La représentativité d un échantillon et son test par le Khi-deux; Testing the representativeness of a sample

The value and credibility of a survey or opinion pool depend to a large extent on the observed sample s representativeness, which in turndepends on sample size. When dealing with an heterogenous population, however, sample size may be insufficient to counterbalance possible biases due to differences in constituent subpopulations. If population s composition is known, the sample s representativeness can be verified by a Chi-square test: the test is illustrated and a special formula is proposed and validated for the case of finite populations.