The reference population, or the study population, is the group to which the results of the study, if valid, will be applied. A sample is a sub-group of the study population that has been selected, randomly or not, to globally represent the study population, when it is not possible, for practical reasons, to study the population in its entirety.
Is the selection biased? Is it random? Do the groups differ by characteristics that are other than those being studied? What is the proportion of subjects that remain at the end of the study? If we are dealing with a diagnostic test, has a wide range of samples been taken into account?
At this step, the external validity of the study should be examined: the conclusions, assuming they are valid, are they applicable to a wider population than the population of samples being tested?