A "Case Control" experimental design is a retrospective investigation. That is, it is a design where the researcher identifies at least two distinct populations (perhaps people with a disease and people without the disease), gets samples from the populations, and then goes back to look at the supposed formative experiences of those populations (for example, were the people exposed to lead-based paints as infants). The experimenter is looking to see if the different populations had similar formative experiences, or if they had different experiences. Finding that the there was a difference at least "suggests" that the differing experiences caused the eventual differences in the populations.

To look at a topic of current interest (current as of November, 2013), there is a concern that playing football may lead to early and severe brain damage. Rather than trying to construct an experiment to test this (a process full of ethical and practical complications), we could just choose two populations (perhaps males between the ages of 40 and 50 separated into groups that do and do not have detectable brain damage). Then we could get a sample of people in each group. For each of the sample subjects we could go back and determine the extent to which they played tackle football. Then we could compare the results between the two sample groups. If we believe that playing football is related to increased brain damage, then we should find a significantly higher level of football playing history in the brain damaged sample than we find in the sample of people without brain damage.

We note that the "inference" of such an investigation is not as striking as is the result of an experiment. However, to design an experiment to test our example we would have to randomly assign people to play or not play tackle football at various levels and we would have to wait 20 to 30 years to see the results. This is not only impractical, but it is not ethical to assign people to involvement in an activity that we believe causes permanent and sever damage. A case control study can be done now and it certainly does not cause additional damage to the subjects.