Box 4-1 Choosing the Null Hypothesis in Epidemiology

How could such therapy have been approved Pike explained that tests required by the U.S. Food and Drug Administration are designed to test the null hypothesis that the treatment does no harm. Instead, he said, the null hypothesis should be the treatment increases the likelihood of causing breast cancer.  

White regions of mammogram are denser tissue than dark regions. [allOver photography/Almy.] 

Bicarbonate in replicate samples of horse blood was measured four times by each of two methods with the following results  

---