The phenomenon of qualitative interaction between the research activity and the studied object exists in veterinary epidemiology, as in all observation-based disciplines. Observational studies carried out under farm conditions strongly involve the farmers and the farm's usual advisors in the data collection. This leads to various biological and sociological interactions between the observer and the observed that can modify the results of the study. Some methodological considerations are proposed in order to consider such qualitative interaction, from the designing of the study through to the interpretation of the results. Examples are presented to illustrate these ideas.