RD回归断点处预测值超出合理范围，是什么情况？为什么会出现这种问题？如何解决呢

. rd fp3_p retire margin,z0(0) gr
Three variables specified; jump in treatment  
at Z=0 will be estimated. Local Wald Estimate
is the ratio of jump in outcome to jump in treatment.

Assignment variable Z is margin
Treatment variable X_T is retire
Outcome variable y is fp3_p

Command used for graph: lpoly; Kernel used: triangle (default)
Bandwidth: 2.8162297; loc Wald Estimate: -.08233519
Bandwidth: 1.4081148; loc Wald Estimate: -.03076593
Bandwidth: 5.6324593; loc Wald Estimate: .03033591
Estimating for bandwidth 2.816229665526795
A predicted value of treatment at cutoff lies outside feasible range;
switching to local mean smoothing for treatment discontinuity.
Estimating for bandwidth 1.408114832763397
Estimating for bandwidth 5.63245933105359
------------------------------------------------------------------------------
       fp3_p |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
       numer |  -.0628363   .1154589    -0.54   0.586    -.2891316     .163459
       denom |    .683511    .074211     9.21   0.000       .53806     .828962
       lwald |  -.0919316   .1682421    -0.55   0.585    -.4216802    .2378169
     numer50 |  -.0216501    .050602    -0.43   0.669    -.1208282     .077528
     denom50 |   .7037037   .0926974     7.59   0.000     .5220202    .8853872
     lwald50 |  -.0307659   .0717796    -0.43   0.668    -.1714513    .1099195
    numer200 |   .0200117   .0656933     0.30   0.761    -.1087448    .1487682
    denom200 |   .6596689   .1337316     4.93   0.000     .3975597    .9217781
    lwald200 |   .0303359   .1001459     0.30   0.762    -.1659465    .2266183
------------------------------------------------------------------------------

如上图

我也遇到了这种情况，请问楼主解决了吗

hyoni 发表于 2020-10-9 19:59
我也遇到了这种情况，请问楼主解决了吗

我也遇到了同样的问题，请问您解决了吗？