The design results are not in accordance with the actual

This forum is for posts covering broader stated choice experimental design issues.

Moderators: Andrew Collins, Michiel Bliemer, johnr

The design results are not in accordance with the actual

Postby fengjie » Tue Mar 22, 2016 1:07 pm

Dear all,
I still have some questions with the design result when i conduct my experiment.
After the design, I found that some of the designed alternative was not in accordance with the actual, and then how to choose the result?
I do not know if you have understand my question.
For example, there are two attributes, fare and time, and these two attributes have three levels, fare[500,750,1000],time[1.5h, 2.5h, 3.5h], and there are two alternatives.
In reality, alternative with fare for 500& corresponds to 1.5h, but some of the designed alternative were not in accordance with the actual, what should I do?
Thanks in advance for your suggestion.
Feng
fengjie
 
Posts: 4
Joined: Sun Jan 10, 2016 6:25 pm

Re: The design results are not in accordance with the actual

Postby johnr » Sat Apr 02, 2016 7:32 am

Hi Feng

This is the difference between stated preference (SP) and revealed preference data (RP). In RP data, the levels are those in the real market. In SP, you are creating scenarios where the level combinations form hypothetical situations. The thing to not is that in econometrics, you need variation (in the independent variables - IV) to explain variation (in the dependent variable - DP). SP allows you to create the variation in the attributes that you can use to test how the attributes (the IVs) impact choice (DP). You also need to understand that if the levels of cost and time are perfectly correlated, you won't be able to estimate a model with both as the two are exactly the same. So you would never do what you want say you want to do.

If you are looking at producing more realistic tasks, then you can nest the attributes so that 500 will be linked to times 1.5 and 2.5 (still some variation), 750 will be linked to all 3, and 1000 is linked only to 2.5 and 3.5 hours. You can do this via constraints which is explained the manual.

John
johnr
 
Posts: 168
Joined: Fri Mar 13, 2009 7:15 am


Return to Choice experiments - general

Who is online

Users browsing this forum: No registered users and 3 guests

cron