Yes, but one status alternative with status attribute levels and none-these or neither alternative. This means you can have 4 alternatives: 2 main or competing alternatives, 3rd status-quo alternative and 4the none of these or neither alternative with no levels. So, 4th alternative will serve as a reference alternative, which means you can have 3 alternative specific constants (ASCs) for the 2 main alternatives and for status-quo alternative.

I hope this helps. Good luck.

Cordula Hinkes. Thanks for involving me in the discussion further. This is very interesting as I have never read about it before! I smell that something is wrong with the inclusion of Status-quo in the design!!!

So, I think we can have the following measures:

1. When calculating WTP, we can still use that choice card and using Poe et al test, we can measure WTP equality. Also, we can apply Mann-Whitney test and Kernel density plots to further explain the differences;

2. When calculating CV, we can exclude those respondents falling in that category and estimate separate suitable Discrete Choice models, such as RPL, RPL-EC, LCM, models;

3. We can also use the above strategy to measure preferences equality applying the Convergent Validity using Swait and Louviere (1995) test;

But, I am still sure that there's a problem with

the experimental design ! Under what circumstances, she included Status-quo 3rd alternative in the design using Ngene software. Normally, the Status-quo alternative is not included in the experimental design simply because it has the fixed attribute levels as compared to other two alternatives in this case with varying attributes' levels. So, now it's very simple! Ask her to implement an experimental design based on 2 alternatives with varying levels. After this, simply include a Status-quo alternative with its fixed attributes' levels in each and every choice

card along with the 2 main alternatives.

I think the problem is eventually solved.

Warm regards and best wishes

Cordula Hinkes

I still haven't applied the survey, I am trying to design my choice sets.

Musharaf A. Talpur you said that the Status-quo alternative is not included in the experimental design, so in the example of biodiversity only the alternative 3 (SQ) should have the 10 birds, and alternative 1 and 2 should vary between 20 and 30?

Best regards

I still haven't applied the surveyI still haven't applied the survey I still haven't applied the survey

Cordula Hinkes

I have one more question, its correct to include 0 value as a level of cost attribute?, as I could have improvement alternatives with 0 cost.

Best

Two versions can be implemented:

Version 1. That includes status-quo levels, e.g. 0 cost and

Version 2. That doesn't include status-quo levels.

Finally, in the experimental design don't include Status-quo as an alternative simply because it has fixed levels. If you do so, then status-quo will also take Improvement levels!!! Which is very strange!!

Based on (i) varying attributes' levels including 2 improvement levels and also the fixed or initial level of status-quo, and (i) including 2 competing alternatives, say project-A and project-B, run minimised Orthogonal blocked design. Now, you have a design that involved 2 competing alternatives with varying attributes' levels including the fixed levels of status-quo alternative. Let's say, you have 36 choice cards with 2 alternatives, 3 attributes (i.e. one cost attribute including 0 cost level of status-quo). Divide 36 by 6 = 6 blocks . Each respondent will face 6 choice cards. But, remember in each choice card include status-quo as 3rd alternative with it's fixed or initial levels based on reality.

Afterwards, implement a pilot survey, say 60 respondents for each version using MNL model to generate prior values for the next D-Efficient design. Once again, generate 36 cards blocked into 6 choice cards for each respondent to face.

1. Using version 1 and 2, implement Swait and Louviere (1995) test to check convergent validity of Preferences across both versions,

2. Using both versions, implement Poe et el (2005) test to check WTP equality, and

3. Using both versions, calculate CV (compensating variation) and compare them.

---- All the above mentioned tests will reveal that to what extent Preferences, WTP and CV are affected if we include and exclude the fixed levels of status-quo ! -----

I hope this helps.

Best of luck

As Cordula Hinkes mentioned that there can be a tendency among the respondents to select the dominant (e.g. probably selecting an alternative with Improvements at 0 cost). In contrast, there can also be a tendency using the Version 2 that doesn't include 0 cost in the 2 competing alternatives. So, this will probably result in 'protest' ! In such a case exclude both types of respondents from both versions, and the reapply the tests I mentioned above.

Best

Very simple. If we include status-quo alternative in the experimental design, its levels which are fixed in reality will also vary :-) As a result, status-quo alternative will serve as an another competing alternative. So, rationally, don't include status-quo when running experimetal design in Ngene or Sawtooth.