How to choose model result from Ngene?

[sukunta]

Dear Prof. Michiel Bliemer and all,

I have a problem with my project that I finished face validity of model 1 (d-error 0.188046, A-error 6.571256 ,B-estimate 100, S estimate 0)
for 20 participants and then i try run Ngene again I found that model 2 has d-error 0.13121, A-error 2.487805 ,B-estimate 100, S estimate 0 . Could I change my model 1 to model 2 for pilot study? or i will go on model 1 or I start new face validity model 2

Best Wishes,

Sukunta (thailand)

[Michiel Bliemer]

I do not understand your question. What is model 1 and what is model 2?

[sukunta]

i am sorry for miss communicate to you . I mean one model can run for many scenarios by Ngene as we want then I choose first scenario and do face validity already.
and later I run the same model I got second scenario . first and second are difference value above question .then my question are
1.what 's scenario should I choose?
2.if I change my mind to choose second scenario I must do face validity again ?
3. Or I can use first scenario because they are not difference.

thank you for your kindness
Sukunta

[Michiel Bliemer]

I am afraid i still do not understand. What do you mean with a scenarios? How do you define a scenario? And what do you mean exactly with face validity?

[sukunta]

ok we are start new time. I ran 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] $
I ran twice I found that
first time present d-error 0.188046, A-error 6.571256 ,B-estimate 100, S estimate 0
and second time present d-error 0.13121, A-error 2.487805 ,B-estimate 100, S estimate 0
what 's design choice situation I choose?
thank you for your kindness

Best Wishes,
sukunta

[Michiel Bliemer]

I ran the syntax and after a few minutes running you will get a design with a D-error of around 0.13.
Each time you run it of course it shows different designs, since there are millions of possible designs to generate and Ngene randomly (in a smart way) picks one and evaluates it. I do not think it is possible that after a few minutes running it gives you a design with 0.18.
Of course the design with the lowest D-error is most efficient, so you may want to choose that design. You need to inspect whether the design makes sense, since you have not entered any priors and therefore Ngene does not know what the parameter values mean and cannot calculate probabilities.

[sukunta]

thank you very much for your quick reply.

sukunta