Design which includes WTP and WTA tasks sequentially

[izakro]

Dear all,
We are trying to estimate the Value of Lost Load (VoLL) for the Ministry of Energy.
The Ministry is interested in both values of WTA (for the interruption in electricity supply) and WTP to avoid it.
The question is: does it make any sense to ask the same respondents to answer, say, 3 scenarios of WTP followed by 3 of scenarios WTA
(the order can be determined randomly).
I have noticed that some studies have taken this approach but I'm not sure.
Thanks in advance,
R.

[Michiel Bliemer]

WTP consists of asking decision-makers how much they are willing to pay to avoid a negative or to accept a positive outcome, while WTA goes for compensation and asks how much a decision-maker would like to be paid to accept a negative outcome or to forego a positive one. Whether you ask for WTP or WTA is a matter of framing.

Framing effects often occur, for example you can frame a choice task in such a way that you say something in a positive way ("your health with increase") or in a negative way ("you have a lower risk of dying"). Framing may influence choice behaviour. So WTA and WTP could be considered as asking the same question using positive or negative framing and they may give different results.

If you are interested in investigating what is the best way to convince consumers/producers to get the best outcome, then you could consider both WTA and WTP. However, if the Ministry of Energy already has a specific framing in mind to explain things to consumers/producers, then perhaps best use that. For example, in the transport domain, it is more natural to speak of WTP a toll to save travel time instead of WTA more travel time at a lower toll cost.

Michiel

[izakro]

Thank you Michiel. That helped a lot.

[izakro]

Hi,
Based on the discussion on whether to use WTA or WTP framed designs to estimate the value of lost load, we ran a pilot which indicated extremely high values for the WTA approach.
We therefore decided to cont. with the WTP experiment.
The design below is a Bayesian efficient one, which provides the respondents with two generic profiles of electricity supply, each characterizes by different number of shutdowns (per year), different duration per shut down, the period in the day (morning, evening, night), whether it will be on weekend/weekdays, whether there will be an announcement before the shutdown, and the cost that the respondents will have to bear (per bill period) in order to avoid this profile and have no planned disruptions (based on the pilot we know what is the preference order regarding morning, evening, night, and weekend/weekdays)

The question is how to describe the status quo option. Since there is no certainty regarding what will be the extent of disruption in electricity supply, we can either add a status quo option which comprised the 'worst' level of all attributes (max shutdown, max duration, morning, weekdays and no alert) which will be fixed for all meus, or use the max level across the other two alternatives in each menu. Perhaps there are other options for the status quo which we haven't thought about.
Here is an example menu without the status quo option [img]C:\Users\Anat\Dropbox\PC%20(2)\Downloads\Picture2.jpg[/img]
and this is the design

