The literature distinguishes between many different types of experimental designs. Some terminology is focused on the location of the experiment: is it performed in a lab, out in the field, or is it a natural experiment? Other terms pay more attention to the assignment of individuals to different conditions/groups.
- For example, there is the pre-experimental design, where an intervention is performed for the entire sample of individuals (i.e. everyone is part of the so-called intervention group). Thus, there is no “control group” that one uses for comparison.
- The quasi-experimental design includes an intervention group and a control group, but the participants are not randomly assigned to the groups.
- Then we have the true experimental design: this includes a control group and an intervention group to which the participants are randomly assigned. Another term commonly used for the true experimental design, is randomised controlled trial (RCT).
Intervention group vs control group
What is the purpose of dividing the participants into an intervention group and a control group?
Well, the aim with an experimental study is to see whether the intervention (e.g. a treatment) has had any effect on the outcome. We therefore collect information before the intervention as well as after the intervention; has the outcome of interest changed across measurement points? What kind of participant conditions can predict this change?
However, even if we can detect a change, how do we know that it is not by chance? It could be something else than the intervention that has caused this change. That is why we have a control group. If the same change is not observed in the control group, we can be more confident that the change in the intervention group is actually caused by the intervention.
Randomisation
So, why is it necessary to randomly assign individuals to the groups? This is because we want to be sure that a difference in the outcome between the intervention group and the control group is not simply because the groups are very different when it comes to the distribution of conditions (in epidemiology, one would be concerned with the equal distribution of risk factors in the two groups). Accordingly, randomisation has a great influence on the reliability of the results.
Blinding
Apart from randomisation, RCT also draws heavily on the concept of blinding. Some studies are blinded, meaning that the assignment to intervention group/control group is unknown for the participants, and sometimes also to those who provide the treatment and/or the researchers. This is to minimise the effect of expectations on the outcome. In the social sciences, blinding is usually not feasible.