e(h_msecomb1)        for min(mserd,msesum).
      e(h_msecomb2_l)      for median(msetwo,mserd,msesum), below the cutoff.
      e(h_msecomb2_r)      for median(msetwo,mserd,msesum), above the cutoff.


      e(h_cerrd)           CER-optimal bandwidth selector for the RD treatment effect estimator.
      e(h_certwo_l)        CER-optimal bandwidth selectors below the cutoff for the RD treatment effect estimator.
      e(h_certwo_r)        CER-optimal bandwidth selectors above the cutoff for the RD treatment effect estimator.
      e(h_cersum)          CER-optimal bandwidth selector for the sum of regression estimates.
      e(h_cercomb1)        for min(cerrd,cersum).
      e(h_cercomb2_l)      for median(certwo_l,cerrd,cersum), below the cutoff.
      e(h_cercomb2_r)      for median(certwo_r,cerrd,cersum), above the cutoff.


      e(b_mserd)           MSE-optimal bandwidth selector for the bias of the RD treatment effect estimator.
      e(b_msetwo_l)        MSE-optimal bandwidth selectors below the cutoff for the bias of the RD treatment effect estimator.
      e(b_msetwo_r)        MSE-optimal bandwidth selectors above the cutoff for the bias of the RD treatment effect estimator.
      e(b_msesum)          MSE-optimal bandwidth selector for the sum of regression estimates for the bias of the RD treatment effect estimator.
      e(b_msecomb1)        for min(mserd,msesum).
      e(b_msecomb2_l)      for median(msetwo,mserd,msesum), below the cutoff.
      e(b_msecomb2_r)      for median(msetwo,mserd,msesum), above the cutoff.


      e(b_cerrd)           CER-optimal bandwidth selector for the bias of the RD treatment effect estimator.
      e(b_certwo_l)        CER-optimal bandwidth selectors below the cutoff for the bias of the RD treatment effect estimator.
      e(b_certwo_r)        CER-optimal bandwidth selectors above the cutoff for the bias of the RD treatment effect estimator.
      e(b_cersum)          CER-optimal bandwidth selector for the sum of regression estimates for the bias of the RD treatment effect estimator.
      e(b_cercomb1)        for min(cerrd,cersum).
      e(b_cercomb2_l)      for median(certwo_l,cerrd,cersum), below the cutoff.
      e(b_cercomb2_r)      for median(certwo_r,cerrd,cersum), above the cutoff.



二维码

扫码加我 拉你入群

请注明:姓名-公司-职位

以便审核进群资格,未注明则拒绝

全部回复
2020-5-12 00:20:05
同问~
二维码

扫码加我 拉你入群

请注明:姓名-公司-职位

以便审核进群资格,未注明则拒绝

相关推荐
栏目导航
热门文章
推荐文章

说点什么

分享

扫码加好友,拉您进群
各岗位、行业、专业交流群