Dear collegues,

As PhD student, I want to set up a Discrete Choice Experiment in a consumers survey. The main objective is to measure consumer willingness to pay for multi-labelled seafood products. Following focus group and expert opinion, my attributes and their corresponding levels will be the following :
Attributes Level
Price Market price; +10%; + 20%; +30%
Labels None; Ecolabel; Health Allegations; Both labels
Production Method Wild cod; Raised salmon; Wild seabass; raised seabass

My full factorial desing is composed by 64 attribute combinations (4*4*4). To generate it, i used the following code

Code:
matrix attrib = 4,4,4
genfact, levels (attrib)
list, separator (4)
I obtained a table which present the entire combination.

I therefore have to select a fractonnial factorial design for my survey. To do so, i used the following code:

Code:
rename x1 price
rename x2 labels
rename x3 species

recode price (1=0) (2=1) (3=2) (4=3)
recode labels (1=0) (2=1) (3=2) (4=3)
recode species (1=0) (2=1) (3=2) (4=3)

******* Test de Defficient *******

matrix b = J(1,9,0)
dcreate i.price i.labels i.species, nalt(2) nset(12) bmat(b)
I aim to propose to each individual in my survey to choose between two options and one output option. Each individual will have to do this operation three times.


I don't understand how to select the fractionnal design with the higher Defficenty ... Each time I run the last two lines of my code, the Defficienty score changes. I am confused about the procedure to follow. And I haven't introduced the opt-out yet for the moment in my design ...


Sorry for these questions, I am just starting to get interested in these issues of DCE.

If you want more information to clarify my position, i remain available

Best Regards

Jean-François DEWALS