For a single factor design I would only have grain as my factor and the levels would be the grain types listed above. The experimental units will be randomly assigned to the listed treatments (the different types of grain). This experiment will use subset randomization and a paired design to see the mean difference in weight for the treatments. The benefits of a single factor experimental design are that it would have less treatment groups. Less treatment groups could potentially have lower costs. The pitfalls of the single factor design are that I wouldn’t be able to see how the amount of grain effects the response (mean weight change of the horse). I could potentially miss out on results that warrant better results. For a factorial design I would …show more content…
Since there will be many horses I will use blocking and will keep one horse from each treatment group in a pasture with grass, since horses are foraging animals having a pasture that they can graze at is crucial. I will have multiple pastures depending on the number of replicates I have. I will repeatedly measure the horse’s weight every week for eight weeks. Since horses vary by breed I would have a mixed effects model and treat the horses as a random effect. Since repeated measures provide more power I will be able to have less replicates. Using blocking for my pastures will help to control for the differences in the pastures. Using blocking for my pastures will help to control for the differences in the pastures. By treating the horses as a random effect, I will be able to use horses of different breeds and ages, therefore taking care of the variability that the horses can bring. A problem I may encounter is if my pastures have more than one extraneous source of variability it can be difficult to detect treatment