Experimental design is a traditional approach to conducting quantitative research. Whether an educational practice or idea makes a difference for individuals can be tested by quantitative researchers. Also known as group comparison studies, experimental design research allows the researcher to determine whether an activity will make a difference in results for participants. One group is given a set of activities, or intervention, and the other group is not.
The first step in experimental design is deciding if an experiment addresses a research problem; experimenters need to know if a practice will influence an outcome. The researcher then will form a hypothesis about outcomes, which will test cause-and-effect relationships. Once a researcher has decided on the experimental unit of analysis, they will identify study participants. Next, researchers will set levels of treatment and introduce it to one or more experimental groups. Choosing a type of design requires several decisions based on the experience level of experiments, and conducing the experiment likely involves several procedural steps. Upon concluding researcher, data is organized and analyzed and compiled into an experimental research report.
When researchers want to establish possible cause and effect between dependent and independent variables, they use an experiment to determine whether the independent variable causes the dependent variable. Some experiments require random assignment, where participants are assigned to random groups. Random assignment is the “most rigorous” approach. Through random assignment, researchers control any influences in the selection of participants that are likely to affect the outcome. Random assignment allows for this control of extraneous variables. After selecting participants, researchers randomly assign them to treatment condition or the experimental group. The researcher physically alters the conditions experienced by the experimental unit through the process called experimental treatment. The outcome is the dependant variable that is the effect of the treatment variable. It is also the effect that is predicted in the original hypothesis. Group comparison, or the process of obtaining scores for individuals and groups, allows for comparison between the means and variance. Researchers hope to design experiments so that the inferences are true or correct. Threats to validity are reasons for why researchers can be wrong, relating to covariance, causation, or whether there is a casual relationship between variations, settings, and treatments.
There are several types of experimental designs, each with their own use and application. Between group designs include true experiments, quasi-experiments, and factorial designs. Between group designs are most frequently used when the researcher is comparing two or more groups. Within group, or individual designs, involve time series experiments, repeated measures, or single subject experiments. Individual designs are used when the number of participants is limited, and if the examiner is interested in single individuals.
The Department of Pensions and National Health, now Health Canada, ran nutrition experiments on First Nation communities in The Pas and Norway House in northern Manitoba and Residential Schools between 1942 and 1952. These experiments involved nutrient-poor isolated communities, and were designed to uncover the importance and optimum levels of the then-newly discovered vitamins. Finding the results of these studies, and the journal articles is difficult because these experiments resulted in a 1946 trial through the Nuremberg Code of Medical Ethics.
This research is an example of experimental design. Two physicians created an experiment which included control and treatment groups of malnourished children. One group was given supplements of riboflavin, thiamine, and ascorbic acid supplements. In another, children were given thiamine, niacin, and bone meal. Efforts were made to control as many factors as possible, including withholding dental care to see if there were further relations between dental hygiene and malnutrition. Despite parents not being informed, or giving consent, the research is experimental design because at least one group was given a set of activities, or intervention, and one group was not.
A correlation is a test that determines the tendency or pattern for two or more variables to vary consistently. Correlational Research allows researchers to relate variables, rather than manipulate a variable in an experiment. Researchers use the correlation to describe and measure the degree of association or how variables influence each other. Correlational research is less rigorous, but it allows for predicting outcomes.
Once a correlational study has been determined as the best way to address a research problem, researchers must identify individuals to study. Individuals should be randomly selected. Two or more measures for each individual in the study should be identified, as the idea of correlational research is to compare participants. Through data collection, researchers must also monitor for potential threats. Then, a researcher will analyze the data and represent the results, concluding with an interpretation.
Explanatory design, or relational research, explains the association between or among variables. Relational research is used when a researcher is interested in the extent to which changes in one variable is reflected in changes in the other. Researchers collect data at one point in time for relational researcher, and analyzes all participants in a single group. At least two scores are obtained for each participant – one for each variable. The use of the correlation statistical test is reported in the data analysis, and finally, conclusions are drawn.
Prediction design is used when researchers aim to anticipate outcomes by using certain variables as predictors. The purpose is to identify variables that predict the outcome, also known as the criterion variable. Researchers typically include the word prediction in the title of their research reports. Researchers measure the predictor variable at one point, and the criterion variable later in time. Researchers conclude with a forecast for future performance.
Correlational designs contain specific characteristics, including displays of scores, the association between scores, and multiple variable analysis. Often, if there are two or more scores, the results are plotted on a graph or table. After graphing, researchers can then interpret the meaning of association between the scores through examining the direction (positive or negative) of the association, the form (linear or non linear), the form of the distribution, and the degree and strength of association. As researchers predict outcomes based on more than one predictor variable, researchers additionally need to account for the impact of each variable through multiple variable analysis.
“A total diet study and probabilistic assessment risk assessment of dietary mercury exposure among First Nations living on-reserve in Ontario, Canada”, a 2017 study, is an example of a correlational study. This diet study was constructed based on a 24 hour recall from the First Nations Food, Nutrition, and Environment Study (FNFNES), and measured contaminant concentrations from Health Canada for market foods, and FNFNES for traditional foods.
Through this research, dietary mercury intakes form the total diet study were paired with participant’s hair mercury results. Correlation analysis was conducted through contrasting percentiles of hair mercury to percentiles of seasonal mercury dietary estimates from traditional food consumption. Models including linear regression were fitted to the data, and a slope was observed as a measure of the explained variability.