A Quasi-Experimental Research Design
A quasi-experimental design is similar to a true experimental design because it also defines the cause-and-effect relationship between dependent and independent variables.
However, there is a significant difference between these two designs. The matter is that a quasi-experiment does not use random criteria. Subjects relate to groups on the basis of non-random assignments. A quasi-experimental design is used when a researcher cannot conduct real-life experiments because of some practical or ethical restrictions.
Why Quasi Experiments and True Experiments Are Different
There are three main differences between these two types of experiments. One of them is in the assignment to treatment. In the true experimental design, you will randomly assign subjects either to treatment or control groups. You use non-random methods to relate subjects to groups in the quasi-experimental design.
In the true design, you develop the treatment yourself, so you can control it. In the quasi design, you cannot control the treatment because you study only pre-existing groups that have already received various kinds of treatment.
Control groups are important in the true experimental design. In the quasi design, control groups are not mandatory, though you can usually use them.
Suppose you need to research how a certain psychological influence can treat depression. In a true experiment, you should go to a mental health clinic and, by dividing all patients into halves, assign this treatment to a half of them and leave the other half with the standard course making them perform as a control group. However, there are certain ethical reasons why the director of the clinic will not permit you to assign the treatment to patients randomly.
So, a true experiment is impossible here, and you need to use a quasi-experimental design. You may have found several psychotherapists in the clinic who are using this new therapy, while others stick with the traditional protocol. It means that a treatment group exists already. So, you can study the progression of symptoms both in it and in the group that uses normal methods. It means that the treatment and control groups have not been randomly defined. However, if you observe both groups systematically, you will see the differences and make the conclusion that these differences are a result of the treatment, and no other confounding variable is involved.
Types of a Quasi Design
There are many types of quasi-experimental designs. Let’s consider the three most common types.
The Design Based on Nonequivalent Groups
Here a researcher needs to pick out already existing groups that seem to be similar but only one group undergoes the treatment. Usually, when you deal with a true experiment and the groups are assigned randomly, the control and treatment are similar. However, quasi-experiment groups, which are not random, display differences. Such groups are nonequivalent ones.
As a researcher using a quasi-experiment, you should rely on any confounding variable and control it in your analysis. Or you may pick out groups that are quite similar. This nonequivalent group design is widely used.
You would like to test your hypothesis about a new after-school program. You believe that it will improve students’ grades.
Therefore, you need two similar groups of students from different schools. One of them uses a new program, while the other does not. When you compare the results of those children who study according to a new program and those who attend the traditional one, you can see whether the new program has affected students’ grades.
Experimenting with Regression Discontinuity
Most researchers use the scenario when one group receives a potential treatment and the other does not. There is a kind of threshold, with the group treated above it and the group unaffected below. The difference between these two groups is minimal immediately near this threshold. Therefore, those individuals who are placed above the threshold are considered a treatment group, and those below it are a control group.
There is a practice in some American high schools to enroll only high-achieving students to their programs. Therefore, to be admitted, all the candidates should pass a test to be allowed to attend this program. It means that students who have passed this test are considerably different from those who have not.
Nevertheless, the exact score for the test is never objective, so the students who have barely passed the test and those who have lost some points only and, accordingly, failed, do not differ much. All of them are near the threshold. So, your conclusion can involve the random chance and any slightest differences in the schools they come from.
Though, you need to continue research with the consideration of the long-term outcomes of attending a certain school for those students who have nearly passed the test and those who have nearly failed. Only after completing the research with the regard to this threshold, your results will be objective.
Natural Experiments
Such experiments can be field and laboratory ones. As a researcher, you can control the assignment of objects to groups. The assignment to a treatment group can be random or random-like because of some external factors or situations. Due to this, natural experiments cannot be fully true because they are observational and cannot be controlled. The only thing a researcher can do is to exploit the event after the completed fact to conclude about the effect of its influence (treatment).
Let’s consider the Oregon Health Study conducted in 2008. It is a well-known example of a natural experiment. The state of Oregon decided to introduce Medicaid as a public health insurance program to the most low-income people. Though, they were not able to cover all the adults they considered eligible for this innovation. So, they had to arrange a random lottery just to allocate some spots for this program.
Therefore, it was a favourable ground for researchers to study the impact of the innovation on those who were randomly assigned to the program and those who were to be assigned but did not succeed in the lottery. The latter individuals made up a control group, while the former ones were considered a treatment group.
Where a Quasi-Experimental Design Can Be Used
It is understandable that true experiments are more internally valid. However, using a quasi-experimental design is inevitable when there are ethical or practical reasons for such a choice.
Ethical Reasons
A random basis provokes many ethical restrictions that do not allow you make true experiments. A quasi experiment can help you do the same study in a casual relationship without breaking ethical rules.
The Oregon Health Study showed that random provision of only several individuals with health insurance, while preventing others from getting it, would have been unethical. Therefore, to meet the purpose of research, it was decided to have a lottery because the Oregon government could not provide such insurance to everybody who needed it. That turned out to be a far more ethical approach that, in its turn, allowed for the resultative study of the problem.
Practical Reasons
A true experimental design may be unavailable in certain situations or turn out to be too expensive if you do not have access to large funds. Or there may be too much work needed for recruiting helpers and implementing the design of a true experiment. You also may not have much time and power to research an adequate number of subjects so that your experiment should be justified.
Quasi-experimental designs are great here because they allow you to take advantage of the data that has previously been collected with great effort and paid for by others.
Pros and Cons of a Quasi-Experimental Design
As other research designs, a quasi-experimental design has its advantages and disadvantages. Let’s consider some of them.
| Pros✔️ | Cons❌ |
|---|---|
| ✔️ The design can boast higher internal validity than non-experimental designs because you can put more control to confounding variables than in other types of research. | ❌ Retrospective data collected by someone else and for different purposes can be incomplete, inaccurate, or partially impossible to obtain. |
| ✔️ The external validity is higher than in true experiments because the latter may involve unpredictable real-world interruptions and interventions, even despite the artificial settings developed in laboratories. | ❌ The internal validity is lower than in true experiments because you may not verify that all the confounding variables have been taken into account. |
Final Thoughts
A quasi-experimental design can be helpful in many situations when real experiments cannot be conducted because of some practical or ethical issues. You need to know the main differences between real and quasi experiments and their designs described in this article to be sure that you have chosen a proper design for your research. If the design is appropriate, you are sure to finish your research with better results and succeed in your academic career.