Binomial test.

H0 could be p = 0.5, to see if your data sufficies to claim that organic goods were selected more (less) often that non-organic goods, or p = 5/15, to see if your data suffices to conclude a selection favouring organic over non-organic goos (or the other way around).

Jochen Wilhelm

Hi! Thank you for the answer. I may be wrong but I think it will only say me "people select or don't select organic products (= yes/no)". But it will not take into account the number of organic/non-organic products selected.

If each person can select several products, it's a beta-binomal model. The proportion of selected organics is modelled by a binomial distribution with parameter p (the probability to select an organic). This parameter p itself varies from person to person, and this variation is modelled by a beta distribution.

As being a "non-standard" problem and analysis, I suggest you seek help from a local statistician.

A crude workaround would be to claculate the proportions (organic:non-organic) for each person, logit-transform these values and use a t-test with weights proportional to the total number of selected items per person.

Jochen Wilhelm

Hi! Really appreciate your help.

Can you clarify:

A crude workaround would be to claculate the proportions (organic:non-organic) for each person, logit-transform these values and use a t-test with weights proportional to the total number of selected items per person.

To be more specific, I had 24 products on my table and 5 products were organics.

Let's say that participant 1 had selected 2 organic and 1 non-organic product.

How would you do the transformation of the data to make the products balanced?

Thank you so much!