choice-metrics.com

Dear all,

i´ve been runnig in some issues to create a d-efficient design for the following stated choice experiments using the software package Stata.

SCE for 4 labeled alternatives (first { } referes to alt1, second { } to alt2 and so forth):

Attribute1: {2, 7, 14}; {0} ; {0, 1, 2}; {0}
Attribute2: {0}; {2, 4}; {0}; {0}
Attribute3: {10, 20, 60}; {0}; {0}; {0}
Attribute4: {0}; {2, 3, 5}; {1, 2, 3}; {1, 2, 3}
Attribute5: {30} ; {16, 24, 32}; {80, 90, 100}; {80, 90, 100, 122.5, 125}


The goal is to present each respondent a block of 6 choice sets. Each Prior can be assumed as a very small negative.

Unfortunatly, I have no experience with Ngene. That is why I would like to ask if Ngene is capable of creating a d-efficient design for this particular SCE. And if so, is the code rather trivial or is it more complex?

Many thanks in advance,

Jonas

That would be fairly straightforward in Ngene. If I understand your setup correctly, the following script in Ngene would generate the required design (assuming that all variables are numerical, if some are categorical then dummy coding is required):

Code: Select all

design
;alts = alt1, alt2, alt3, alt4
;rows = 30
;block = 5
;eff = (mnl,d)
;model:
U(alt1) = asc1[0]
        + b1[0]   * A1[2,7,14]
        + b2[0]   * B1[0]
        + b3[0]   * C1[10,20,60]
        + b4[0]   * D1[0]
        + b5[0]   * E1[30]
        /
U(alt2) = asc2[0]
        + b1      * A2[0]
        + b2      * B2[2,4]
        + b3      * C2[0]
        + b4      * D2[2,3,5]
        + b5      * E2[16,24,32]
        /
U(alt3) = asc3[0]
        + b1      * A3[0,1,2]
        + b2      * B3[0]
        + b3      * C3[0]
        + b4      * D3[1,2,3]
        + b5      * E3[80,90,100]
        /
U(alt4) = b1      * A4[0]
        + b2      * B4[0]
        + b3      * C4[0]
        + b4      * D4[1,2,3]
        + b5      * E4[80,90,100,122.5,125]
$

I used zero priors in the above. Using small positive or negative priors to indicate the sign of the parameter is only useful when checking for dominant alternatives (which typically do not exist in a labelled experiment) so you may as well be using zero priors in that case until you have done a pilot study.

I used 30 rows since you have attributes with 2, 3, and 5 levels. 30 is divisible by 2, 3, and 5 and hence the design will be attribute level balanced. I used 5 blocks because that will give you blocks of 6 choice sets.

Michiel

Dear Michiel,

thanks you so much for your reply and your uselful tips.