做异质性分析，是在总样本缩尾以后再用keep if 把异质性数据给筛选出来分析，还是先把异质性数据筛选出来再缩尾？
我用前一种方法得出来的结论是我想要的，第二种的出来的结论不显著。但是我不知道哪种数据处理是对的。nature是产权异质性。
Example generated by -dataex-. To install: ssc install dataex
clear
input float id str69 name byte nature double SA byte Level double(Size LR ReGrowth)
2 "万丰奥威" 0 3.836977 4 139.88931004505744  .52385  .170325
2 "万丰奥威" 0 3.865084 5 159.73240324650527 .529911  .138336
2 "万丰奥威" 0 3.902603 5 163.89316942736988 .557458  .317489
3 "万华化学" 1 3.633503 5  968.6528060582796 .546471  .147723
3 "万华化学" 1 3.588238 7 1337.5276302168859 .613834   .32067
3 "万华化学" 1 3.524808 8 1903.0960940887135 .623268  -.03636
4 "万向钱潮" 0 4.168675 5  120.2817778992541  .54705  .200287
4 "万向钱潮" 0 4.187972 5 154.90065763914936 .428987  .117614
4 "万向钱潮" 0 4.220038 6  168.2078685583612 .477766  .140225
5 "万达信息" 1 4.112137 5  76.14024795473954 .617534 -.095705
5 "万达信息" 1 4.154569 5   69.7831168455024 .787028  .626601
5 "万达信息" 1 4.193976 5  71.44765068286267 .784656  .268288
6 "万里扬"    0 3.775238 5 109.01720217892023 .428061  .498964
6 "万里扬"    0 3.811303 6 115.80851819946936 .422466  .214224
6 "万里扬"    0   3.8603 6  100.0262614668575 .465693 -.265438
7 "三一重工" 0 3.570211 4  905.4135737196206 .497172  .109137
7 "三一重工" 0 3.524141 4 1262.5448120561787 .539124  .115818
7 "三一重工" 0 3.538371 4  1385.566120501335 .530191 -.106613
8 "三七互娱" 0 4.099512 4 101.45253108824814 .281072  .051113
8 "三七互娱" 0 4.137161 5 105.64334676386156 .425856 -.057768
end