Week 6: Defensible Quantitative Design

Explain how you will select treatment and control groups if the design is a field experiment

            Control experiment is where a researcher holding or fixing constant elements in a data generating process in an experiment to measure the effects of manipulations or treatments in the best way. Control experiment is done to reduce the effects of unobservable and observable confounding attributes of an experiment. In this case, researchers take control of laboratory environment where they measure more factors that make attributes observable and hold other factors that influence the change of characteristics of the elements that received treatments (Creswell, 2013).

            In field experiment, there are two selection criteria through which samples from a target population are selected for the purpose of conducting a given study. In this project, if the design is a field experiment, random assignment and random selection will be used to select treatment group and control groups. Random sampling is the process through which subjects are selected from a population whereby any participant in the study stands an equal chance of selection (Campbell & Stanley, 2015). Although random sampling is not required in the in experiments, external validity of the experiments can be achieved if random samples are used. On the other hand, random assignment is the procedure of determining how subjects used in the experiment are assigned to treatment groups. There are within and between subject designs whereby in between subject design, subjects are assigned single treatments and researchers compare treatment groups and to the control group or to each other. On the other hand, in within subject designs, subjects serve as the control of each other where the characteristics are monitored and compared before and after treatment (Creswell, 2013).
Explain what techniques you might use to address selection bias if the design is a quasi-experiments

Creswell (2013) stated that selection bias in quasi experiments exists when treated repossesses differ from control experiments in ways that treatments are not involved. Selection bias occur when subjects are not assigned randomly in experiments. To deal with selection bias in quasi experiments, researchers need to add a pre-or post-treatment measurement in the experiment. Addition of pretest measure may help in rectifying the selection-by-time interaction. On the other hand, similar method is used in randomized experiments to improve efficiency in the absence of selection bias (Punch, 2013).
Explain how you might address internal validity if the design is a non-experimental design

            Campbell & Stanley (2015) argue that internal validity is the measure of how research design of a study concurs with the conclusion that changes in the independent variable causes any observed differences in the dependent variable. Therefore, the study of internal validity needs to address the ability of its design to concur with causal conclusion. There are 8 threats to internal validity relevant to non-experimental designs. They include maturation, regression to the mean, and selection of subjects, selection by maturation interaction, mortality, instrumentation, testing, and history. In maturation, there are psychological processes that occur within participants during experiment as a matter of change of time. On the other hand, regression to men is the tendency of participants recording extreme scores when tested and in the nest testing they record slightly lower score.

            Moreover, Punch (2013) adds that selection of subjects refers to selection of participants that lead to differences between participants in same groups. Elsewhere, selection by maturation interaction is where the treatment and non-treatment groups grow differently with no treatment administered. Also, mortality refers to the dropping of participants from the study before the completion of the experiment. Moreover, instrumentation refers to the changes observed when different measurement procedures are used. Besides, testing is when participants are tested repeatedly and recorded changes in test scores. Finally, history refers to the events that take place during the time of conducting the experiments and may affect the responses from participants on the dependent measure (Campbell & Stanley, 2015).

Reference

Campbell, D. T., & Stanley, J. C. (2015). Experimental and quasi-experimental designs for research. Ravenio Books.

Creswell, J. W. (2013). Research design: Qualitative, quantitative, and mixed methods approaches. Sage publications.

Punch, K. F. (2013). Introduction to social research: Quantitative and qualitative approaches. Sage.