Skip to content

Science · NCEA Level 1

91923 · Evaluating science claims

Judge how scientific language and conventions are used in claims.

Open LearnOpen PractiseOpen Exam

Key concepts

  • Find the claim before you judge it

    A claim is a statement that something is the case - that a product works, that a level is safe, that one thing causes another. Advertisements and articles wrap claims in description, story and opinion, so the first move is to write the claim in one plain sentence: "drinking this powder makes you recover from exercise faster". Only then can you ask what evidence would be needed to support it. Vague marketing language often hides the fact that no testable claim was ever made - "supports your natural wellbeing" cannot be checked, which is frequently the point.

  • Source: who is saying it, and what do they know?

    Weigh the source on expertise, independence and transparency. Does the person or organisation have relevant qualifications or a track record? Are they independent of the product, or do they sell it? Do they name their method and let anyone see the data? A company reporting a trial it ran on its own product, without saying who conducted it or how many people took part, is the weakest end of this scale. A monitoring report that names its methods, its laboratory and its funding sits at the strong end - even if the organisation is small.

  • Purpose: what does this claim want you to do?

    Purpose is what the claim is for - to sell, to inform, to persuade a decision, to attract attention, to reassure. Purpose does not automatically make a claim false, but it tells you where to look for weakness. A claim written to sell will present its most favourable number and quietly omit the rest; a headline written to attract clicks will state a tentative finding as a certainty. Naming the purpose specifically ("to sell a subscription", not just "to persuade") is what turns this into a real evaluation point.

  • The language of science used as decoration

    Scientific-sounding words carry authority even when they mean nothing in the sentence. Watch for terms borrowed from science and used loosely - "energy frequencies", "activated", "quantum", "structured" - and for words with a technical meaning being used to imply safety or effectiveness: "detox", "boosts immunity", "chemical-free". Everything is made of chemicals, including water; "detox" has no accepted meaning outside medicine, where it describes a clinical treatment. The test is simple: can you say precisely what the word claims is happening, and how you would measure it?

  • Vague terms, missing comparisons and unstated baselines

    "Up to 84% of users saw an improvement" tells you the best result anyone got, not the typical one. "Twice as effective" invites the question: than what? "9 out of 10 people agree" is meaningless without knowing how many were asked, how they were chosen, and what they were asked. "Clinically proven" sounds regulated but is not a guarantee of anything without a named trial. Each of these is answerable: name the missing number, and say what it would take to make the statement checkable.

  • The conventions science uses to make evidence trustworthy

    Science has agreed conventions because they are what make a result believable to a stranger. A stated sample size, so you can judge whether the result could be chance. A control or comparison group, so an effect can be told apart from what would have happened anyway. Blinding, so expectation does not create the result. A described method, so others can repeat it. Units, so numbers mean something. Publication and peer review, so specialists have looked for the flaws. When a claim is missing several of these, say which ones and why each matters here.

  • Reading graphs and statistics carefully

    A graph is an argument. Check the y-axis: starting it at 78 rather than 0 turns a two-point difference into a cliff. Check the units and whether they change part-way. Check the range shown - a decade of flat data cropped to the three years that rise is cherry-picking. Ask what a percentage is a percentage of: 100% more sounds enormous, but not if it is 2 people rather than 1. And ask whether an average hides the spread, since a mean improvement can come from a handful of large responses among many with none at all.

  • Correlation is not causation

    Two things varying together can mean one causes the other, or that both are caused by something else, or that the causation runs the other way, or that it is coincidence. The useful move in an answer is not to recite the phrase but to name the alternative: students who buy a tutoring app may already be the students with the most family support and study time, which is a confounding variable. Then say what evidence would settle it - typically a comparison between similar groups where only the factor of interest differs.

  • Audience: who is this aimed at?

    Claims are written for someone. A supplement post aimed at teenagers before exams, a fertiliser advertisement aimed at farmers reading a trade magazine, and a media release aimed at parents all choose different evidence and different language for a reason. Identifying the audience explains the choices in the text - which fears it plays on, which words it assumes will impress, and what it leaves out because that audience is unlikely to ask.

