Quasi-Experiment Essay ⋆ Political Science Essay Examples ⋆ EssayEmpire

Quasi-experimental design refers to situations in which an experimenter can assign treatment conditions but, for certain reasons, is not able to randomize subjects across conditions. Variants on this design include interrupted time series designs, which can either be simple or equivalent types, and nonequivalent control group designs.

Quasi-experiments represent one of a broader range of nonexperimental alternatives to experimental work. They strive to incorporate as much of standard experimental design as possible but, or practical or ethical reasons, fail to achieve full experimental control in one area or another. These kinds of alternatives include correlational work, pseudo-experiments, and quasi-experiments.

In correlational work, researchers observe how variables correlate to each other, as with smoking and lung cancer. However, because it is not ethical to force people to smoke to see if the behavior gives more of them cancer, there is no treatment manipulation; thus experimenters are not able to prove causality. This is because problems exist in determining causal direction, and because spurious associations remain possible. For years, tobacco companies asserted that some third, unknown case made certain people both more likely to smoke and more likely to contract lung cancer, but there was no direct relationship between smoking and cancer. Often, statisticians can correct problems in determining causal direction with a cross-lagged panel design. But it took systematic experimental research in a laboratory to show that exposure to substances in cigarettes caused cancer in lung tissue in a petri dish to prove causality.

Regression-discontinuity designs compare treatment groups between pretests and posttests but use a distinct method of subject assignment. Subjects are placed into groups based on their preintervention score; those above a particular threshold are placed in one group, whereas those below are placed in another. While selection threats to internal validity exist, this method offers the advantage of targeting those most affected by a particular outcome.

According to Elliot Aronson and colleagues in their book Methods of Research in Social Psychology (1989), pseudoexperimental design exerts some control over administering manipulations to subjects. However, in this design, there is no control group, so comparisons between conditions remain impossible. Such studies often occur with natural experiments, where a real-life occurrence only makes one measurement

possible. There are three kinds of pseudo-experiments. The first, a one-shot case study design, occurs when investigators can only administer a single posttest, after a given disaster like a fire, hurricane, or earthquake has taken place. The second pseudo-experiment design is known as the one-group pretest–posttest. In this variant, the same group is studied before and after a manipulation, but no comparison control group exists so many threats to internal validity remain. A final kind of pseudo-experiment is the static group comparison design. In this design, two different groups are tested at the same time on the same variable, but because the subjects were not randomized to condition prior to the observations, there is no way to know whether preexisting systematic differences between subjects caused any observable differences between the groups.

Quasi-experimental designs give experimenters more control than either correlational or pseudo-experimental designs, but not as much as a fully controlled experiment. In quasiexperiments, experimenters can assign treatment and control conditions to subjects and collect the dependent measures they desire, but they cannot randomize subjects to condition. The main difference between real and quasi-experiments thus lies in the lack of randomization to condition in the latter design. This often makes quasi-experiments more feasible for real-life studies in field environments where such randomization remains impractical, illegal, or unethical, but this does limit the strength of the causal claim that can be made concerning any findings.

There are two types of quasi-experimental studies. The first is the interrupted time series, which involves a within-subject comparison using the same independent variables on the same group of subjects over time. Dependent measures are taken both before and after the manipulated treatment. Sometimes, measures are simply made after different levels of manipulation have been applied to different groups, as when more of a particular drug may be given to one group than another.

A second type of quasi-experiment is called the nonequivalent control group design. This design takes the form of a between-subjects model to measure the impact of a given independent variable on different groups of subjects. Comparisons of these groups are made at the completion of the study, but these subjects have not been assigned at random to condition, so threats to internal validity remain.

Bibliography:

Aronson, Elliot, Phoebe Ellsworth, J. Merrill Carlsmith, and Marti Hope Gonzalez. Methods of Research in Social Psychology. New York: McGrawHill, 1989.
Roth, John, and Alvin Kagel. Handbook of Experimental Economics. Princeton: Princeton University Press, 1998.
Shadish,William,Thomas Cook, and David Campbell. Experimental and Quasi-Experimental Designs for Generalized Causal Inference. New York: Wadsworth, 2001.
Trochim,William, and Jane Donnelly. Research Methods Knowledge Base. 3rd ed. Mason, Ohio: Atomic Dog Publishing, 2006.

This example Quasi-Experiment Essay is published for educational and informational purposes only. If you need a custom essay or research paper on this topic please use our writing services. EssayEmpire.com offers reliable custom essay writing services that can help you to receive high grades and impress your professors with the quality of each essay or research paper you hand in.