Quasi-experiments include certain control over variables and are used to test causal hypotheses. The main goal of quasi-experiments is to reduce bias and other factors that can threaten causal inference. Since the participants of quasi-experiments are not aware of the fact that experimentation is being conducted, the factor of bias on their part is removed. To define and imply causal inference, several designs can be used.
Some of the most common designs include the one group posttest only design, the nonequivalent control group design with pretest and posttest, the interrupted time-series design, and the regression discontinuity design. The first design consists of one posttest observation of participants receiving the treatment and does not exclude the majority of plausible threats. The second design uses a pretest, which allows improving causal inference since it can help “to determine whether people changed from before to after treatment.” In the third design, the same variables are repeatedly observed over time, with statistical procedures used to assess the treatment better. In regression discontinuity design, experimenters would base treatment assignment on “a cutoff score on an assignment variable measured prior to treatment.”
Regardless of what design is used to define and imply causal inference to a quasi-experimental designed project, it is essential to rule out alternative explanations. For that, it would be necessary to enumerate all of them, identify the plausible ones, and evaluate which of them are operating in a way that may explain any of the effects observed. Thus, there are different ways to define causal inference while conducting a quasi-experiment, and the choice of steps to take depends on the study itself.