Nature and meaning of science

Baseline differences may be confused with the treatment effect. Check initial similarity before claiming causation.

Multiple analyses of the same data are not independent replications. Check data independence along with the number of studies.

Fitting past data is insufficient if new predictions fail. Evaluate predictive power along with explanation.

Scattered values indicate low precision while clustering around the true value may indicate accuracy on average. Distinguish the two.

Changing a claim after seeing the result weakens testability and transparency. Look for predefined claims.

Hiding unsuccessful results can make the success rate appear inflated. Identify publication bias.

Convergence of independent evidence increases confidence in a conclusion. Recognise converging evidence.

Knowledge of treatment may alter participants' feelings or behaviour. Understand the value of blinding.

The direct answer is B: low measurement validity. Measurement validity asks whether an instrument actually measures the concept it is supposed to measure. A very large sample improves the amount of data and may support generalisation, but it cannot repair a basic mistake in what was measured. The reasoning is: the study collects many observations; the instrument consistently records the wrong concept; therefore the claim is not a valid measure of the intended idea. Option A, low external validity, concerns whether findings generalise to other people, places, or situations. A large sample may even help external validity, but that is not the stated problem. Option B is correct because the instrument lacks correspondence with the intended construct. Option C, high precision, would mean closely clustered or finely measured readings; even precise readings can be precisely wrong and do not solve invalid measurement. Option D, high reproducibility, means others can repeat the procedure and obtain similar results; repeatability also cannot make the wrong concept the right one. Exam cue: ask first, “Did we measure the right thing?”

The direct answer is A: a real effect may remain undetected. If two levels of an independent variable are very close, the true change may be smaller than the instrument can reliably distinguish. The reasoning is: make a small change, take a reading, compare it with another reading, and consider the instrument’s resolution and measurement error. If the difference is hidden within that error, the experiment may show little or no visible effect even when an effect exists. A is correct. B is wrong because no result automatically proves causation; controls and sound design are needed. C is wrong because a weak instrument does not become accurate by itself. D is wrong because a control group remains useful for comparison. Memory cue: small effect plus insensitive instrument can mean “not detected,” not “not present.”

Simplicity is useful but the claim must adequately explain evidence. Evaluate both adequacy and simplicity.

The direct answer is A: participants may give socially acceptable answers. A teacher’s presence can create pressure, fear of judgment, or a wish to please authority. The reasoning is: a sensitive question is asked, the participant notices the teacher, the participant worries about how the answer will look, and the reported answer moves toward what seems acceptable rather than the true view. This is social desirability bias. A is correct. B is wrong because the physical unit does not change merely because a teacher is present. C is wrong because teacher presence does not make a sample random. D is wrong because pressure can reduce objectivity rather than improve it. Privacy, neutral wording, and anonymous responses can reduce this bias. Memory cue: authority nearby may change answers, even when the question is unchanged.

A comparison group is needed to separate natural change over time from treatment effects. Before-and-after comparison alone may be insufficient.

Vague predictions may fit almost any outcome and weaken testing. Choose clear measurable predictions.

A robust result should generally remain stable across reasonable analytical choices. Identify analysis dependence.

Different measuring procedures may introduce observer-related variation. Understand the need for standardised procedures.

Precisely measuring the wrong variable does not properly test a hypothesis. Distinguish precision from validity.

Scientific consensus is evidence-based and may change with strong new evidence. Treat evidence as the final basis.

Environmental differences between groups may be confused with the treatment effect. Keep relevant conditions similar.

The direct answer is A: balanced evaluation of evidence is weakened. A scientific conclusion should consider supporting evidence, contrary evidence, limitations, and reasonable alternative explanations. Step by step, gather relevant results, compare evidence on both sides, judge its quality, state uncertainty, and then form a conclusion. Ignoring contrary evidence creates one-sided reasoning and may produce a biased conclusion. A is correct. B is wrong because the report’s treatment of contrary evidence does not itself change measurement units. C is wrong because sample size is a separate design feature; it may matter, but it is not the quality named here. D is wrong because laboratory safety concerns safe procedures, not balanced evidence evaluation. Memory cue: a strong conclusion faces inconvenient evidence instead of hiding it.

A shared wrong measurement method may reproduce results without measuring the intended concept. Check validity along with replication.

The scope of a claim should match the scope of the evidence. Identify condition-specific conclusions.

Personal experience is a limited sample and systematic evidence may be more reliable. Judge the quality of evidence.

Direct answer: Option A, no outcome would count against the claim. A scientific claim is useful only when it makes a clear prediction that could possibly be shown false. If a person changes the claim after every result so that every result is said to support it, the claim cannot be seriously tested. Step by step: state the claim before collecting results; identify what result would support it and what result would challenge it; compare the actual result with those predictions; revise or reject the claim if appropriate. Option A is correct because a claim that explains every possible outcome is not vulnerable to evidence. Option B is wrong because flexible explanations do not make every result more accurate. Option C is wrong because sample representation depends on sampling method, not on changing a claim. Option D is wrong because a hypothesis does not become a law merely by explaining outcomes. Memory cue: a scientific claim must be able to lose.

An average summarises the whole group in one number, so it can conceal the fact that some participants improved greatly while others became worse. Here, the worsening of half the participants shows that individual results, data spread, and subgroup patterns must also be examined. Option C is incorrect because an improved average does not mean that every individual improved. Exam tip: Along with the average, check the range, distribution, and results for different groups.