clogit回归中为什么没有替代常数项呢

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

如题，用stata跑数据做clogit回归，发现没有asc替代常数项。。。但是在书中看clogit原理中是应该有的。。。什么原因。。。非常感谢！

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

分享0 收藏0 回帖

关键词：clogit logit Log CLO 常数项

you cannot estimate the intercept in clogit, and the intercept term doesn't impact the result.

jjjj6666 发表于 2014-4-2 03:17
you cannot estimate the intercept in clogit, and the intercept term doesn't impact the result.

but why the explaination in book<Introduction to Econometrics> says it will have a ASC?and I see there are ASCs in asclogit results in official asclogit reference.
it will be so great if you can leave your contact message. thx so much.

jjjj6666 发表于 2014-4-2 03:17
you cannot estimate the intercept in clogit, and the intercept term doesn't impact the result.

and I am trying to construct an utility function. does that mean that there is no ASC in the function?

that's for choice model, but if you apply asclogit to the data used by clogit, you have to use noconstant option, otherwise, I think the estimating procedure will not converge

jjjj6666 发表于 2014-4-3 07:56
that's for choice model, but if you apply asclogit to the data used by clogit, you have to use nocon ...

hi,thx again.there are lines in the mixed logit command book.
the fitted models have no alternative-specific constants. this is common practice when the data come from a so-called unlabeled choice experiments, where the alternatives have no utility beyond the characteristics to them in the experiment.
is that the reason?
but what it if I do want ASCs to perform a function? should I label it first or sth?

how many alternatives in your dependent variable? if 2, you may use clogit, asclogit, and mixlogit.  If more than 2, I think you have to use asclogit or mixlogit.

jjjj6666 发表于 2014-4-3 19:36
how many alternatives in your dependent variable? if 2, you may use clogit, asclogit, and mixlogit.  ...

three,two alternatives and one status quo. other literatures suggested two ASCs. one for the function of the two alternatives,and one for the status quo function.

jjjj6666 发表于 2014-4-3 19:36
how many alternatives in your dependent variable? if 2, you may use clogit, asclogit, and mixlogit.  ...

  here is my data.PID is person's ID. CID is choice pair's ID, CHO is choice, VN VH VF VC are some attributes, and EDU INC AGE are personal information.
  +-------------------------------------------------------+
     | PID   CID   CHO   VN   VH   VF   VC   EDU   INC   AGE |
     |-------------------------------------------------------|
  1. |   1     1     1    1    1    2    2     4     3     3 |
  2. |   1     1     0    2    1    1    1     4     3     3 |
  3. |   1     1     0    0    0    0    0     4     3     3 |
     |-------------------------------------------------------|
  4. |   1     2     1    1    2    2    2     4     3     3 |
  5. |   1     2     0    2    1    1    1     4     3     3 |
  6. |   1     2     0    0    0    0    0     4     3     3 |
     |-------------------------------------------------------|
  7. |   1     3     1    1    2    1    1     4     3     3 |
  8. |   1     3     0    2    1    2    2     4     3     3 |
  9. |   1     3     0    0    0    0    0     4     3     3 |
     |-------------------------------------------------------|
10. |   1     4     0    1    1    1    2     4     3     3 |
11. |   1     4     0    2    2    2    1     4     3     3 |
12. |   1     4     1    0    0    0    0     4     3     3 |
     |-------------------------------------------------------|
13. |   1     5     1    1    1    2    2     4     3     3 |
14. |   1     5     0    2    1    1    1     4     3     3 |
15. |   1     5     0    0    0    0    0     4     3     3 |
     |-------------------------------------------------------|
16. |   1     6     1    1    1    2    2     4     3     3 |
17. |   1     6     0    2    2    1    1     4     3     3 |
18. |   1     6     0    0    0    0    0     4     3     3 |
     +-------------------------------------------------------+

jjjj6666 发表于 2014-4-3 19:36
how many alternatives in your dependent variable? if 2, you may use clogit, asclogit, and mixlogit.  ...

here is my command and results.

. clogit CHO VP VN VH VF VC, group(CID)
Iteration 0:   log likelihood =  -2153.701  
Iteration 1:   log likelihood =  -2114.572  
Iteration 2:   log likelihood = -2107.1917  
Iteration 3:   log likelihood = -2107.1903  
Iteration 4:   log likelihood = -2107.1903  

Conditional (fixed-effects) logistic regression   Number of obs   =       8424
                                                                  LR chi2(5)      =    1955.43
                                                                 Prob > chi2     =     0.0000
Log likelihood = -2107.1903                          Pseudo R2       =     0.3169

------------------------------------------------------------------------------
         CHO |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
          VP |  -.0047628   .0004674   -10.19   0.000    -.0056789   -.0038467
          VN |   .7408797   .0433607    17.09   0.000     .6558943    .8258651
          VH |   .4605629   .0407377    11.31   0.000     .3807184    .5404073
          VF |   .3646686   .0401135     9.09   0.000     .2860476    .4432897
          VC |   .5338821   .0407072    13.12   0.000     .4540975    .6136666
------------------------------------------------------------------------------