Quasi-Experimental Design: A Sneak Peek to Research Technique that Tests a Causal Hypothesis
Experimental research design is widely used by the research community to manipulate one or more variables (independent) and determine their effect on the other variables (dependent). This type of research design is often used in two cases: (1) When time priority exists in a causal relationship and (2) there exists a consistency in a causal relationship.
Experimental research design is classified into a pre-experimental, true experimental and quasi-experimental research design. Although true experimental design is regarded as the most accurate design, it is a quasi-experimental design that is widely used in the research in social science, psychology, etc. as it doesn’t involve random assignment of participants to orders of condition.
A quasi-experimental research design, also known as ex-post-facto design involves the selection of groups without a random pre-selection process. This type of research design doesn’t allow the research scholar to control the assignment of participants to groups or conditions. It identifies a comparison group that is similar to the intervention group in terms of baseline features.
Types of quasi-experimental research designs
Quasi-experimental research designs are of various types and have a variety of applications in a specific framework.
- Non-equivalent group, posttest only design – This type of research design includes administration of outcomes measures to treatment group or two groups and a comparison. For instance, a group of students might receive coding instruction using the latest programming language whereas another group of students might receive writing instruction using standard programming language. After a week, a writing comprehension test is administered to determine which programming language is more effective. However, the major setback of this design is that the two selected groups may differ in important ways that impact the writing process. For example, if a group of students who code using standard programming language, then we cannot determine if the students are well-prepared or if it’s the result of utilization of standard programming language.
- Non-equivalent control group design – Here we compare the control group and the experimental group. The groups are selected & assigned through convenience rather than choosing via randomization. But the problem arises when the results of the control and experimental group are to be compared. For instance, a researcher wants to determine the impact of online training on the average grade point of class 11 students. Here the researcher cannot randomly collect the participants as the institution will not permit to regroup the classes. Therefore, the researcher chose two sections of class11 from the same institution. Since the participants were not randomly allocated, the researcher cannot conclude that two groups were equivalent before the manipulation. Also, it cannot be determined if the difference in the grade point is only due to the impact of online training as other variables such as intelligence, motivation, etc. can bring about the difference in learning.
- Separate pretest-posttest sample design – The main idea of this research design is that the individual selected for the pretest is not the same as that of the posttest. For example, assume that the two companies are similar. You can use one company to implement your study and second company as a control. A customer satisfaction program is developed. Initially, the customer satisfaction is measured without implementing the program. This is followed by the implementation of program and measuring the client satisfaction. Here the clients for each company will be different for pretest and posttest. The response of the client from the pretest and posttest cannot be compared. As a result, nonequivalence exists not just between companies but also between the pre and post groups.
- Double pretest design – This type of quasi-experimental research design is regarded as the strongest design with respect to internal validity. This is because, in the pre-post nonequivalent research design, the nonequivalent groups may differ from one another. Pretest does not tell us if the changing at the same rates prior to the program. But with a double pretest design, we can determine the rate of change through pretest1 and pretest2. Therefore, this research is explicitly used to control the selection of maturation threat.
- Repeated treatment design – Here the treatment is withdrawn and then presented for the second time. Simply said, treatment is presented more than once. The response of the participant is measured prior to and after the implementation of treatment. The treatment program is withdrawn and then the process is performed again. Consider an example where the government has banned the consumption of alcohol. To analyze the impact of this policy, problems pertaining to alcohol consumption were assessed. The outcome of the policy was the health of individuals improved as a result of the ban.
A quasi-experimental design is widely used to produce results for general trends. The added advantage of this design is that it doesn’t require randomization, prescreening and minimizes the threat to internal validity.
Although this type of research design is easy to implement, one should take care to address the internal validity threat and utilization of additional data.
