constrain design

[Rex Zhang]

Dear professor
I am generating an choice set with six attribution.
A,B and C are three policy attributions.D is the policy cost sharing. E is the policy effect. F is the willingness to pay.
A=1 means to implement this policy,A=0 means not to implement this policy.
B=1 means to implement this policy,B=0 means not to implement this policy.
C=1 means to implement this policy,C=0 means not to implement this policy.
E=0 means to brings 25% improvement,E=1 means to brings 50% improvement,E=0 means to brings 75% improvement,
The levels of payment attributes are 50, 100, 150, 200 respectively
My ;cond: command is to prevent A=B=C=0 from appearing in alt1 and alt2,but the new question is coming.
alt1 alt2
A 1 1
B 1 0
C 1 0
D 1 0
E 1 2
F 3 0


Like the choice set above.
alt1.A=alt1.B=alt1.C=1,
alt2.A=1,alt2.B=0,alt2.C=0
alt2.E>alt1.E,
this is obviously unreasonable.
So I went ahead and added the ;cond: comand.
;cond:
if(alt1.A+alt1.B=0,alt1.C=1),
if(alt1.A+alt1.C=0,alt1.B=1),
if(alt1.B+alt1.C=0,alt1.A=1),
if(alt2.C+alt2.B=0,alt2.A=1),
if(alt2.C+alt2.A=0,alt2.B=1),
if(alt2.A+alt2.B=0,alt2.C=1),
if(alt2.A+alt2.B+alt2.C=3,alt1.E<alt2.E),
if(alt1.A+alt1.B+alt1.C=3,alt1.E>alt2.E)
But new problem arise.
alt1 alt2
A 0 0
B 1 1
C 1 0
D 2 1
E 0 1
F 0 3

Like the choice set above
alt1.A=0,alt1.B=alt1.C=1,
alt2.A=0,alt2.B=1,alt2.C=0
alt2.E>alt1.E,
this is also obviously unreasonable.
What should I do to make the policy attributes and policy effects of my choice set more reasonable?
Thank you for your help,
Rex



The first code is as follow:

Code: Select all

design
;alts = alt1*, alt2*
;rows=12
;block=2
;eff=(mnl,d)
;cond:
if(alt1.A+alt1.B=0,alt1.C=1),
if(alt1.A+alt1.C=0,alt1.B=1),
if(alt1.B+alt1.C=0,alt1.A=1),
if(alt2.C+alt2.B=0,alt2.A=1),
if(alt2.C+alt2.A=0,alt2.B=1),
if(alt2.A+alt2.B=0,alt2.C=1)
;model:
U(alt1) =
 b2.dummy[0.0001]*A[1,0]
+b3.dummy[0.0001]*B[1,0]
+b4.dummy[0.0001]*C[1,0]
+b5.dummy[0.0001|0.0002] * D[1,2,0]
+b6.dummy[0.0001|0.0002] * E[1,2,0]
+b7.dummy[-0.0001|-0.0002|-0.0003]*F[1,2,3,0]/
U(alt2) =
b2.dummy[0.0001]*A
+b3.dummy[0.0001]*B
+b4.dummy[0.0001]*C
+b5.dummy[0.0001|0.0002] * D
+b6.dummy[0.0001|0.0002] * E
+b7.dummy[-0.0001|-0.0002|-0.0003]*F$

[Michiel Bliemer]

Sorry but I do not understand why your attribute level combinations are unreasonable. Are you worried about attribute level overlap? Are you worried about dominant alternatives?

Michiel

[Rex Zhang]

I'm sorry for not making it clear.
For example, in this choice set:
Alt 1 Alt 2
A 1 1
B 1 0
C 1 0
D 1 0
E 1 2
F 3 0
Alt 2 adopts only Policy A, while Alt 1 adopts three policies: A, B, and C.
Policy E represents the governance effectiveness of the policies.
Although Alt 2 adopts fewer policies, its governance effectiveness (Policy E) is higher than that of Alt 1. Both Alt 1 and Alt 2 include Policy A, but Alt 1 also incorporates Policies B and C. Since Policies A, B, and C all contribute to improving governance effectiveness, the scenario raises the question: Is it reasonable for Alt 2, with fewer policies, to achieve better governance effectiveness than Alt 1?
Thanks for your help
Rex

[Michiel Bliemer]

So you mean something like:

;cond:
if(alt1.A+alt1.B+alt1.C > alt2.A+alt2.B+alt2.C, alt1.E >= alt2.E)

Ngene gives you a lot of flexibility to formulate constraints, but it is up to you to formulate exactly what you want.

Michiel

[Rex Zhang]

Dear Professor,

Thanks for your help.

Regards
Rex