Question 6.18
6.18 Flu shots. A New York Times article reported a study that investigated whether giving flu shots to schoolchildren protects a whole community from the disease. Researchers in Canada recruited 49 remote Hutterite farming colonies in western Canada for the study. In 25 of the colonies, all children aged 3 to 15 received flu shots in late 2008; in the 24 other colonies, they received a placebo. Which colonies received flu shots and which received the placebo was determined by randomization, and the colonies did not know whether they received the flu shots or the placebo. The researchers recorded the percentage of all children and adults in each colony who had laboratory-confirmed flu over the ensuing winter and spring.
(a) Outline the design of this experiment. You do not need to do the randomization that your design requires.
(b) The placebo was actually the hepatitis A vaccine, and “hepatitis was not studied, but to keep the investigators from knowing which colonies received flu vaccine, they had to offer placebo shots, and hepatitis shots do some good while sterile water injections do not.” In addition, the article mentions that the colonies were studied “without the investigators being subconsciously biased by knowing which received the placebo.” Why was it important that investigators not be subconsciously biased by knowing which received the placebo?
(c) By June 2009, more than 10% of all the adults and children in colonies that received the placebo had had laboratory-confirmed seasonal flu. Less than 5% of those in the colonies that received flu shots had. This difference was statistically significant. Explain to someone who knows no statistics what “statistically significant” means in this context.