| Abstract: | The question whether it is allowed to use methods of statistical inference in the study of entire populations is disputed in the scientific community. Such a procedure can be justified only if the generation of the data of the population can be regarded as the realisation of a stochastic data generating process, analogously to drawing a sample out of a population by random. This article specifies under which circumstances these conditions can be regarded as given. Furthermore, it shows under which circumstances we should be at least careful, if not reluctant, to interpret the results of significance tests in the usual manner. |
