choice-metrics.com

Hi, i want to create a desing results from the pairwise comparison of just one attribute of 9 levels. The choice cards simply show the first attribute level against all the others (8) and an opt out option. The number of possible pairwise comparison (without any repetition) is 36 if I am not wrong (8+7+6+5+4+3+2+1). I can do that by hand, but I wish to block it (efficiently) in 6 blocks.
Well, Ngene does not find a design with 36 rows (if I reduce the number to 24 it is fine, but obviously there are some pairs that are not compared!).
I use the following simple code:

Design
;alts = alt1*, alt2*,alt3
;rows = 36
;block=6
;model:
U(alt1) =b1.dummy[0|0| 0| 0|0|0|0|0] * A[0,1,2,3,4,5,6,7,8]
/
U(alt2) =b1.dummy[0|0| 0| 0|0|0|0|0] * A[0,1,2,3,4,5,6,7,8]
$

Two things:

1) You need to tell Ngene what type of design you want, a random factorial, orthogonal, or efficient design. Either a random factorial design or an efficient design will give you the design you want.

Add the following to your syntax:
;fact

Or add this to your syntax:
;eff = (mnl,d)
;alg = mfederov

2) Note that blocking has nothing to do with efficiency. Blocking is related to orthogonality and aims to create attribute level balance within each block (which in your case cannot be perfectly satisfied since you only have 6 choice tasks per block).

Michiel

Hi Michiel,
Thanks very much for your help.
Sergio.