Experimental, Quasi-Experimental, and Non-Experimental Designs

Experimental, quasi-experimental, and non-experimental designs are all utilized in nursing research. An example of an experimental design is found in a study by Bird, Jones, Korten and Smithers (2007). These researchers explored the outcomes of a predominantly psychosocial approach to the treatment of dementia. The intervention group received treatment that focused on problem behavior from a psychosocial angle along with medication, while the control group did not receive this treatment and only received psychopharmacological treatment. The results indicated that the intervention (psychosocial approach) was effective in treating dementia, and allowed the treatment group to reduce medication levels.

An example of a quasi-experimental design is found in a study by Mounir, Mahdy, and Fatohy (2003). This study assessed whether a reproductive health education program could impact the knowledge and attitude of Egyptian women. Since it is not practical to randomly sample the entire female Egyptian population, the researchers sampled female students at Alexandria University, and conducted a quasi-experimental study using pre-post testing control group. An example of a non-experimental design is found in a study by Monina, Falzetti, Firetto, Mariani, and Caputi (2006). These researchers analyzed an existing relationship between subjective and objective measures of pain through an observational correlational study.

The results yielded by experimental designs allow researchers to infer true cause and effect due to the manipulation of an independent variable in order to observe changes in a dependent variable. Quasi-experimental designs yield trends. The decision to use an experimental vs. quasi-experimental research design hinges on certain factors. Matching instead of randomization necessitates a quasi-experimental design, as does the measurement of concepts that are not concrete, such as quality of life or context.

Non-experimental designs may provide the first step in the exploration of possible causes of phenomena through the identification of a correlational relationship. Causation cannot be inferred from non-experimental designs, only relationships. This type of design allows researchers to observe how certain factors co-exist in a natural environment, without any experimental intervention. This allows researchers a first step in testing hypotheses and theories as to what causes a certain phenomenon.

When a study is lacking in internal validity, it means that confounding variables within the study were not successfully eliminated. In other words, certain factors are present in the study that puts into question whether or not the outcomes being measured or observed are really due to the manipulated variable or the condition under study. When a study is lacking in external validity, the findings of the study can not be readily generalized to the population at large. This may be due to a lack of randomization in the sample used for the study. Internal and external validity are inextricably intertwined, and affect each other. The presence or absence of internal validity can especially influence the degree to which a study demonstrates a high level of external validity. In order to balance out the demands of internal and external validity, researchers must work diligently to ensure that the effect under study is truly due to the manipulation of the independent variable and not any confounding factors. It is also important to ensure that there is a level of randomization among the participants in the study. Overall, it is more important to design a study with strong internal validity, since without internal validity the findings of the study are uninterpretable. Once internal validity has been established, researchers can then take measures to ensure that the findings of the study can be generalized to the population at large, therefore