Dear Listers,
I have a pressing problem! Our client wants to test in a Central
Location 5 products. His budget
permits a sample of only 125 respondents.
5 products generate 120 different rotations. I can add another
5 rotations (extract randomly from this 120: 1 considering product A on first
place, 1 considering product B on first place. 1 considering product C, 1 for
product D and 1 for E). In this way, I will have exactly 125 rotations (120
unique) for exactly 125 respondents.
Is this method correct? I’m not so sure, because I consider
that a minimum overlap is strongly necessarily! How can I reproduce the
variance between groups if I don’t have enough cases per group?
I was thinking at another method: I can extract randomly from
this 120 possible rotation a number of 25 rotations. In this way I assure an
overlap (5 cases per 1 rotation).
What’s your opinion about this situation?
Which is the best method?
Thanks a lot,
Alexandra
Locked
