Scientific Inquiry and Investigation
Scientific Inquiry and Investigation tests the design, critique, and interpretation of empirical work: variables, controls, sampling, measurement, uncertainty, data displays, and evidence-based conclusions. It also tests whether you can separate what the data support from what you wish they showed.
Core rules and concepts
- Start with a testable question and operational definitions. 'Growth' must become, for example, 'change in dry biomass after 14 days.'
- Identify the independent variable, dependent variable, controlled variables, and control group. Random assignment reduces confounding in experiments; random sampling supports generalization.
- Replication, repeated measurement, and calibration improve reliability; valid tasks measure the intended construct.
- Report units, precision, and uncertainty; graph continuous data appropriately; avoid causal claims from correlational designs.
- Follow safety and ethical review, especially with organisms, chemicals, and human participants.
Worked example
A student tests fertilizer concentration and duckweed growth using one jar per concentration, placed on a windowsill, counting fronds after 14 days. Critique: one jar per level gives no replication; light and temperature may differ; frond count is an incomplete operational definition; no uncertainty estimate. A stronger design uses multiple jars per concentration, random placement and rotation, controlled water volume and light, a 0 g/L control, consistent counting, and a pre-specified analysis. The student can claim an association under these conditions but should not generalize to all ponds.
Traps that cost marks
- Calling every kept condition a 'control group'; the control group is the comparison treatment.
- Changing several variables and attributing the result to one.
- Treating a small, nonrandom sample as representative.
- Ignoring uncertainty or overstating statistical significance.
Key points
- Operationalize variables so another investigator could repeat the study.
- Separate control group from controlled variables.
- Replication and randomization address different threats to validity.
- Match the graph and statistic to the data type.
- Conclusions should not exceed the design and sample.