What is Experimental Design?
The purpose of experimental design is to test and validate the effects of one or more variables (independent) on another variable (dependent) (Miller et al., 2020).
An experiment involves an organized inquiry focused on testing a hypothesis or discovering new information (Casler, 2015).
This design follows rules and procedures when assigning interventions or treatments to variables (Casler, 2014).
A researcher manipulates one or more variables to determine the effects on another variable while randomly assigning subjects (Miller et al., 2020).
A statement indicating that the study's purpose is to test or evaluate something such as an intervention, suggests an experimental design (Polit & Beck, 2021).
Variables
Independent variable - The cause or treatment
Dependent variable - The effect or outcome that is measured
Extraneous variable - Confounds the relationship between the independent and dependent variables and needs to be controlled; often called confounding variable.
Experimental design consists of:
Control Group: Subjects are not given a stimulus
Treatment/Experiment Group: Subjects receive one or more experimental stimulus. Treatment involves the experimental intervention and condition manipulated by the researcher.
If the treatment group rates more favourably on the outcome variable than the control group, then the treatment deems successful.
Randomization
Random Selection: This approach ensures each individual participating has an equal chance of being selected for the sample
Random Assignment: Members of the population are designated to the experimental or control groups
Randomization is essential to allow researchers to draw conclusions about the causality of variables
Treatment Manipulation
Manipulation influences or directs the independent variable
Intervention is directed as such one group receives it and one does not
Allows for control of the "cause" in cause-effect relationships
Placebo
A measure or pseudointervention sometimes used as a control group condition with no therapeutic value
*Placebo effect: Changes in the dependent variable attributable to the placebo.
(Bhattacherjee, 2012; Polit & Beck, 2021)
Types of Experimental Design:
True Experimental: characterized by control, manipulation, and randomization
Quasi-Experimental: uses control and manipulation to establish a cause-and-effect relationship, does not use randomization
Nonexperimental: includes research that does not employ manipulation or random assignment, researchers collect data without introducing treatments
(Thompson & Panacek, 2006).
The most effective in demonstrating the efficacy of an intervention or treatment (Thompson & Panacek, 2006).
Focused on a specific research question and prospective in nature (Thompson & Panacek, 2006).
Two-Group Experimental Designs
Pretest-posttest control group Design
Test the effects of an independent variable that can be manipulated as a treatment
Pretest-posttest control group uses random assignment of subjects to treatment and control groups
An initial measurement of the dependent variable prior to treatment (pretest) and measurement conducted after treatment (posttest) is used
Posttest-only control group design omits the pretest measurement and uses the posttest scores after intervention
When measures of dependent variables are influenced by extraneous variables (covariates)
Covariates need control to eliminate influence on the dependent variable
The pretest measurement occurs on the covariates rather than the dependent variables
Each independent variable is a factor and each subdivision is a level
Enables researchers to examine the joint effect (interaction effects) on the dependent variables
Research requires manipulating two or more independent variables for their effects on one or more dependent variables
Incomplete factorial designs do not receive any treatment
Main Effect: dependent variables show the difference between multiple levels of one factor
No change in the dependent variable is a null care
Interaction effect: One factor depends on the level of another factor
Hybrid Experimental Designs
Formed when features of established designs are combined
Randomized Block Design: the population is grouped into homogeneous subgroups known as blocks, in two or more stages to equally distribute a potentially identified extraneous variable among the groups
Solomon Four-Group: The sample is split into four groups, two control and two treatment groups. One control group and treatment group receive the pretest while the others do not.
Switched Replication Design: Implemented over two phases and three waves of measurement. The treatment group in the first phase is used as the control group in the second phase. The roles switch again for the next wave where the first phase's control group functions as the treatment group in the second phase. Essentially, all participants receive the intervention and replication of treatment/control is switched between groups. Also called Crossover design.
(Bhattacherjee, 2012; Gray et al., 2020)
Essential elements of experimental research:
(1) Researcher-controlled manipulation of the independent variable
(2) The presence of a distinct control group
(3) Random assignment of subjects to either the experimental or the control condition
(Gray et al., 2020)
When do we use Experimental Design?
Ideal use in explanatory research/causal relationships (cause-effect relationships) and tests whether an intervention causes changes in the outcome
Researchers actively introduce an intervention or treatment to address a hypothesis
Experimental design is conducted in laboratories or field settings
Experimental design can be used in physical sciences, education, and psychology
In medical and epidemiological research, the term clinical trials is used for experimental studies
Randomized controlled trials use experimental designs
(Bhattacherjee, 2012; Polit & Beck, 2021)