choice-metrics.com

Dear all,

I found in NgeneManual112 in page 108 that Ngene ignores any constants in designs for estimating multinomial logit models .
constants that is mean alternative-specific constants but I found information any way that it is must to put in syntax and my model should be no or put alternative-specific constants in syntax.

Design
;alts=cer, ba
;rows=14
;eff = (MNL, D)
;model:
U(cer) =b1.dummy[0 I 0]*I1 [1,2,3] +b2.dummy[0I0I0]*L1 [1,2,3,4] +b3[0]*salarycer[8985,13800] /
U(ba) = b0+b4.dummy[0I0I0]*I2 [4,5,6,7] +b5.dummy[0 I0I0]*L2 [5,6,7,8] +b6[0]*salaryba[11000,15000,18000] $

thank you very much.
Sukunta

You should always put constants in the model if you intend to estimate alternative specific constants.

What is referred to in the manual is that Ngene does not optimise for estimation of the constants by default. However, if want to also optimise on the constants, you simply add ;con to the syntax.

Dear Michiel Bliemer,

thank you very much for your kind and quick reply my question. I will learn more about this point.

Best Wishes,
Sukunta

Dear Prof. Michiel Bliemer ,
then from the answer above I can write syntax for eff design follow

Design
;alts=cer, ba
;rows=14
;eff = (MNL, D)
;model:
U(cer) =b1.dummy[0 I 0]*I1 [1,2,3] +b2.dummy[0I0I0]*L1 [1,2,3,4] +b3[0]*salarycer[8985,13800] /
U(ba) = b0+b4.dummy[0I0I0]*I2 [4,5,6,7] +b5.dummy[0 I0I0]*L2 [5,6,7,8] +b6[0]*salaryba[11000,15000,18000] $

your answer is reference for my thesis later.

thank you for your kindness
Best Wishes,
sukunta