choice-metrics.com

Good morning,

I am creating a choice experiment where I show monthly savings in terms of the share of the monthly electricity bill (in percent). I would then like to know whether showing these same savings in monetary values makes a difference in stated preferences. My idea was to show the same choice experiment to all respondents and add an additional row with "monetary savings" to a randomly selected treatment group. The idea comes from this paper: https://www.sciencedirect.com/science/a ... 8323007752.

I would then include the percentage attribute in my model while omitting the "monetary savings" treatment attribute due to perfect colinearity.

An alternative idea was to replace the percentage values with monetary values in the attribute "monthly savings" instead of additionally showing the monetary values in a new row.

My question is whether this is feasible and if anyone maybe knows of another study where something like this was done.

Thanks in advance
Tom

You can only make an appropriate comparison if the information you are showing is consistent and complete in both cases. You will need to ask respondents about the cost in their last or typical electricity bill in the first part of the survey, say that they report X. Then you can apply either percentages or absolute values as monthly cost savings in your experiment.

For example, if your choice experiment contains savings levels 10%, 20%, 30%, then in the choice experiment you will either show these percentages or you show absolute values that are calculated within the survey platform as 0.1*X, 0.2*X, 0.3*X. That way, the information you are providing in the choice experiment is the same, except that the framing of the levels is different and you can analyse whether there are any framing effects.

Michiel