statistical analysis of allocation data: next x purchases
Wednesday November 26, 2008 by Elsbeth Black
I'm struggling with the correct analytic approach to take with allocation data. The basic design is that the subject allocates their next say 6 purchases across a number of products within a category. Everyone has to allocate all 6 of their purchases but they can allocate all 6 to one product or distribute them across products.
At first this seems like a repeated measure design - so I would do a repeated measures ANOVA, but that feels wrong.
Right now I am thinking that this is really a Chi-Square test of independence. But maybe I should be thinking about these as proportions of allocations - so z - test.
I could really use some direction.
» Post to this Topic
Thursday December 4, 2008 by David J. Mangen
This is a reply to: statistical analysis of allocation data: next x purchases
I am assuming that the research question revolves around whether or not purchases are randomly distributed. It sounds like you need to determine the random expectation of a purchase of each product. If we assume that there are 10 different possible products in the category, then you would want to compare the observed average to the theoretically expected value of 0.6 (6 purchases divided by 10 product options).
Having said that, however, this is really more an issue of "what are you trying to determine" than it is an issue of what is the correct technique. Clarification is probably needed, because I was guessing as to the objectives.
» Reply to this Post