This module describes methods typically used to construct a comparison control group for quasi-experimental research.

Learning Objectives

• Describe the purpose of a comparison group
• Define matching
• Explain matching methods and provide examples

In quasi-experimental research, it is not possible to randomly assign subjects into experimental and control groups. Subjects may be selected rather easily for the experimental group based on their membership in an existing group or due to convenience. In many research problems, however, it is important to have a control group for the purpose of comparison with the experimental group. The problem in quasi-experimental research is that without random assignment to control for confounding variables, inherent differences in the experimental and control (comparison) group test subjects may impact the results. For example, if a researcher wanted to explore the difference in overall health of children who ate 5 or more servings of fruits and vegetables every day with children who did not, the researcher would need an experimental group and a control group. Parents concerned about healthy eating are not likely to participate if assigned to the control group so random assignment is not possible in this example. However, the selection of a control group may introduce many confounding variables. For instance, healthy diets may be related to the education levels and incomes of the parents. It may even be related to whether or not the children attend a daycare or stay home with one of the parents. These confounding variables must be addressed and therefore, methods such as matching have been developed to assist in the construction of a valid comparison control group.

Matching refers to the selection of control group subjects based on the similarity of specific criteria. It is a process whereby the confounding variables of all participants are considered, and members are selected for the control group based on how well they “match” the overall characteristics of the experimental group. In the above example, members of both groups should have similar education, income, and time spent with their children.

One of the most common methods is pairwise matching, in which pairs of participants are identified that are similar in other ways but differ in their attention to the fruits and vegetables their children consume. The data from these paired samples can be used in statistical analyses to determine significance difference. Another way to utilize matching is to attempt to achieve similar groups that have matching characteristics regarding variables deemed important in the study. The most significant of confounding variables can be controlled in either of these manners. However, while matching guarantees that the subjects are not different in those ways identified by the study, it does nothing to equalize the conditions created by other variables not considered in the matching because perfect matching is not possible. For this reason, random assignment is still superior, but matching does help construct better comparison groups.

There are many other types of more complicated matching techniques including judgmental matching, propensity score matching, matched comparisons, and sequential allocation, which are discussed in the resources and readings associated with the module. In general, matching in quasi-experimental designs can be classified as Normative Group Matching and Normative Group Equivalence.

• Normative Group Matching is a procedure where the experimental group is compared with a sample from a normative population. This procedure is when specific criteria are identified in the experimental group individuals and a person with matching criteria is selected for the control group. For example, if there is a 25-year-old female in the experimental group, a 25-year-old female should be selected for the control group. Because it may become difficult to match on more than one variable, careful selection of matching criteria is important.
• Normative Group Equivalence is another matching procedure, but it is based on identifying test subjects based on selected equivalent demographic characteristics to create a control group that is overall similar in regard to those key characteristics. These factors may include variables such as an age range, gender, income range, education levels, and so forth.

The construction of a valid control group for comparisons is important to the validity of the study. For the results of the study to be meaningful, it is important that the researcher carefully considers how the test subjects for the control group are chosen.

Suggested Readings

Bernard, H. R., & Bernard, H. R. (2012). Social research methods: Qualitative and quantitative approaches. Sage.
Brown, L. (2010). Quasi-experimental research. Doing Early Childhood Research: International perspectives on theory and practice, 345.
Campbell, D. T., & Stanley, J. C. (2015). Experimental and quasi-experimental designs for research. Ravenio Books.
Creswell, J. W. (2013). Research design: Qualitative, quantitative, and mixed methods approaches. Sage publications.
Fraenkel, J. R., Wallen, N. E., & Hyun, H. H. (1993). How to design and evaluate research in education (Vol. 7). New York: McGraw-Hill.
Lipsey, M. W. (1990). Design sensitivity: Statistical power for experimental research (Vol. 19). Sage.
Punch, K. F. (2013). Introduction to social research: Quantitative and qualitative approaches. Sage.
Stuart, E. A., & Rubin, D. B. (2008). Best practices in quasi-experimental designs. Best practices in quantitative methods, 155-176.
William R. Shadish, Cook, T. D., & Campbell, D. T. (2002). Experimental and quasi-experimental designs for generalized causal inference. Wadsworth Cengage learning.