Hi,
I have a design I'm trying to create that has 17 brands. I'd like a labeled design with the brands as the alternatives. But with 17 brands, this is not practical. Can I create a partial profile design in which only, say, 5 brands appear in a choice set? There is some literature on this, but I don't see this capability in Ngene.
Thanks,
Walt Paczkowski
partial profiles
Moderators: Andrew Collins, Michiel Bliemer, johnr
-
Michiel Bliemer
- Posts: 2057
- Joined: Tue Mar 31, 2009 4:13 pm
Re: partial profiles
This is something we are working on, but is not available in the current version of Ngene.
What you could do is using the brand as an attribute instead of an altenative by treating it as a dummy variable in an unlabelled altenative.
For example:
U(coca) = b1*x1 + b2*x2 /
U(pepsi) = b1*x1 + b2*x2 /
would then become:
U(cola) = b1*x1 + b2*x2 + b3.dummy*[0,1] / ? where 0 = coca and 1 = pepsi
U(cola) = b1*x1 + b2*x2 + b3.dummy*[0,1]
So you go from a labelled experiment to an unlabelled experiment. This would work well if all attributes appear in all alternatives. However, if some attributes are specific to the brand, then it will not work, as that requires a labelled experiment.
Michiel
What you could do is using the brand as an attribute instead of an altenative by treating it as a dummy variable in an unlabelled altenative.
For example:
U(coca) = b1*x1 + b2*x2 /
U(pepsi) = b1*x1 + b2*x2 /
would then become:
U(cola) = b1*x1 + b2*x2 + b3.dummy*[0,1] / ? where 0 = coca and 1 = pepsi
U(cola) = b1*x1 + b2*x2 + b3.dummy*[0,1]
So you go from a labelled experiment to an unlabelled experiment. This would work well if all attributes appear in all alternatives. However, if some attributes are specific to the brand, then it will not work, as that requires a labelled experiment.
Michiel