Product development

Participants rate many similar crackers in one long session, and ratings become less consistent near the end. Which adjustment is most directly relevant?