Introduction
Experiments are designed to test the hypothesis. In an experimental design or design of experiment (DOE), participants are assigned to variable conditions in an experiment. Usually, experiments are designed in such a way that participants are divided into two groups, i.e. a control group and an experimental group, and then a change is introduced in the experimental group to determine the outcome. There are two important types of experimental design; True experiments and Quasi-experiments. Both types of experimental designs are used to check the cause of some particular phenomenon.
Types of each design
True Experiments
In a true experiment, subjects or participants are randomly assigned to different treatment conditions. In this type of design, all the necessary factors having an important affect on the phenomenon of interest are completely controlled.
Quasi-Experiments
In a quasi-experimental design, subjects or participants are not randomly assigned. This design of experiment is usually performed, when it is not practical or possible to control all the important factors.
Similarities between true and quasi-experimental designs
True and quasi-experimental designs are similar in some aspects such as
In both designs, subjects or participants of the study have to go through certain treatment or condition,
In both designs, certain outcome of interest is determined, and
The experimenters check whether differences in the outcome are caused by the treatment or not.
Differences between true and quasi-experimental designs
True and quasi-experimental designs have some differences such as
Participants are randomly assigned to the control group or the treatment group in the true experiment, whereas in the quasi-experimental design, participants are not randomly assigned,
Quasi-experiment is different from the true experiment in that the groups are different not only in terms of the experimental treatment but also in certain other unpredictable or unknowable ways, and
In case of quasi-experimental design, there could be several “rival hypotheses” to explain the observed results.
Validity of Results
Validity of an experiment helps in determining, whether the results of an experiment can be trusted or not. There are two types of validity of experiments; internal and external. Internal validity is related to the validity of factors from inside the study as, for example, performance of researchers. It shows a good construct of the study. An experiment can have a good internal validity, if bias, confounding variables, and random error, are properly checked. On the other hand, external validity is related to the validity of factors from outside the study as, for example, performance of experiment in other labs or settings. It helps in generalizing the study results to a larger population.
Threats to internal validity
An important threat to internal validity arises, when participants of the study either refuse to take part in the experiment or drop out from the experiment. This threat can be considered as the experimental mortality. An example for this threat is the dropping out of patients from the active treatment group while comparing an intranasal spray with placebo in removing symptoms of allergy. Other threats to internal validity include history, testing, instrumentation, and selection. History can become a threat to internal validity, when certain factors external to the condition of the subject or experiment occur with the passage of time. Testing threat can occur, when repeated testing of an experiment is done without disturbing the intervention. Instrumentation threat can arise, when study results are changed as a result of changes in instrumental calibration or changes in observers. Selection of subjects is an important issue in internal validity. When subjects are not randomly assigned to treatment groups, threats to internal validity can develop. Moreover, interaction of the selection threat with any of the other threats can also affect the internal validity (Thomas, Silverman, & Nelson, 2015).
Confounding variables. Confounding variables are those, which are not controlled or eliminated by the researcher, thereby affecting the internal validity (Mangal, & Mangal, 2013). It can severely affect the connection between the dependent variable and independent variable that can result in incorrect analysis of results. Confounding variables can become a threat for quasi-experimental design. Confounding variables can be removed or reduced by a proper experimental design along with constant checks on the study.
Threats to external validity
Among the important factors that can result in disturbance to external validity include reactive effects of testing and experimental arrangements, interaction of the experimental treatment with selection biases, and multiple-treatment interference (Thomas et al., 2015). As a result of reactive effects of testing, pre-test could change the response of subjects in such a way that it could not be generalized. In case of reactive effects of experimental arrangements, an effect can arise due to the fact that subjects know that they are participating. In case of interaction of the experimental treatment, some selection factors could interact with the experimental treatment that would not be present in case of random selection. In multiple-treatment interference, same subjects could receive two or more treatments making it difficult to generalize the results to single treatments (Michael, n.d.).
Conclusion
Experimental designs are helpful in finalizing the hypothesis. True and quasi-experimental designs have been developed to work on the hypotheses in different settings as, for example, quasi-experimental design is helpful, when true experimental design don’t work or it significantly restricts the external validity.
It has been reported that it is almost impossible to have high degrees of both internal and external validity as, for example, strict controls for internal validity can make it difficult to generalize the results according to the real world. On the other hand, experiments with high level of external validity show weak internal validity. However, random selection of subjects and random assignment to variable treatments can help in controlling most of the threats to internal and external validity (Thomas et al., 2015).
References
Mangal, S. K., & Mangal, S. (2013). Research Methodology in Behavioural Sciences: PHI Learning.
Michael, R. S. (n.d.). Threats to Internal and External Validity. Retrieved June 19, 2015, from http://www.indiana.edu/~educy520/sec5982/week_9/520in_ex_validity.pdf
Thomas, J. R., Silverman, S., & Nelson, J. (2015). Research Methods in Physical Activity, 7E: Human Kinetics Publishers.
