changing a level

[twickr07]

Dear Michiel,

After doing a small pilot study, I would like to change a level of one of my attributes. The attribute was coded as a linear variable:

b1[0.00001] * frequency[1,2,4,6]

I would like to remove 6 weeks and instead add 3 weeks. Can I swap 6 weeks for 3 weeks? If I had coded this as a categorical variable, it would have been easier, but wondering if there is a solution to this without redoing the design and the survey coding again.

Many thanks for your help,
Tara

[Michiel Bliemer]

You can swap levels in the design, but it will make the design somewhat less efficient and you may get dominant alternatives because you are changing the preference order of the attribute levels 6 was preferred over 4, while 3 is less preferred than 4). You can easily check for dominance yourself and if there is no issue then you may just have to accept a decrease in efficiency (but I suspect not a major difference since you used zero priors).

Michiel

[twickr07]

Thank you so much for your advice!

Tara