Quasi-Experimental Research

Logic of Quasi-Experimental Designs

A quasi-experimental research design is intended to approximate a true experiment in real-world settings where complete control or manipulation of some variables is not possible. The critical distinction between true experiments and quasi-experiments is that the units being evaluated are not randomly selected or assigned. An example of a study that cannot construct groups randomly is found in cases when all individuals who meet the eligibility criteria for a social program are required to receive the program. This situation occurred in some evaluations of the Head Start Program, which was targeted at preschool children from low-income families. In many local Head Start Centers, all of the children who met the eligibility criteria were enrolled in the program, which eliminated the possibility of randomly assigning some eligible preschoolers to a control group. Another example occurs when all students performing below a specified proficiency level on an assessment are required to take a remedial course or curricular program. When control groups cannot be formed randomly, there are likely to be several threats to the validity of the causal inferences that can be drawn between the treatments or interventions and the outcomes. Less confidence is placed on the inferences drawn from quasi-experiments than on inferences drawn from true experiments. However, quasi-experiments are more flexible than true experiments and can be more easily implemented in field settings, thus, the results of the studies are often more generalizable. These qualities suggest that quasi-experiments are likely to be a desirable choice for use in curriculum studies.

Threats to Validity

The identification of threats to the accuracy of causal inferences between treatments and outcomes in experiments and quasi-experiments was a significant advance in the social and behavioral sciences. Initially, researchers focused on issues of internal and external validity. Internal validity refers to the confidence the researchers have that the study findings are attributable to the treatment alone. When the researcher can eliminate rival hypotheses that might produce the observed relationship, the internal validity of the study is strengthened. External validity has to do with the generalizability of the study findings to other populations. Quasi-experiments vary in the degree to which they limit the internal and external validity of the studies. Over time, the kinds of validity that could threaten the inferences drawn from experiments and quasi-experiments were expanded to include statistical conclusion validity and cause and effect construct validity. However, experts in research design do not agree on that so many types of threats to validity are valuable.

...