门槛模型回归中遇到的问题

easylife

1799

收藏 2016-10-14

悬赏 2 个论坛币未解决

用连老师的程序做门槛模型，三重门槛跑出的结果异常，------------------------------------------------------------------------------
   lnscore |    Coef. Std. Err.    t P>|t|    [95% Conf. Interval]
-------------+----------------------------------------------------------------
   patent2 | .0407027 .0056751    7.17 0.000    .0295452 .0518602
   three | .0137706 .0030825    4.47 0.000    .0077104 .0198308
   lngdp | .728314 .1109646    6.56 0.000    .5101538 .9464743
      city |  -.0316464 .0112927 -2.80 0.005 -.0538482 -.0094446
   city3 |  -.0020575 .012004 -0.17 0.864 -.0256578 .0215428
people2_1 |  -.0409417 .0027367 -14.96 0.000 -.0463222 -.0355612
people2_2 |  -.0155329 .0014306 -10.86 0.000 -.0183454 -.0127203
people2_3 |       0  (omitted)
people2_4 |  -.0086291 .0014964 -5.77 0.000    -.011571 -.0056871
   _cons | 1.434257 .4037603    3.55 0.000    .6404499 2.228063
-------------+----------------------------------------------------------------
   sigma_u |  .41366473
   sigma_e |  .22548517
      rho |  .77093581 (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0:    F(19, 392) = 14.21          Prob > F = 0.0000
Note: people2_1: people2*I(people2<32.83)
   people2_2: people2*I(32.83<=people2<32.83)
   people2_3: people2*I(32.83<=people2<83.29000000000001)
   people2_4: people2*I(people2>=83.29000000000001)
可以看出  people2_3被剔除了，而且people2_2: people2*I(32.83<=people2<32.83)也很怪，根本没有按照程序跑出的门槛值来划分，请问哪位大神知道这是哪里出了问题吗？请赐教！！不胜感激！！！