Statistical inference means that conclusions about a value in a population are drawn from a random sample or that conclusions about a theoretical value, a parameter, in a probability model are drawn from an observed outcome. One should regard the random variations in the occurrence of disease as an expression of variations in the occurrence of risk factors for the disease which for the moment are not known or even suspected. The magnitude of the "random" variation in the occurrence of disease can be evaluated with the aid of statistical inference. One precondition for this is an underlying or theoretical measure of the occurrence of disease which is estimated from an observed measure of disease occurrence. The chief justification for carrying out empirical studies is the belief that these underlying theoretical relationships or processes do actually exist. If epidemiologic studies were used exclusively to describe the population which is actually under study, their value would be severely restricted.