Route choice with fixed time and cost

[prashmg]

Hi Prof,

I am new to choice modeling, and currently trying to design a labeled experiment for calculating WTP values for different charging infrastructures as well as preferences.

The alternatives are Routes with Level 2, Routes with Level 3 charging, and dynamic wireless charging. I am trying to design two experiments according to length of the route 100 mi vs 30 mi and scenarios varying ACC. to EV's initial state of charging, say 80%, 50% and 30%.

For 100 miles,
The attributes are:
SOC(80%,50%,30%)- The time and cost vary according to initial SOC level, so I haven't specified them in the utilities. Should I add them?
1. Charging Time : (Fixed for the length for each alternative, and for each SOC)
2. Charging Cost: (Fixed for each alternative)
3. Coverage of charging stations: (25%,50%,75%) as of conventional gas stations
4. Serving speed: 5,15,30 minutes (Time to wait for charging)
5. Presence of restrooms and convenience store: Dummy variable (0/1)
6. Presence of entertainment options: Dummy variable (0/1)

My current design looks like this (for 100 mi)

design
;alts = level2, level3, dwpt
;eff=(mnl,d)
;rows=12

; cond:
if(level2.TT=240, level3.TT3=15 AND level2.C=1 AND level3.C3=1.5 AND dwpt.Cd=2),
if(level2.TT=300, level3.TT3=20 AND level2.C=2 AND level3.C3=2.5 AND dwpt.Cd=3),
if(level2.TT=400, level3.TT3=30 AND level2.c=3 AND level3.C3=3.5 AND dwpt.Cd=4)

;model:
U(level2)= b2*TT[240,300,400]+b3*C[1,2,3] + b4*Density[25,50,75] + b5*Servsp[5,15,30] + b6*RP[0,1] + b7*S[0,1] /
U(level3)= c1+ b2*TT3[15,20,30]+b3*C3[1.5,2.5,3.5] + b4*Density[25,50,75] + b5*Servsp[5,15,30] + b6*RP[0,1] + b7*S[0,1] /
U(dwpt)= c2 +b3*Cd[2,3,4] + b4*Density[25,50,75] + b5*Servsp[5,15,30] + b6*RP[0,1] + b7*S[0,1]

$

However, NGENE returns the following error: No valid design has been found after 1000 evaluations. There may be a problem with the specification of the design. A common problem is that the choice probabilities are too extreme (close to 1 and 0), perhaps because some or all of the prior values are too large......

My few questions are:
1. There are too many constraints because for each scenario the time and cost for each alternative are fixed, is there a way to relax these constraints? Also, is there a way we can vary the time and cost within the alternatives so that we can get WTP values for each alternative, rather than a generic one which we will get according to my specification?
2. Is the design code feasible?

Best,

[Michiel Bliemer]

Because of your conditional constraints it is not possible to estimate constants c1 and c2 because they are no identifiable because you have created some perfect confounding in your constraints. If you remove c1 and c2 Ngene can find a design, or if you remove the constraints Ngene can find a design. If parameters are not identifiable, you will not be able to estimate the model.

Michiel