choice-metrics.com

Hi All,

I am a very new user of NGENE and never did SP design before. Therefore, the questions may be a bit naive.

I am trying to design a Stated Preference experiment of 6 alternatives (Bus, train, shuttle, taxi, car, bike) and 5 attributes (travel cost, time, transfers, congestion, parking cost). The levels are different for different alternatives for same attributes. Some of the parameters are generic and some are alternative specific.

1. When I have completed the design, I found lots of dominant alternatives in 3-4 choice tasks! Is there any way to get rid of them ? (I know it will reduce if I increase the levels) But, can I change it manually once the design is completed?

2. I have used efficient design to generate 8 choice tasks. But, I want to pivot the design based on some other features. For example, the parking cost is 10$/day and when the respondent answers he is going to park for 7 days (through RP) then the SP will show parking cost is 70$ in his choice task. Another example, the bus travel cost is 5$/person. If the respondent answers they traveled 3 people together than the travel cost of bus will show three times higher in each choice task. Is it okay to do this kind of pivoting?

3. I am going to do online survey. Is there any online survey tool which I can use to capture this pivoting?

4. How can I design a survey if I want to show 4 alternatives out of 6 in each choice task? I am considering 8 choice tasks (number parameters/(no of alternatives-1)) using the equation. When I will consider only 4 alternatives for each choice task do I have to increase the number of choice tasks?

5. If I have 8 choice tasks and want to present 4 to an individual, how do I select which 4 should be presented?

Thanks in advance.

Best,
Pasha

1. Yes feel free to remove choice tasks that you think do not make sense. It is important to have realistic choice tasks.

2. Yes you can create such respondent-specific attribute levels to make the choice tasks more familiar/realistic.

3. Stated choice experiments are not supported by most survey instruments. I could not tell you which tools would support pivoting. We program all our online surveys ourselves without any commercial tools.

4. This is called an availability design or a partial choice set design. It is possible to generate such designs with a slightly modified version of Ngene (and also with the new version Ngene 1.2 that we intend to release very soon). In the equation you will need to assume 4 alternatives yes, so this means that the number of choice tasks may need to increase.

5. This is called blocking. In Ngene you can use ;block = 2 to do this, although it is never perfect. What you want to achieve is to have all attribute levels appear as equally as possible across each block such that each respondent sees high and low levels of each attribute.

Michiel

Thanks very much Michiel.

Just following up with your answers:

3. Is there any reading material where I can have a look to develop my own online SP survey instruments? or Can you give me some insights about this? (if possible)

4. I am using Ngene 1.1.2. Is it possible with this version? and Is there any manual for this type of design?

Thanks in advance.

Best,
Pasha

3. I am usually not involved in the implementation. It either involves HTML coding or using commercial software. There is SurveyEngine that specifically does stated choice experiments that you may wish to look into.

4. Yes please email contact@choice-metrics.com and request for the special Ngene version that can read in candidate sets. We also have an example project that we can share and draft manual text.

Michiel

Thanks very much Michiel.