用probit做回归分析，在加入控制变量，发现观测变量减少！
我原来的样本量是4225个
note: 2013.year != 0 predicts failure perfectly
      2013.year dropped and 908 obs not used
note: 4.industry1 != 0 predicts failure perfectly
      4.industry1 dropped and 1 obs not used
note: 5.industry1 != 0 predicts failure perfectly
      5.industry1 dropped and 4 obs not used
note: 10.industry1 != 0 predicts success perfectly
      10.industry1 dropped and 1 obs not used
note: 17.industry1 != 0 predicts failure perfectly
      17.industry1 dropped and 2 obs not used
note: 19.industry1 != 0 predicts failure perfectly
      19.industry1 dropped and 1 obs not used
note: 22.industry1 != 0 predicts failure perfectly
      22.industry1 dropped and 1 obs not used
note: 24.industry1 != 0 predicts failure perfectly
      24.industry1 dropped and 1 obs not used
note: 26.industry1 != 0 predicts failure perfectly
      26.industry1 dropped and 6 obs not used
note: 30.industry1 != 0 predicts failure perfectly
      30.industry1 dropped and 4 obs not used
note: 33.industry1 != 0 predicts failure perfectly
      33.industry1 dropped and 7 obs not used
note: 35.industry1 != 0 predicts success perfectly
      35.industry1 dropped and 1 obs not used
note: 36.industry1 != 0 predicts failure perfectly
      36.industry1 dropped and 1 obs not used
Probit regression                                 Number of obs   =       3187
                                                  Wald chi2(97)   =     469.46
                                                  Prob > chi2     =     0.0000
Log pseudolikelihood = -1727.3179                 Pseudo R2       =     0.1393
求大牛帮我看看如何能使得观测量不发生变化！