[Limdep Nlogit List] imposing a constrain on the distribution

[sally alain]

Hello,

Following Hensher and
Greene (2003), I tied to use conditional parameter estimates to derive plausible
WTP values.  I tried to restrict the sign
and the range of a triangular parameter by constraining the spread to that of the
mean of the random parameter. However the specification Fcn=name(t,1) does not
work with binary models. It fixing the parameter than imposing a constrain on
the distribution.

 

calc;ran(12345)$

--> logit;lhs=choice;rhs=one,x1,x2,x3,x4

    ;rpm

   
;fcn=x1(t,1),x2(t),x3(t),x4(t)

    ;pts=20

    ;pds=4

    ;shufffled

   
;parameters

    ;matrix$

+--------+--------------+----------------+--------+--------+

|Variable| Coefficient  |
Standard Error |b/St.Er.|P[|Z|>z]|

+--------+--------------+----------------+--------+--------+

---------+Nonrandom parameters

 Constant|   -1.32856117       .24237666    -5.481  
.0000

---------+Means for random parameters

 X1      |    
.70710678    ......(Fixed
Parameter).......

 X2      |    
.03499364       .00173257    20.198  
.0000  

 X3      |    
.05693492       .00792475    
7.184   .0000  

 X4      |   
-.19909318       .00875393   -22.743  
.0000  

---------+Scale parameters for dists. of random parameters

 X1      |   
2.22470699       .01125844   197.604  
.0000

 X2      |    
.06967358       .00412668    16.884  
.0000

 X3      |    
.32085919       .01092655    29.365  
.0000

 X4      |    
.03397756       .01162298     2.923  
.0035

 

I wonder if there is a
way to solve this problem.

Cheers,

Sally