Validating Scientific Knowledge: a skill companion
A small set of reusable sheets that grow one way of thinking about science at Year 10 depth: how a claim becomes accepted knowledge (submission, peer review, revision, publication, replication) and how accepted knowledge keeps being refined, sometimes over decades, as better measurements arrive. Print the scaffolds once and slot them into the science lessons you already teach.
What a skill companion is
Ideas about how science works are not a topic of their own. They grow inside the units a class teaches all year. So this pack is not a full term of lessons. It is three reusable scaffolds, a map of where they fit, and a short stand-alone lesson for teaching the idea first.
Start here: five minutes
- Read the pairing map on the next page: it gives an idea from each Year 10 topic and how it was validated and refined.
- Print the referee report when students review each other’s work, and the case study when a topic touches the expanding universe.
- Cut out the cards once. They are reused all year.
- Open the free interactive unit on your board when you want a worked example.
- Run the one-page mini-lesson first if you want to teach the idea before folding it into a topic.
The step up from Year 9
The descriptor is the same in Years 9 and 10; the depth is not. In Year 9 students checked a claim with a reviewer’s checklist and followed a claim that failed. In Year 10 they write a referee report with a recommendation, and follow a measurement that was validated and then refined many times, to a disagreement between careful teams that is still open.
A validated, refined idea in every Year 10 topic
Every idea a Year 10 class learns was once a claim that had to be checked, and most have been refined since. This map gives one for each topic: how it was validated, and how it was refined.
| When you teach | The idea | How it was validated | How it was refined |
|---|---|---|---|
| Heredity: Cell Division and Inheritance (AC9S10U01) | Mendel’s rules of inheritance | Published in 1866 and largely ignored; three researchers independently got the same ratios around 1900 | Joined to chromosomes, then to DNA, in the twentieth century |
| Evolution by Natural Selection (AC9S10U02) | Natural selection | Darwin and Wallace presented it together in 1858; checked since against fossils, anatomy and DNA | Combined with genetics in the 1930s and 1940s |
| The Big Bang and the Universe (AC9S10U03) | An expanding universe that began hot and dense | A glow left from the hot beginning was predicted in 1948 and found by accident in 1965 | The expansion rate is still being measured; two methods disagree |
| Energy Flow and Climate Change (AC9S10U04) | Rising carbon dioxide in the air | Measured on Mauna Loa from 1958; matched by stations such as Cape Grim in Tasmania from 1976 | Ice cores extended the record back hundreds of thousands of years |
| Newton’s Laws of Motion (AC9S10U05) | Newton’s laws | Published in 1687 and tested for two centuries on everything from cannonballs to planets | Refined by relativity for very high speeds; still used everywhere else |
| Atomic Structure and the Periodic Table (AC9S10U06) | The periodic table | Mendeleev’s 1869 table predicted elements later found: gallium, scandium, germanium | Reordered by atomic number, then explained by electron shells |
| Reaction Types and Rates (AC9S10U07) | Collision theory of rates | Proposed around 1916 to 1918 and tested against rates measured in many laboratories | Refined in the 1930s to explain why the angle of a collision matters as well as its energy |
The three moves, and the picture that backs each one
When you want a worked example on the board, open the interactive unit and use the picture that matches the move.
- Follow a claim from rejection to acceptance as evidence builds: “From a rejected idea to plate tectonics”.
- Know what review checks, and what it does not: “Which of these does peer review judge?”.
- Release a surprising result responsibly: “How should a surprising claim be released?”.
Review it like a journal would
A student research group has sent this short paper to the class journal. As a referee, you summarise it, list its major faults (ones that would change the conclusion) and minor ones, and recommend: accept, minor revisions, major revisions, or reject.
Referee report
Teacher note: referees judge the work, not the people. Ask for a report the authors could act on.
How fast is the universe expanding?
The Hubble constant says how fast galaxies move apart for every megaparsec of distance between them (a megaparsec is about 3.26 million light-years). Its first measured value was published in 1929. Since then it has been checked, corrected and refined, and it is still being argued about. Values are in kilometres per second per megaparsec, rounded.
| Year | Value | Who | How |
|---|---|---|---|
| 1929 | about 500 | Edwin Hubble | Distances from variable stars in nearby galaxies, speeds from red shifts |
| 1952 | about 250 | Walter Baade | Found there are two kinds of those variable stars; the distances had been too small |
| 1958 | about 75 | Allan Sandage | Corrected another error: some “stars” were glowing gas clouds |
| 2001 | 72 ± 8 | A Hubble Space Telescope team | Many distance methods checked against each other |
| 2018 | 67.4 ± 0.5 | The Planck satellite team | From the cosmic microwave background and a model of the universe |
| 2022 | about 73 ± 1 | A supernova team | From variable stars and exploding stars in nearby galaxies |
What does each step of validation do?
Cut out the twelve cards. Six name a step by which scientific knowledge is checked; six say what a step does. Pair each step with what it does. Then put the six steps in the order a new claim usually meets them.
