I recently did a study where I asked participants to edit a partially written article. They had to select at most three sources that they found most useful for a written task. Participants were divided into 4 conditions: either they receive a reading guide for sources, the partially written article, both or neither. The sources were of three types: very relevant, tangentially relevant, not relevant.
I am interested to know the likelihood of participants to select the very relevant sources over the tangentially and not relevant sources. Does this mean that I need a statistical method that can accommodate multiple count variables (which in this case is the source types)?