Experimental Design is a structured approach to planning and conducting experiments in order to:
Unlike observational studies, experimental design allows researchers to control conditions and draw cause-and-effect conclusions.
The factor that the researcher manipulates or changes. It is the presumed cause.
Example: In testing a new website layout, the layout design is the independent variable.
The outcome that is measured to see if it changed. It is the presumed effect.
Example: The number of user clicks is the dependent variable.
A group that does not receive the experimental treatment. It provides a baseline for comparison, allowing researchers to isolate the effect of the independent variable.
The group that receives the treatment or intervention being tested.
All other factors that are kept constant to ensure that only the independent variable affects the dependent variable.
Randomization is the process of assigning subjects to groups by chance. This:
A/B Testing is a common experimental design used in data science and technology:
Example: A company tests two email subject lines to see which generates more opens.
In data science, original data can be gathered through several approaches:
| Method | Description | Example |
|---|---|---|
| Survey | Structured questionnaire given to a sample population | Likert-scale satisfaction survey |
| Qualitative Interview | In-depth conversation to gather detailed opinions | User experience interview |
| Prototype Testing | Testing a working model to collect performance data | Testing a new app feature |
| Simulation | Using software to model real-world scenarios | Simulating traffic flow |
| Advanced Search | Using search operators and databases to locate existing data | Boolean searches in academic databases |
When designing a data collection plan: