I am interested in investigating the effects of regular administration of metformin on blood sugar levels (glycosylated hemoglobin, HbA₁C) in obese marmoset monkeys. I measure HbA₁C (mg/dL of blood) in 10 obese marmosets before exposure to metformin and again after 3 months of regular metformin administration.

1. What is the null hypothesis? 2. What type of experimental design is outlined? What threats to internal validity are associated with this type of design? 3. Of the statistical tests (Fisher's exact test, binomial test, and three types of t-test), which is most appropriate for the experiment outlined and why?