Quasi-Experimental Design: Types, Examples, Pros, and Cons

A quasi-experimental design is a type of research methodology. The best way to explain this approach is to understand the difference between experimental and quasi-experimental designs. As the name suggests, a quasi-experiment is almost a true experiment. The primary difference between the two is that researchers do not randomly select specific elements (frequently the participants) in a quasi-experimental design. In true experiments, the selection is random.

In both experiment types, an independent variable gets manipulated to see the cause-and-effect of the dependent variable. (Learn more about independent and dependent variables.) The internal validity (the confidence you have that the causality in your study is not due to outside factors) of quasi-experimental research is not quite as strong as in an actual experiment due to confounding variables inherent in preselected participants or elements.

The most common quasi-experimental designs are:

1. 1. Nonequivalent groups design: This design uses a pretest and posttest for participants to gauge cause and effect.
2. 2. Regression discontinuity design: Regression discontinuity design assigns participants to a particular treatment using the propensity score of a pretreatment variable.
3. 3. Interrupted time series design: In this design, researchers track participants for a lengthy period, both pre-intervention and post-intervention.

There are two reasons why using a quasi-experimental design may be preferable to a true experimental design: ethical or practical. There are certain situations when research methods using a random assignment would be unethical, such as providing public health care to one group while withholding it from another treatment group. In a quasi-experimental study, you can examine a causal relationship without putting anyone in physical danger. Learn more about the ethical decision-making process.

On the other hand, a randomized controlled trial may not be the best choice for researchers for practical reasons, such as the work involved in group design or the cost of weeding through a large sample size of participants without a particular attribution for data collection. In a nonequivalent groups design, for example, you may need to have two experimental groups with similar conditions (as opposed to two groups of random people who may or may not have that condition) to study the causal inference of a treatment.

Let’s say you want to study the effects of a motivational reward on students who are frequently late to class. First, you would choose two classes of similar age, size, and makeup, then assign both classes a pretest, with research questions such as what time they arrive every day, reasons for tardiness, and general enjoyment of the class.

One class would then receive the motivational reward for being on time, making this class the intervention group. You would give nothing to the second class for arriving on time, making this class the comparison group. You would then administer a posttest with questions assessing the same factors as the pretest to see if the motivational reward affected tardiness.

A quasi-experimental design has several advantages, including:

1. 1. Higher external validity: Quasi-experimental research designs tend to have more real-world applications, especially within the social sciences.
2. 2. Higher control over targeted hypotheses: Because the participants in the control group or comparison group are not randomized, the nonequivalent dependent variables in your study design can be more controlled, targeted, and efficient.
3. 3. Can be combined with other methodologies: Quasi-experiments can lean on statistical analysis and alternative explanations from other true experiments, which cuts down on the time needed to determine your outcome of interest.

The disadvantages of a quasi-experimental design are as follows:

1. 1. Lower internal validity: Because the researchers control the variables, it’s hard to know if they have included all confounding variables.
2. 2. Risk of inaccurate data: Because a quasi-experimental design often borrows information from other experimental methods, there’s a chance that the data is not complete or accurate.
3. 3. Risk of bias: Because researchers choose baseline elements and eligibility, there’s a risk of human bias in selection. Learn more about different types of bias.

Get the MasterClass Annual Membership for exclusive access to video lessons taught by science luminaries, including Neil deGrasse Tyson, Chris Hadfield, Jane Goodall, Bill Nye, and more.