[Alireza Rh]

Dear Arne,

Is it possible to determine combinations that cannot be presented as packages? Let say we have two attributes each with 3 levels X1= (e,f,g) and X2=(a,b,c). Can we exclude the X1=e & X2= b in the choice sets?

Thanks in advance for your support.

[Arne Risa Hole]

Hi Alireza,

In response to your two questions in posts #75 and #76:

- #75: I am afraid that is not possible to do with dcreate.

- #76: Yes, you can remove alternatives with certain combinations of attribute levels by simply removing them from the candidate set before running dcreate.

Arne

[Markus Hendsler]

Dear Arne,

thank you so much for this package and your related work!

I plan to conduct a DCE with both continuous and categorical variables as part of my Master's project to examine consumer's preference between external cues in an e-commerce environment.

I would like to use categorical variables x1, x2, x3 (2, 2, 3 levels; x1 free shipment, x2 free return, x3 sustainability) together with two continuous variables x4 (product Rating with "3", "4", "5") and x5 (allowed return policy in days with "14", "30", "100").
I am very much interested in measuring effects such as one additional unit of x4 (product Rating) on the preference for x3 (categorical, sustainability) and vice versa the "value" of x2 (free shipment) expressed in units of an additional day of x5 or one additional Rating unit in x4.

There are no interaction terms.

Here are my questions:

(1) When creating the coefficient matrix, i believe that the coefficients for the categorical variables x1, x2 and x3 should be 0, right?
(2) How can i best determine the coefficients for the continuos variables x4 and x5, especially as they are on two different scales?
(3) In the dcreate command, i start with i.x1 i.x2 i.x3 c.x4 c.x5 right?

Again, thank you so much - your research and the associated literature helped me so much already, these are my final questions before starting the design.

Best regards
Markus

[Arne Risa Hole]

Hi Markus,

I am pleased to hear that you have found dcreate useful. In response to your questions:

- Unless you have any prior estimates of the coefficients (e.g. from a pilot study) I would recommend setting them all to zero.

- I would recommend specifying all of the attributes to be categorical (i.x1 i.x2 i.x3 i.x4 i.x5) as this gives you more flexibility later - you can still model the attribute as continuous if you wish.

Arne

[Faizan Ahmd]

Dear Arne,

Thank you for dcreate. I have a question. My model has 3 alternatives where each alternative has a ASC. I am trying to understand your example 3 but unable to implement it in my following model:

V1= b0 + b1x1 +b2x2 +b3x3
V2= b4 + b1x1 +b2x2 +b5x5
V3= b6 + b1x1 +b7x7

Can you share the syntax?
All of my variables are continuous. I read your previous comments that even continuous variables should be considered as dummy for flexibility. Can you share some more light on this?

[Chris Weinmann]

Hi Arne,

thanks for creating the dcreate command. Somehow my dcreate command always results in a syntax error, even with the example values.
Do you have an idea why this is the case?
Thanks!!
Chris

[Adam Ray]

Hi Arne,

Thanks for putting this package together!

I have one question for you. When trying to do the experimental design in SAS, there is a macro called 'mktruns' which outputs potential sizes for the experimental design. My question is, is there an equivalent for dcreate? And if not, how does one choose the appropriate value for 'nset' or compare different designs (by the number of choice sets).

Thanks in advance!
Adam

[Shira Haderlein]

Hi, Arne!

Thanks for creating such a useful package and for providing so much support in using it! This is my second time using dcreate to design a DCE and I am running into some issues.

I have 6 attributes, with the following levels: matrix levmat= 3,3,2,2,2,5

First, I am having trouble figuring out the correct number of coefficients to be estimated (in the b matrix). I know you describe this in post#9, but I don't quite follow your logic there.

Second, I run my command as follows:
matrix b = J(1,11,0)
dcreate i.price i.numratings i.avgrating i.align i.tguide i.format i.comments , nalt(2) nset(504) bmat(b)

and get the following error message:
The random starting design does not identify all of the effects to be estimated.
Possible options include:
- Simplify the model
- Increase the number of choice sets in the design
- Run dcreate again specifying a different random number seed


Any idea what is going on there? Any advice you have would be so useful!

Thanks!
Shira

[sebastian zimmer]

Hello together,
I also applied the DCREATE function for a stated preference experiment in my PHD. I am wondering how I can calculate the D-error with Stata and whether there is a way to obtain the Fisher-Information Matrix.
Thank you so much in advance,
Sebastian

[Maja Keppler]

Dear Arne,

thank you for creating dcreate. It is really helpful for my master thesis!

