An experimental research design is a set of protocols to be used in a research experiment to test a hypothetical causal relationship. Researchers must design their experiments before they can initiate them to maintain the integrity and viability of the study. They can rely on protocols from older experiments, general standards and practices, and guidance from assistance and supervisors. The process of experimental research design can take weeks or months as researchers explore all aspects of a research project to develop the best design.
The first step in experimental research design is defining the phenomenon to study. A researcher may want to examine anything from human behavior to a topic in biology. She must describe what she wants to study and develop some parameters to define the testing and determine what a successful outcome might look like. Researchers want to confirm or deny a correlation and determine whether it is causal in nature, or merely coincidental. The research may provide definitive results or material that contributes to a body of knowledge on the subject.
Experiments need to take place in tightly controlled environments where every possible variable is addressed. The researcher does not want to end up with false results on the basis of poor experimental controls. Thus, experimental research design requires the researcher to outline all the variables with a possible impact and provide a mechanism for controlling them. In a test on human behavior in the grocery store, for example, the researcher might need to set up a fake grocery store to carefully control for lighting, sound, encounters with other people, and other factors.
The experimental research design will show what is to be studied, how to study it, and how the researcher will measure outcomes. With some experiments this may be easy, as the research should generate quantitative data. In other cases, the experiment relies on interviews with subjects, observations, and other methods of data collection. These must be appropriately collated and analyzed to yield useful study results.
Researchers preparing an experimental research design must be prepared for any outcome in the study. If the design tends to slant the results toward a specific outcome, it is not valid, and may lead to bad data. Personal bias can be an issue in research, especially when money rides on a specific outcome. The researcher should be able to show how he plans to correct for bias, such as running a double blind medical study to prevent researchers from influencing their subjects.