Teacher note: the answer sheet lists the pairs and the usual order. Some steps, such as sharing data, can happen at several points.
Submit, review, revise
The class runs its own small journal. Each group tests a simple question (does a two-ply paper towel hold more water than a one-ply one?), writes it up as a five-sentence paper, and submits it to another group, which reviews it with the referee report. Groups then revise and reply to every point. The lesson shows what peer review catches, what it cannot catch, and why replication by others still matters.
We are learning to
- describe the steps by which a claim is checked before and after publication,
- review someone else’s work fairly, by its method and evidence,
- revise our own work in reply to a review.
Success criteria
- I can write a referee report with a major fault, a minor fault and a recommendation.
- I can say one thing peer review cannot check, and what would.
You need
- for each group: one-ply and two-ply paper towels of the same size, a measuring cylinder, a tray, and water,
- the referee report and case study (scaffolds 1 and 2), and the cards from scaffold 3, one set per table. Wipe up spills.
Lesson flow (about 30 minutes)
| 8 min | Test and write Groups soak each towel, let it drip for 30 seconds, squeeze the water into the cylinder, and write five sentences: question, method, results, conclusion, one limitation. Ask: “Would another group be able to repeat your test from your five sentences alone?” |
| 10 min | Review Groups swap papers and complete the referee report, with a recommendation. Show the board (see the box). Ask: “Did the reviewers see the experiment? What can a review check, and what can it not?” |
| 7 min | Revise and reply Papers go back. Authors revise and write a one-line reply to each point. While they work, tables match the validation cards. Ask: “One reviewer asked for something you disagree with. How do you answer without just refusing?” |
| 5 min | Replicate Two groups that tested the same towels compare results: does the finding replicate? What would the class journal publish? |
Running it shorter? Stop after the review, and revise and replicate in your next science lesson.
Watch for these ideas
- Thinking peer review proves a result true: it checks the reasoning on paper; replication tests the result.
- Reviews that judge the authors, or only say “good job”.
- Treating a corrected or refined value as proof that science was wrong.
Make it easier, make it bigger
- Easier: give groups a five-sentence template and a referee report with tick boxes for common faults.
- Bigger: one group secretly leaves a fault in its method. Does review catch it, or only replication?
Answers and look-fors
The next sheet has a model referee report, the case study answered, the card pairs in order, and a quick three-level guide.
Answers and look-fors
A model referee report
Summary: the authors left three bottles of different colours in the sun and found the black one warmest, and conclude that black bottles always warm fastest. Major faults: No starting temperatures were recorded, so the rise in each bottle is unknown. One reading of one bottle of each colour: no repeats, so chance cannot be ruled out. The bottles may differ in material, size or position on the sill, not just colour. “Always” and “all” go far beyond three bottles on one afternoon. Minor faults: The volume of water and the bottle material are not stated, so others cannot repeat it. The title states the answer more strongly than the data allows. The authors should record the start temperature, use identical bottles that differ only in colour, repeat at least three times, swap positions, and state the conclusion for these conditions only. Recommendation: major revisions, since the question is good but the method cannot yet support “black bottles warm fastest in sunlight”.
Case study
The value fell because the distances were wrong: Baade found there are two kinds of the variable stars used to measure distance, so galaxies were further away than thought, and Sandage found that some “stars” used as markers were glowing gas clouds. Hubble’s discovery, that more distant galaxies move away faster, was right and has been confirmed again and again; his number was refined. The 2018 value (67.4 ± 0.5) and the 2022 value (about 73 ± 1) do not overlap, even with their uncertainties: a real disagreement, often called the Hubble tension. It could come from an unnoticed error in one method’s distance ladder, or from something missing in the model of the universe used to read the microwave background. It will be settled by more measurements from independent methods, published, reviewed and replicated.
Validation cards (pairs, in the usual order)
| Cards | Step | What it does |
|---|---|---|
| G and A | A journal editor reads the submission | Turns away work that is off-topic or clearly not ready, before reviewers see it. |
| D and J | Two or three independent experts review it | Look for flaws in the method, analysis and reasoning, and ask for changes. |
| E and L | The authors revise and reply to every point | Fix the problems, or explain with evidence why a change is not needed. |
| K and F | The data and code are shared openly | Let anyone check the analysis and reuse the data. |
| B and H | Other teams repeat the work | Show whether the result holds up when people with no stake in it try again. |
| I and C | A paper is corrected or retracted | Fixes the record when errors or misconduct are found after publication. |
A quick three-level guide
| Move | Working towards | At standard | Beyond |
|---|---|---|---|
| Review | praises or criticises the authors | names major and minor faults in the work and recommends | suggests the changes that would make the conclusion sound |
| Validation | thinks publication makes a claim true | explains what review, replication and correction each add | explains why replication by independent teams outranks one paper |
| Refinement | sees a changed value as science being wrong | explains how better methods refined a value | weighs an open disagreement and says what evidence would settle it |
In the mini-lesson, review usually catches a missing volume or towel size, but only another group’s test shows whether the result replicates. A fault hidden in the method often slips past review and shows up only when someone repeats the work.