Scientific Inquiry and Experimental Design
This domain tests whether you can turn a question into a testable hypothesis, identify what was changed and what was measured, and judge whether a stated conclusion is actually supported by the data. Items usually present a scenario - a student comparing detergents, a trial of a fertilizer, a survey of two neighborhoods - and ask you to name variables, spot a design flaw, or choose the better of two designs.
Variables and controls
- Independent variable: the factor the experimenter deliberately changes.
- Dependent variable: the response that is measured; it should be quantitative wherever possible.
- Controlled (constant) variables: everything held fixed so the comparison is fair.
- Control group: receives no treatment or a placebo; it supplies the baseline against which the treated group is compared.
- Confounding variable: an uncontrolled factor that changes along with the independent variable, so cause cannot be assigned.
What makes a design trustworthy
- Random assignment of subjects to groups spreads nuisance variation evenly across groups.
- Replication - many trials, many subjects - damps random error and raises reliability.
- Blinding, single or double, stops expectations from shaping measurement or reporting.
- A large, representative sample supports generalization; volunteers from one site do not.
- Validity means the test measures what it claims to measure; reliability means it gives consistent results. A balance reading 0.5 g high every time is reliable but not valid.
Worked example
A student predicts that more light raises the rate of photosynthesis. Pondweed is placed in water and oxygen bubbles are counted per minute at 20, 40, 60 and 80 cm from a lamp, with water temperature, sodium bicarbonate concentration, plant species and counting time held constant.
Reasoning: the independent variable is distance from the lamp, used as a proxy for light intensity; the dependent variable is bubbles per minute; the controlled variables are temperature, bicarbonate, species and time. Because distance is only a proxy, the conclusion should read "as distance from the lamp increased, bubble rate fell" rather than asserting a mechanism. A dark-trial control confirms the bubbles are light-driven and not caused by handling.
Traps that cost marks
- Naming the dependent variable as the thing that was manipulated.
- Calling a repeated measurement a "control."
- Concluding causation from a correlation - ice-cream sales and drowning rates both rise in summer.
- Missing a second unintended difference, such as fertilized plots that also received more water.
- Writing "the hypothesis was proven"; data support or fail to support a hypothesis.