Investigation Design and Methodology
Turning a Question into a Plan
A sound methodology lets someone else repeat your work. Specify variables (independent, dependent, controlled), your procedure step by step, your materials, your sample and sampling method, and your analysis plan before collecting data. Choose a design that fits the question: experimental (manipulate a variable with a control and, ideally, random assignment) to test cause; observational or correlational to describe relationships; survey to measure attitudes or self-reports; case study for depth on a single instance.
Concepts to Know
- Control and comparison groups; random assignment vs. random selection.
- Validity (does it measure what it claims?) and reliability (consistency across repeats).
- Confounding variables and how controls reduce them.
- Sampling: random, stratified, convenience; sample size and representativeness.
- Ethics: informed consent, confidentiality, minimal risk, and review approval for human subjects.
Traps in Questions
- Claiming causation from a correlational design.
- Confusing random assignment (to groups) with random sampling (from a population).
- Omitting a control, so improvement cannot be attributed to the treatment.
- Convenience samples presented as representative of a whole population.
- Ignoring consent or privacy when surveying peers.
Worked Example / Rule of Thumb
Rule of thumb: ask "What else could explain this result?" and build a control for it. Example: testing whether a review game raises quiz scores - give both classes the same quiz, same time, same teacher, and randomly assign which class gets the game. Without the comparison class, higher scores could reflect an easier quiz or a stronger group.