  • Putting it together: one whole, justified evaluation

    This external is judged as one response, so a strong answer runs as a single argument. State the claim and its source and purpose. Take the two or three features that matter most - the missing control group, the truncated axis, the vague quantifier - and explain what each one does to the claim's credibility. Weigh the genuine strengths against the weaknesses, since almost no claim is entirely one or the other. Say what better evidence would look like. Then give an overall judgement that follows from what you actually said, rather than a verdict announced at the start and repeated at the end.

Assessment

External · grade-score marked.

This is an external achievement standard. Whetū offers original exam-style questions marked on the NCEA grade-score scale. Grades are estimates, not official results.

A mock paper is available in Whetū: Mock paper · Evaluating science claims.

Learn

4 authored Learn units for this standard.

  • Source, purpose and who is speaking

    Before you can say how well a piece of communicated information supports a science-related claim, you have to know who is speaking and why they published it. This page is the first move of Standard 91923: name the source, name the intended purpose specifically, and write the claim in one plain sentence. Later pages take the language, the graphs and the whole judgement. This one owns the speaker.

  • The language of science

    A science communication is not a pile of facts with decoration on top. It is a set of word choices: discipline-specific names, operational labels, verbs that state a relationship, and words that state how sure the writer is and how far the claim is allowed to travel. Science language is that wording. Your job is to quote the phrase from the resource and say what it is doing in that sentence, not to hand in a list of science-y words or a dictionary definition pulled from another topic.

  • Conventions, data and graphs

    A graph, a table, a unit beside a number, or a chemical formula written the agreed way are all science conventions. This page is about reading those conventions in communicated information — a constructed harbour bulletin, a visitor-centre pamphlet, a product caption — and saying what each one lets a stranger check. It is not about designing a trial, drawing a graph from raw data, or deciding whether a product is good. The job is to treat the display as evidence: a convention either makes a number checkable, or it hides the check.

  • Evaluating claims in full

    The three neighbouring pages taught you to name who created a communication, why it was published, the science-related claim it makes, and the language or conventions it uses. This page assumes those can already be named. Its job is the last move: carry that inventory into one connected evaluation of evaluating how well the language or conventions support the claim you actually wrote down.

Practise

32 Practise questions in “Evaluating science claims”. Feedback here is formative and is not an official NCEA grade.

  • Source and purpose

    Ask who is making the claim and what they gain from it.

  • The language of science

    Strip science-sounding words back to what they actually state.

  • Vague terms

    Name the number or comparison a vague phrase leaves out.

  • Conventions and data

    Hold evidence to sample size, comparison groups and a stated method.

  • Judging evidence

    Tell a testimonial apart from a result anyone could check.

  • Graphs and statistics

    Read the axis, the units and the base number before the shape.

  • Correlation and causation

    Name the alternative explanation instead of reciting the slogan.

  • Audience

    Work out who a claim is aimed at, and what that explains.

  • Judging fairly

    Credit a strong source as readily as you flag a weak one.

  • The full evaluation

    Weigh it up and land on a judgement that follows from your reasons.

Sample questions

  1. What is the main purpose of this post?
  2. “We paid for the trial, ran it in-house, and chose which results went into the campaign. It's our product, and the post goes live before winter.” Whose viewpoint is this?
  3. Which changes would turn this into an evaluation? Select all that apply.
  4. Why is a “chemical-free” label scientifically meaningless?
  5. Match each phrase to the problem with it.
  6. Which changes would make this a stronger critique? Select all that apply.
  7. What does “up to 84% of users saw an improvement” actually tell you?
  8. Name two pieces of missing information you would need before this claim could be judged.

Practise 91923 in Whetū

Exam-style questions

470 original exam-style questions in Whetū. These are practice papers, not official NZQA assessments.

Back to Science.