Let’s now look at a type of study design that is like experiments but lacks random assignment. This design is called quasi-experimental. The quasi-experimental design is used when it is impossible or unethical to randomly assign people to groups.
Perhaps you are involved in research conducted by the military to determine when infected soldiers start showing cognitive changes associated with HIV infection. The study will use a quasi-experimental design. There can be no random assignment to groups because the study demands specific participants in each group. One group comprises soldiers who are infected and the other group comprises soldiers who are not. Participants in both groups must be matched in age and other factors.
Since you must ensure that participants in both groups are matched in age and other factors, you cannot randomly distribute the soldiers. You must also ensure that one group has infected participants and the other group has healthy participants. You have no control over the manipulation of the independent variable, the HIV infection.
Both groups come in for testing twice a year. Each time, their results are compared to their own previous scores and to the scores of the matched-control group. The study design enables you to see when cognitive changes occur.
In many ways, these studies are like experiments and can provide useful information. You can make cause-and-effect statements using such studies, but they must be made with caution.
Since you could not control the random assignment of people into groups, you have no idea whether other subject characteristics are equally distributed across groups. The presence of these other characteristics in only one group can lead to confusion and could be responsible for the quantity or quality of cognitive change. This results in limitations to internal and external validity.