Goat breeders investigated the effects of three different additives on the number of somatic cells in goats’ milk, additive DHA which is of fish origin, additive ALFA of plant origin, and additive EPA of fish origin. A control treatment contained no additive. The objective of this experiment was to answer two questions: Do any of these additives significantly reduce the number of SC in goats’ milk? For treatments resulting in a reduction, how long does the effect persist?
Standard statistical methods used in the first phase did not give satisfactory results, therefore we analyzed the experiment in the context of exploratory data analysis. We used several graphical displays to establish which transformations have to be used in the preliminary phase and what kind of statistical methods should be applied to answer the questions of the experimenters. The results showed that ALFA treatment is the best, its effect was significant up to 54th day of the experiment. Exploratory approach generated a new hypothesis: the procedure used for drug administration may increase the number of somatic cell due to the stress caused to the animal.
In this paper we tried to show that in the analysis of experiments the common approach mostly based on modeling and hypothesis testing can be expanded by an intense exploration of data.