Code: Select all

design ;alts = optiona, optionb
;rows = 36
;block = 6
;eff = (mnl,d)
;bdraws = halton(500)
;alg = mfederov(candidates = 750)
;reject:
optiona.shutdowns=1 and optiona.cost=296,
optiona.duration= 15 and optiona.cost>152,
optiona.duration= 60 and optiona.cost>200,
optiona.duration>= 120 and optiona.cost<224,
optionb.duration= 15 and optionb.cost>152,
optionb.duration= 60 and optionb.cost>200,
optionb.duration>= 120 and optionb.cost<224,
optionb.shutdowns=1 and optionb.cost=296
;model:
U(optiona) = cost[(n,-0.0304, 0.0494)] * cost[104, 128,152, 176, 200, 224, 248, 272, 296]
         + duration[(n,-0.320, 0.01)]  *  duration[15,60,120, 240]
         + shutdowns[(n, -0.418, 0.078)] * shutdowns[1,2,3]
         + dayperiod.dummy[0.231|0.636] * dayperiod[2,3,1] ? 1 =evening , 2 = morning, 3= night
         + weekend.dummy[-0.727] * weekend[0,1] ? poeple prefer to have electricity on weekdays
          + alert.dummy[0.630]*alert[1,0]   
        /
U(optionb) = cost* cost
         + duration* duration
         + shutdowns* shutdowns
         + dayperiod* dayperiod       
         + weekend * weekend
         + alert*alert               
;formatTitle = 'Scenario <scenarionumber> <blocknumber>'
;formatTableDimensions = 3, 8
;formatTable:
1,1 = ''/
1,2 = 'Cost in USD' /
1,3 = 'Average duration per shutdown in minutes'/
1,4 = 'Number of shutdowns in a season'/
1,5 = 'Part of the day in which shutdown may occur' /
1,6 = 'Shutdown will occur in weekdays or weekend'/
1,7 = 'Shutdown will be announced'/
1,8 ='Choice question&:' /
2,1 = 'Option A' /
2,2 = '<optiona.cost>' /
2,3 = '<optiona.duration>' /
2,4 = '<optiona.shutdowns>'/
2,5 = '<optiona.dayperiod>' /
2,6 = '<optiona.weekend>' /
2,7 = '<optiona.alert>' /
2,8 = '<choice 1>' /
3,1 = 'Option B' /
3,2 = '<optionb.cost>' /
3,3 = '<optionb.duration>' /
3,4 = '<optionb.shutdowns>' /
3,5 = '<optionb.dayperiod>' /
3,6 = '<optionb.weekend>' /
3,7 = '<optionb.alert>' /
3,8 = '<choice 1>'
;formatTableStyle:
1,1 = 'default' /
1,2 = 'headingattribute' /
1,3 = 'headingattribute' /
1,4 = 'headingattribute' /
1,5 = 'headingattribute' /
1,6 = 'headingattribute' /
1,7 = 'headingattribute' /
1,8 = 'headingattribute' /
2,1 = 'heading1' /
2,2 = 'body1' /
2,3 = 'body1' /
2,4 = 'body1' /
2,5 = 'body1' /
2,6 = 'body1' /
2,7 = 'body1' /
2,8 = 'choice1' /
3,1 = 'heading2' /
3,2 = 'body1'  /
3,3 = 'body2' /
3,4 = 'body2' /
3,5 = 'body2' /
3,6 = 'body2' /
3,7 = 'body2' /
3,8 = 'choice2'
;formatStyleSheet = Blue buttons.css
;formatAttributes:
optiona.cost(104=#, 128=#, 152=#, 176=#, 200=#, 224=#, 248=#, 272=#, 296=#) /
optiona.duration(15=#,60=#,120=#, 240=#) /
optiona.shutdowns(1=#,2=#,3=#) /
optiona.dayperiod(1=Evening, 2=Morning, 3=Night) /
optiona.weekend(0=weekdays, 1=weekend) /
optiona.alert(1=No, 0=Yes) /
optionb.cost(104=#, 128=#, 152=#, 176=#, 200=#, 224=#, 248=#, 272=#, 296=#)/
optionb.duration(15=#,60=#,120=#, 240=#) /
optionb.shutdowns(1=#,2=#,3=#) /
optionb.dayperiod(1=Evening, 2=Morning, 3=Night) /
optionb.weekend(0=weekdays, 1=weekend)/
optionb.alert(1=No, 0=Yes)
$


Thanks much
R.

[Michiel Bliemer]

Willingness to accept can indeed be very high because there is no limit on what people are willing to accept (as opposed to a clear limit to what people are willing to pay).

Is including a status quo alternative a must? If there is no particular need for it, you could simply omit it. If the attribute levels are not clear, it may only be confusing. Alternatively, you can make it a respondent-specific reference alternative, where you ask for the levels of each attribute in the first part of the questionnaire and simply present these back to the respondent as their individual-specific status quo alternative in the choice experiment. You could optimise the design around average levels but in the survey instrument show the levels that the respondents provided.

Michiel

[izakro]

Thanks Michiel.
There is no particular need for a status quo actually. I was thinking that reviewers might comment on its absence.
The only issue I have with using respondent-specific reference alternative, is that current levels of blackouts in the country are not representative of future levels (it will probably get worse without voluntary participation in such plans).
So I guess that the solution is not to include a status quo

[Michiel Bliemer]

You could also consider including blackouts as a scenario variable, where you write at the top of the choice experiment something like "Assume that in the near future the level of blackouts is X. Which of the following options do you prefer?" In that case, you have control over the choice environment and you can add the 'current' level of blackouts as a variable in the utility functions.

Michiel

[izakro]

thank!!!