A Quasi-Experimental design is one in which the treatment comparison groups are not assigned by randomization. The groups might be for example an education study, in which your treatment and comparison groups are two different 6th grade classes, or two different grades all together. Another example would be a blood pressure study where the treatment group is everyone with HDL>150 and the comparison group is everyone with HDL<150. The latter example is actually indicative of a cut-off assignment which is actually very good at infering causality, but, it is still technically a Quasi-experimental design because the groups were not assigned randomly.
As we can see, it is possible that pre-existing groups (i.e. the 6th grade example) are dissimilar in one way or another, before the treatment or intervention even occurs. That is why these types of experiments are not as good as a randomized (or cut-off ) experiments for infering causation. The problem is that you may not be able to tell if it is your treatment/program/intervention, or dissimilarities between the two groups that cause the difference in your outcome.
The one advantage to these type of experiments is that they are often cheaper, and more feasible than a randomized experiment, and the differences in comparison and treatment groups can be accounted for in the analysis (see analysis section in the non-equiv group page below).
If you feel that you do want a greater ability to infer cause, you might want to investigate whether it if feasible to do your experiment as a randomized design .
Please Email me with any information
or comments about this web page
4/9/97
