Would you actually choose it? Dual response interpretation

[tomschuette]

Dear all,

I have seen some papers in the past that used forced choices (no opt-out or status quo) and afterwards asked the respondent whether they would actually make this choice in reality. How would I include such a yes/no question in a mixed logit model and how would I interpret the parameter?

If anyone knows a paper where this is done, I would be very grateful.

Thanks in advance

[Michiel Bliemer]

I believe it is quite common to ask for forced and unforced choices. It requires creating two choices, namely:

forced choice: alt1, alt2
unforced choice: alt1, alt2, neither

Essentially, you duplicate all observations, and in the data, you use the same attribute levels for alt1 and alt2 for both observations, the only difference is that in the unforced choice you have 3 alternatives instead of 2. How you set this up in the data depends on the software you use. In Biogeme and Apollo you can define the availability of an alternative, so you would set the availability of alt1 and alt2 to 1 and the availability of alt3 is a column in the data that is zero or one.

In the utility functions, you specify all 3 alternatives, and at most have a constant for the 'neither' option (or if you add constants to alt1 and alt2, you normalise neither to zero). The interpretation of the constant is similar with the interpretation of any constant, namely the relative preference for a labelled alternative, in this case the optout.

Michiel

[johnr]

This was done a lot in transportation studies in the early 2000s-2010s. The reason was that many studies during this time involved route choices examining toll road uptake. Many of the study areas did not have toll roads at the time, and it was uncertain if respondents drawn from the relevant catchment areas would select toll road options when completing SP experiments. For this reason, researchers often had respondents complete dual response questions, so as to obtain forced trade-offs between toll and non-tolled roads in the second choice. We discuss the this in the following paper.

Rose, J.M. and Hess, S. (2009) Dual Response Choices in Reference Alternative Related Stated Choice Experiments, Transportation Research Records, Paper #09-2432, Vol. 2135, 25-33.