An experiment has variables and a control group, and is valuable in that it determines exactly what is causing a phenomenon. Non-experimental research is based largely on uncontrolled observation. While this has disadvantages, non-experimental research allows for the study of phenomena that would otherwise be impossible or immoral to study. While there are many types of non-experimental design, they can be grouped into a few categories and have their own strengths and weaknesses.
Experimental vs. Non-Experimental Designs
A true experiment uses groups to test a hypothesis. Members of these groups are randomly assigned to the control and variable groups, and the researcher is largely in control of the entire experiment. A non-experimental design does not follow this format. There may be only one group, or it may not be possible to randomly assign participants. Non-experimental designs are often used in psychology, but have value in other fields as well. Experimental designs have more validity and can show more definitively the answer to a question than non-experimental designs, though non-experimental designs have their own advantages, such as lower costs in many instances.
A quasi-experimental design is any experiment in which the researcher is not able to divide the subjects of the research into equal groups. In ideal conditions, groups are randomised into control and experimental groups to eliminate different types of bias. Some situations do not allow this. Subjects tested, manipulated with a variable, then tested again and compared to their earlier results are part of a quasi-experimental design. Subjects tested at intervals over time with a variable manipulated at each point are also part of a quasi-experimental design. These designs have potential for bias and external variables. However, in some instances, this is the only practical design for an experiment.
In a developmental design, time is the major variable of the experiment. For instance, the behaviour of a group of children in a certain situation may be compared to that of adults. A group of volunteers may be studied for the same variable, unmanipulated, over a long period of time. These types of studies are good at describing trends or demonstrating the effects of time. The major disadvantage in this type of study is subject mortality or subjects dropping out of the experiment for various reasons as time passes.
Observational research is any type of research in which a researcher studies the behaviour of a subject by observing the subject. The researcher may study behaviour in a natural setting or may join a group in order to study that group. Case studies, in which a single person or small group is studied, are also part of observational research. Observational research can also be conducted by analysing pre-existing data. These kinds of experiments are prone to bias on the part of the researcher as well as to misinformation from the subject of the study. In addition, if a subject is aware of the observation, the subject may behave differently, skewing results. However, in many instances, valuable information is obtained from observation and researchers may discover key information that can lead to future experiments.
- Wadsworth Cengage Learning: Nonexperimental Approaches
- Research Methods Knowledge Base: Types of Designs; William M.K. Trochim; 2006
- The Ohio State University Department of Political Science: Non-Experimental Research Designs
- California State University, Fresno: Observational Research
- California State University, Fresno: Developmental Design
- California State University, Fresno: Quasi-Experiments