[James Cook]

Dear Arne,

Thanks for your wonderful command.

If I only know the signs of my priors (and not the magnitude), should I simply use something like .01 and -.01? Is there a "protocol/best practice" for this?

Thank you!

James

[Arne Risa Hole]

Dear James.

Thanks for the positive feedback.

Is there a "protocol/best practice" for this?

Not that I know of, but I would be interested in hearing about it if anybody knows otherwise. In my own work I have tended to use either zeros or the estimated coefficients from a pilot study as priors.

Arne

[Mwihaki Kihoro]

Dear Arne,

I have used the evaldes command and gotten a D-Efficiency value of 2.06 what does this mean.


Esther

[Arne Risa Hole]

Esther

See post #4 in this thread. The formula used to calculate the D-efficiency is given on slide 9 in this presentation http://www.stata.com/meeting/nordic-...way16_hole.pdf

Arne

[Geir Godager]

Dear Arne,

Thank you for your useful dcreate module. I have a question about evaldes. When evaluating generated designs by means evaldes, I get an error message if the model is introduced with heterogeneity. In the example below, I augment example 2 from the help file by introducing heterogeneity:
Code:

matrix levmat = 4,4,2,2,2,2
genfact, levels(levmat)
matrix b = J(1,11,0)
matrix I11=I(11)
dcreate i.x1 i.x2 i.x3##i.x4 i.x5 i.x6, nalt(2) nset(16) bmat(b) vmat(I11)


The algorithm converge after 10 iterations. When I next run evaldes:

evaldes i.x1 i.x2 i.x3##i.x4 i.x5 i.x6, bmat(b) vmat(I11)

Output:
<istmt>: 3499 dcreate_burn not found
r(3499)

I wonder If I am doing something wrong?

[Arne Risa Hole]

Hi Geir,

You are not doing anything wrong - there seems to be a bug in evaldes that prevents it from evaluating the efficiency of Bayesian designs. I will get this fixed and post back when an update is available.

Arne

[Dimitris Damigos]

Dear Arne,

First of all thank you so much for providing this useful module. I want to design a DCE and I would mostly appreciate your help in clarifying a couple of issues.

The number of choice sets in the design has to be specified by the user. The minimum number of choice sets should be equal or greater than the degrees of freedom of the model. My model has 17 degrees of freedom and thus specifying 18 sets seems to be o.k. (I have some three-level attributes). Nevertheless, the minimum runs required for my model (based on the number of attributes and their levels) when searching for an OMEP to represent the profiles in the first option in the choice sets, is 36.

Do I have to try a number of different candidate sets (starting from the minimum number of choice sets required according to the df of the model) and use the one with the highest efficiency or is it more appropriate to use the minimum number of sets determined by the orthogonal array?

Thank you in advance for your time.

Kind regards,

Dimitris

[Arne Risa Hole]

Dear Dimitris,

There are no hard and fast rules here that I'm aware of. I personally find it "safest" to set the number of choice sets well above the minimum required, so a pragmatic approach would be to set the number to 36 and then block the design into the required number of blocks (depending on how many choices you think it's manageable for a single respondent to complete).

Arne

[Arne Risa Hole]

Thanks to Kit Baum a new version of dcreate is now available on the SSC archive which fixes the bug in evaldes described in posts #20 and #21.

Arne

[Dimitris Damigos]

Dear Arne, thanks a lot for your prompt response!

Dimitris

[Carla Cavallo]

Dear Arne,

thank you, I found dcreate very useful.

I'm aware that the d-efficiency allows to have a measure of efficiency of alternative model specifications, but I'm wondering if there is a way to evaluate the efficiency as the amount of information contained in the design compared with what contained by an orthogonal design.

I will appreciate any help,

Carla.

[Eleanya Nduka]

Arne Risa Hole
Hello Arne,
Thank you for creating the D-efficiency design. I never knew this existed until I read about it in an article. I have read several of your articles on DCE design and models. I really love the way you write. As a beginner in this field I have found your articles very helpful. I have read your presentation on D-efficiency during the Nordic & Balistic meeting and I wish I were there. The step by step explanation is awesome! I should say that I understand the presentation in toto.

However, in the 4 different examples presented in Stata D-efficiency, there is only one thing that I need to understand. I am interested in the 3rd example, but I still don't understand how to form my own b matrix following yours. I understand that you said if priors (coefficients) are not known, you can use zero (which is what I intend to do), but I do no understand how the elements of J(?, ?, ?) were arrived at.

Suppose I have 5 attributes in the order of 2, 2, 3, 3, 4 levels.
Please, how do I form matrix b = J(? ? ?)?
In your own example of 4, 4, 2, 2, 2, 2, you gave J(1, 10, 0). My main challenge is understanding how the three elements 1, 10, 0 were arrived at.


Thank you.

[Arne Risa Hole]

Thanks Eleanya. To answer your question, if all your attributes are categorical your matrix should be "matrix b = J(1, 9, 0)", which is the same as "matrix b = 0,0,0,0,0,0,0,0,0" (see help matrix functions).

See also post #9 in this thread.

Arne

[Eleanya Nduka]

Dear Arne,
Thank you very much

[Maya Gheroufella]

Dear Arne,

First of all, I want to thank you for your helpful command “dcreate”.

I am a PhD student and I am using a discrete choice experiment (DCE) for measuring willingness to pay.

I want to ask about the dimension of matrix of estimators (bmat), since I have 4 attributes x1,x2,x3,x4 and I am assuming interaction effects between x1 and x2. I have no prior information so I set betas equal to zero.
While running the command it said that "There are 10 effects in the design", how this number is calculated ?

Also I am wondering how to introduce the “opt out” option (no purchase) in the command “dcreate” to have two alternative by choice set and the (no purchase) option ?

Maya