Experiments and Experimental Design

Experiments are a type of quantitative research method used to examine questions of causal relationships. Specifically, experimental designs allow researchers to determine whether one or more independent variables (IVs; also called treatment variables or factors) significantly predict one or more dependent variables (DVs; also referred to as outcome or criterion variables) when all other factors are held constant. In true experimental designs, IVs are manipulated by the researcher to determine their effect on DVs. To do so, researchers place participants who have been sampled from a single population with common characteristics into experimental conditions or control groups using random assignment. Preexperimental designs involve the examination of the influence of manipulated IVs on DVs without the element of control required for true experiments. Quasi-experimental designs involve the examination of differences in groups or the predictive value of IVs on DVs without the use of direct experimental manipulation or random assignment. In true experiments, preexperiments, and quasi-experiments, the impacts of the IV on the DV are analyzed using statistical tests, while controlling for any extraneous or intervening variables that might also have some predictive influence on the DV. Projects with factorial designs involve multiple IVs in a single project.

As a simple example of a true experimental design, researchers interested in whether or not a particular training protocol reduces communication anxiety would randomly assign participants who are otherwise similar in characteristics (e.g., age, socioeconomic status) to an experimental condition or a control group. Both groups would first complete measures of their communication anxiety to assess a baseline for all participants. Participants in the experimental group would then be given the training protocol, but those in the control group would not. Finally, both groups would again be given the measure of communication anxiety to assess whether or not the participants in the experimental group had significant reductions to their levels of anxiety as a result of the training as compared with the control group.

To the extent to which researchers randomly assign groups and carefully control for extraneous variables, any statistical differences in the experimental group compared with the control group can be attributed to the experimental manipulation. That is, if results show that participants in the experimental group are lower in communication anxiety than those in the control group, the researchers can assume that the training protocol was effective in lowering anxiety for the treatment group.

This entry outlines the most common types of preexperimental and true experimental designs.