Nature and meaning of science

The method and reliability of an opposing result must be examined. Choose balanced evaluation of evidence in exams.

Keeping temperature constant allows the effect of light to be isolated. Control confounding variables in exams.

A scientific hypothesis should produce a prediction that can be tested. Identify testability in exams.

Closely grouped values show precision but distance from the true value shows inaccuracy. Distinguish accuracy from precision.

Selecting only supportive results is confirmation bias. Choose unbiased reporting of all relevant data.

When two events decrease together, the data may show correlation, but this does not prove that a decrease in one causes the decrease in the other. A third factor may be affecting both events. A causal conclusion needs controlled investigation, adequate evidence, and checks for other possible causes. Exam tip: correlation is not proof of causation.

A scientific theory is supported by many tests and evidence. Distinguish its scientific meaning from everyday usage.

A scientific law describes a regularly observed relationship. Do not treat it as a stage leading to theory.

Obtaining the same result independently demonstrates reproducibility. Link it with reliability.

Without a clear measurement method the quality of the result is difficult to assess. Understand the importance of operational definitions.

Random selection reduces the chance of unintentionally favouring certain types in the population. Link it with unbiased sampling.

A large sample is not automatically a good sample. Sample size tells us how many members were studied, while selection quality tells us whether those members fairly represent the population. If a sample is chosen in a biased way, some groups may be included too often and others may be left out. Repeating that biased selection with many people can produce a precise-looking result that is still unfair or unrepresentative. Increasing size can reduce some random variation, but it cannot remove systematic selection bias.

For example, asking only one neighbourhood about the preferences of an entire city may give a large number of responses, yet the city’s other neighbourhoods are not represented. The conclusion may therefore misrepresent the population. Option B is correct. A large biased sample does not become random by itself, and data remain necessary. Both sample size and the method of selection must be considered.

A model simplifies important features to aid understanding and prediction. Recognise both its usefulness and limits.

Repeatability concerns obtaining similar results while validity concerns appropriateness of measurement. Distinguish the two.

Keeping other variables constant isolates the effect of the independent variable. Identify controlled experiments for causal testing.

Presenting a post-experiment change as predetermined violates transparency. Link accurate records with scientific ethics.

The cause of an unusual value should be checked and any removal scientifically documented. Choose transparent data analysis.

A conclusion from a small region can be applied to the whole country only if the people and conditions in that region adequately represent those of the country. Therefore, similarities and differences in population, social and geographical conditions, and other relevant factors must be checked. The number of diagrams in a report or the researcher’s residence does not determine whether findings can be generalised. Exam tip: For such questions, focus on sample representativeness and generalisation of findings.

Original data and axis scales are needed to interpret a graph correctly. Critically examine visual presentations.

Without a control group it is difficult to establish that the treatment caused the change. Look for a comparison group in causal claims.

Direct answer: Option A, it is difficult to test further. A scientific explanation should not only fit observations already known; it should also lead to new predictions that can be checked by observation or experiment. Reasoning step by step: an explanation describes why events occur; a new prediction follows from that explanation; researchers test the prediction; the result can support, modify, or challenge the explanation. If an explanation makes no new prediction, there is no clear new situation in which it can be tested, so its predictive power and usefulness are limited. Option A is correct. Option B is too strong: lack of new predictions does not prove that the explanation must be false. Option C is wrong because an explanation does not automatically become a law; laws and explanations have different roles. Option D is wrong because an explanation may contain logical reasoning even when its predictions are limited. Exam cue: a useful scientific explanation must be testable and have predictive power.

Similar results from independent institutions strengthen the evidence. Identify independent verification.

Double-blind testing reduces both expectation and observer bias. Remember the purpose of blinding.

Self-reported information can be useful but may contain bias. Identify limitations of evidence.

Showing only data that support a claim can make the conclusion biased or misleading because contrary or different results are hidden. Therefore, this is a misleading presentation of selected data. Accurate measurement concerns the precision of data, not the omission of inconvenient results. Exam tip: When reading a graph, check whether the complete data set and a suitable scale are shown.