Statalist The Stata Listserver


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: asmprobit, problems with correlation patterns


From   "Barbara Hanel " <[email protected]>
To   [email protected]
Subject   st: asmprobit, problems with correlation patterns
Date   Fri, 20 Apr 2007 17:32:01 +0200

Hello everyone,

I would like to estimate a mulitnomial probit model explaining a 
person's labour market status.
There are four different destinations: 
* employed
* unemployed
* retired
* other
that are possibly correlated.

Now I tried to use -asmprobit- and specified different correlation 
patterns. 
But all of them (except "independent") failed to converge. 

I�m not very surprised that "unstructured" didn�t work, but there are 
two different problems I don�t really understand:

1) If I specify the correlation pattern

matrix define B=J(4,4,.)
matrix B[3,2]=1
matrix B[4,3]=1
matrix B[4,2]=1

I think I restrict asmprobit to estimate only one covariance 
parameter and two variances 
(for the two alternatives not being 'scale' or 'base').(?)

For this correlation pattern, the covariance parameter is always 
EXACTLY -0.5
Even with different explanatory variables included in the model.

How is that possible?


2) If I include more than one explanatory variable in the model I get 
the message:
"Note: two or more of the variables are collinear; convergence may 
not be achieved"
That's the case even if I include:
* one variable varying across alternatives and cases and  
* only one other variable that is constant across the alternatives, 
varies only by "i". 

I don�t understand how two such variables could be collinear?


Both problems occur with different optimization techniques.
I�m not sure if it may help, but I have printed the output below.

I would really appreciate any suggestions!

Barbara


The output is as follows:

########################################################
####################
. asmprobit wahl eink AB, case(z1) alternatives(altern) 
casevars(AB) correlation(pattern B) 
> iterate(100) hessian gradient shownrtolerance
Note: two or more of the variables are collinear; convergence may 
not be achieved

(output omitted - Iterations 1 to 31)


Alternative-specific multinomial probit        Number of obs      =       
9700
Case variable: z1                              Number of cases    =       
2425

Alternative variable: altern                   Alts per case: min =          4
                                                              avg =        4.0
                                                              max =          4
Integration sequence:      Hammersley
Integration points:               200             Wald chi2(4)    =          .
Log simulated-likelihood = -2429.6698             Prob > chi2     =      
    .

------------------------------------------------------------------------------
        wahl |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
altern       |
        eink |    .000631   .0001332     4.74   0.000       .00037    
.0008921
          AB |  (dropped)
-------------+----------------------------------------------------------------
ET           |  (base alternative)
-------------+----------------------------------------------------------------
AL           |
          AB |   6.498919   2.957141     2.20   0.028     .7030285    
12.29481
       _cons |   -1.80445   .1309075   -13.78   0.000    -2.061024   -
1.547876
-------------+----------------------------------------------------------------
R            |
          AB |  -4.308268   2.407709    -1.79   0.074     -9.02729    
.4107547
       _cons |  -1.236563   .0867427   -14.26   0.000    -1.406576   -
1.066551
-------------+----------------------------------------------------------------
KS           |
          AB |   15.49664   1.622816     9.55   0.000     12.31598     
18.6773
       _cons |  -.1677057   .0752585    -2.23   0.026    -.3152096   -
.0202019
-------------+----------------------------------------------------------------
  /lnsigmaP1 |   .0504696          .        .       .            .           .
  /lnsigmaP2 |   .1474712          .        .       .            .           .
-------------+----------------------------------------------------------------
   /atanhrP1 |  -.5493061   .0127844   -42.97   0.000    -.5743631   -
.5242492
-------------+----------------------------------------------------------------
      sigma1 |          1  (base alternative)
      sigma2 |          1  (scale alternative)
      sigma3 |   1.051765          .                             .           .
      sigma4 |     1.1589          .                             .           .
-------------+----------------------------------------------------------------
      rho3_2 |        -.5   .0095883                     -.5185563   -.4809729
      rho4_2 |        -.5   .0095883                     -.5185563   -.4809729
      rho4_3 |        -.5   .0095883                     -.5185563   -.4809729
------------------------------------------------------------------------------
(altern=ET is the alternative normalizing location)
(altern=AL is the alternative normalizing scale)

-- -- -- -- -- -- -- -- -- -- -- -- -- -- 
Dipl.-Volksw. Barbara Hanel
FAU Erlangen N�rnberg
Lehrstuhl f�r Statistik und empirische Wirtschaftsforschung; Prof. Riphahn, 
PhD
0911/5302-258
Lange Gasse 20
90403 N�rnberg
[email protected] 

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index