Hypothesis and experimentation

A scientific hypothesis must state measurable variables and a clear relationship. In exams identify whether it can be tested.

A null hypothesis usually states that no effect or difference exists. In exams distinguish it from the alternative hypothesis.

A larger sample and repetition reduce the effect of chance variation. In exams choose replication for reliability.

Contradictory results are also scientific evidence and must be evaluated objectively. In exams place evidence above the hypothesis.

A falsifiable hypothesis can potentially be shown false by observation. In exams choose a clear measurable prediction.

The control group provides a standard condition for comparison with treatment groups. In exams identify the normal baseline condition.

Changing two factors together makes it unclear which factor caused the effect. In exams recognise this as confounding.

In a valid experiment relevant conditions other than the independent variable remain controlled. In exams apply the fair-test principle.

A better hypothesis links a cause to a measurable outcome. In exams replace vague words with quantitative variables.

Random assignment reduces bias and makes groups more comparable. In exams connect randomisation with fairness.

Similar results obtained by independent scientists demonstrate reproducibility. In exams distinguish reproducibility from simple repetition.

An unusual value should be investigated before it is removed. In exams choose retesting when an outlier appears.

Variables changing together do not prove causation because a third factor may be responsible. In exams distinguish correlation from cause.

A pre-decided measurement method reduces selective interpretation and bias. In exams remember the value of an operational definition.

Consistent repeated comparative results provide stronger evidence. In exams prefer replicated evidence over a single observation.

A constant shift in the same direction in every reading is a systematic error. In exams identify errors related to calibration.

The mean of several readings reduces the effect of random error. In exams connect averaging with random variation.

A blind test reduces bias caused by participant expectations. In exams identify subjective tests such as taste or medicine trials.

A double-blind test reduces bias from both participants and investigators. In exams look for both parties being unaware.

A placebo helps separate the effect of expectation from the effect of the actual treatment. In exams treat it as a control comparison.

A stratified random sample gives suitable representation from different classes. In exams choose a representative sample.

A line or scatter graph clearly displays the relationship between two quantitative variables. In exams match the graph to the variable type.

Large overlap indicates uncertainty so a strong claim is not justified. In exams choose a cautious conclusion when uncertainty is high.

The conclusion should match the strength of the evidence. In exams state the degree of support rather than claiming absolute proof.

Independent reproduction is not possible without a clear method. In exams look for precise materials quantities and steps.