Page 1 of 1

dominant choice

PostPosted: Mon Feb 07, 2022 1:28 am
by hefei0913
Hello all,

We are designing a willingness to accept choice experiment. There are five attributes: payment (price), nrate, nmethod, irrigation and cover crop.

We know the coefficient of payment is positive but we don't know other attributes' coefficients, so we just added 0.00001 as the payment prior. However, the output design includes some dominant choices. As each of our attributes has a price (cost), the willingness to accept payment should at least cover part of the cost. Also, a lower cost with a higher payment choice set is also considered as a dominant choice in our case. Here is our code. Could you help us to figure out how to solve this issue? Thank you very much for your time!

Design
;alts=alt1*, alt2*, optout
;rows=12
;block=2
;eff=(mnl,d)
;alg=mfederov
;con
;model:
U(alt1) = bpayment (0.00001) *payment[40,90,140,190](3-4,3-4,3-4,3-4) +
bnmethod *nmethod[0,60,120](4-5,4-5,4-5) +
birrigation *irrigation[0,80] +
bnrate *nrate[0,1] +
bcovercrop *covercrop[0,40] /

U(alt2) =bpayment*payment + bnmethod*nmethod
+ birrigation *irrigation + bnrate*nrate + bcovercrop*covercrop /

U(optout)=bcons[0]
$



Best,
Fei

Re: dominant choice

PostPosted: Mon Feb 07, 2022 7:38 am
by Michiel Bliemer
You mention that you do not know the signs of the parameters but you do have dominant alternatives. If you are able to assess dominance then you must know the contribution to utility of each alternative as otherwise it is not possible to assess dominance. In other words, you either need to specify the signs of more parameters, or you need to impose other constraints.

You mention "cost" a few times, but there is no cost attribute. Perhaps "payment" is cost, but you also mention that the "payment" should at least cover the "cost" and that a higher "payment" should not have a lower "cost", so it is unclear to me what is the "cost" attribute? If "nrate" is the cost, then perhaps you need to use something like

bnrate.dummy[-0.0001] * nrate[1,0] ? 0 = no cost (base), 1 = cost

If you really do not know the preference order of the attribute levels, then you can impose conditional constraints via ;cond or reject constraints via ;reject. If you can clarify your attributes, I can assist.

Michiel

Re: dominant choice

PostPosted: Tue Feb 15, 2022 5:28 am
by hefei0913
Hi Michiel,

Thank you very much for your reply! I used "reject" to impose conditional constraints and it worked!

Best,
Fei