I have one question. Is it somehow possible to create a design with a fixed alternative with only one level for one attribute that is held constant?
I have chosen four attributes (2*4 levels and 2*3 levels) and I would like to have 3 options and "opt-out".
Ideally, option 1 would always be level 1 (out of 4) for attribute 1 but at random levels for attribute 2-4. Is that even possible?
Thank you in advance!

Best
Maja

[Alene Yong]

Hi Arne,

I am trying to include prior values from the pilot study into my choice-set. One of my variables (x7) is "cost" with 4 levels. However, when I create "cost" as a continuous variable, the choice sets generated only included the highest and lowest levels (no in-between) for for the cost variable.

The commands are as follows:

matrix levmat = 3,3,3,3,3,3,4
genfact, levels(levmat)
list, separator(2)
matrix b = 0.55,-0.86,-0.09,-2.18,-0.04,-1.73,0.72,-0.45,0.02,2.13,-2.11,-1.38,-0.001
dcreate i.x1 i.x2 i.x3 i.x4 i.x5 i.x6 c.x7, nalt(2) nset(16) bmat(b)


My question is:
Am I supposed to specify "cost" as categorical variable instead? If I do so, how am I supposed to input its prior value (since when I run regression for pilot study, it only generated one coef for cost)?

Thank you in advance for your help!

[Bryan Parthum]

First, thank you Arne for creating this very useful tool.

I have a question regarding dominating choice alternatives. When I run the following code 5 out of the 12 choice tasks created contain strictly dominating alternatives.

matrix levmat = 3,3,3

genfact, levels(levmat)

matrix optout = J(1,3,1)

matrix b = J(1,7,0)

dcreate i.x1 i.x2 i.x3, nalt(2) nset(12) fixedalt(optout) bmat(b) asc(3)

blockdes block, nblock(2)

list, separator(3) abbreviate(16)

I have two questions:

1. I noticed from your excellent presentation from the Nordic and Baltic Stata Users Group meeting on dcreate (https://www.stata.com/meeting/nordic...way16_hole.pdf) that your created choice sets on slide 20 also contain two dominating alternatives (choice_set 3 and 6), but I wonder if it can be prevented in dcreate? (The ngene manual states that in that program identical and dominating alternatives will be excluded.)
2. I would prefer not to include any choice sets where one alternative dominates the other. I have tested to increase the number of levels and/or reduce the number of choice sets in nset(), but dominating alternatives remain. Suggestions I have seen are to remove dominating choice sets by hand, or to by hand change levels to remove dominance. What would be your recommendation?

Again, thanks for a great program.

Henrik

Hi Henrik, I'm sure this is a little late as you posted this question in 2019, but for others who are looking to remove dominated alternatives (or any unwanted combinations) you can simply remove those from the candidate design and then continue with `dcreate` as usual. Doing so drops those combinations from the full factorial and they will no longer be considered in the design. For example:

matrix levels = 4, 4
genfact, levels(levels)
drop if (x1 == 1 & x2 == 1)

[Francesca DAddia]

Hi, thanks a lot Arne for creating this very useful tool!

My analysis comprises one attribute with 6 levels, one attribute with 2 levels, and the last one also with 2 levels. All are categorical variables and I wrote "matrix b = J(1, 23, 0)".
I would like to know if I made it right. I need it because I have to create 10 vignettes for respondent (for factorial analysis), in which the structure is orthogonal.

ssc install dcreate
matrix levmat = 2,6,2
genfact, levels(levmat)
matrix b = J(1, 23, 0)
dcreate i.x1##i.x2##i.x3, nalt(6) nset(16) bmat(b)
list, separator(10) abbreviate(16)

Any advice?

Thank you in advance.

Francesca

[Francesca DAddia]

Hi, thanks a lot Arne for creating this very useful tool!

My analysis comprises one attribute with 6 levels, one attribute with 2 levels, and the last one also with 2 levels. All are categorical variables and I wrote "matrix b = J(1, 23, 0)".
I would like to know if I made it right. I need it because I have to create 10 vignettes for respondent (for factorial analysis), in which the structure is orthogonal.

ssc install dcreate
matrix levmat = 2,6,2
genfact, levels(levmat)
matrix b = J(1, 23, 0)
dcreate i.x1 i.x2 i.x3, nalt(6) nset(16) bmat(b)
list, separator(10) abbreviate(16)

Any advice?

Thank you in advance.

Francesca

[Rune Lomholt]

Hello. I am doing vignette experiment where my outcome variable is the Danish grade scale (-3, 00, 02, 4, 7, 10, 12) which is equivalent to the international ECTS scale (F, Fx, E, D, C, B , A). Can I use dcreate to create ad-efficient design for a study with this type of outcome? In terms of analysis technique I was going to use mixed linear regression, so not logistic regression.