Answer:
Take a village and divide the population in two, with an experimental group that will be given a different diet, and a control group which will continue with their normal diet.
Explanation:
In experiment design, when we want to test whether or not certain factor actually has a real effect on a population, one of the best methods to follow is to split the population in two groups: an experimental group, where we will apply the treatment, and a control group, where either a placebo or no treatment at all will be applied. The advantage of having both an experimental and a control group is that this way we can ascertain with much more clarity whether a change in the indepent variable (the treatment) has an observable effect on the dependent variable (the effect on the population).
In our example, a practical manner to test whether longevity in Russian Georgian villages is due to genetic or dietary factors, is to take one of the villages, and split the population in two groups: an experimental group, that will be given a higher calorie diet, and a control group, which will continue to consume their normal low-calorie diet. Then we would track the differences in age expectancy over time. If after a long enough period of time there's a difference in longevity between both groups, in other words, whether one of them leads longer or shorter lives than the other, then we can conclude that our test has proven that diet has a definitive effect on their longevity. If there's no significant difference, then we can conclude that diet has no relevance and it's all due to genetics.
