Variables, Controls & Repeated Trials · Grade 6 Science
Chapter: Inquiry, Measurement & ModelsVariables, Controls & Repeated Trials
Learning goals
- I can name the independent, dependent and controlled variables in an investigation.
- I can explain why a control group is needed even when the treatment obviously works.
- I can explain what repeating a trial tells you that one trial cannot.
Readings for this unit
Before this unit
Read the chapter opening and the named observation or diagram. Clarify one word or unit that blocks understanding. Recall the relevant earlier measurement, matter, or energy idea.
Study the unit across three 35–45-minute class sessions. Use the worked examples after an attempt and the supplied records when actual equipment is not available.
In 1954 the United States ran the largest medical trial ever attempted up to that point. Polio was paralysing tens of thousands of children a year, a vaccine existed, and the question was whether it worked. About 1.8 million children took part, and in the strictest part of the trial 623,972 were split at random into two groups.
One group received the vaccine. The other received an injection of salt water that looked identical, and neither the children nor the doctors giving the injections knew which was which. At the end, paralytic polio occurred in about 16 children per 100,000 in the vaccinated group and about 57 per 100,000 in the group given salt water.
Every element of that design exists for a reason you will use in far smaller investigations. One thing changed, everything else held steady, a comparison group that got nothing active, and enough children that a difference this size could not be chance.
Build your understanding
Where this fits
The last lesson was about what makes a question answerable. This one is about building the test once you have the question, and it introduces the vocabulary used in every investigation in this course: independent, dependent and controlled variables, control groups and repeated trials. The chapter investigation will ask you to design one of your own.
Use three flexible 40-45-minute sessions: the three kinds of variable; control groups and why they are needed; repeated trials and confounding variables, then the practice set, revision and the exit check. Workshop, review, quiz and exams have separate time.
Part 1 · One thing changed, everything else held
Every investigation has three kinds of variable and it is worth being strict about the words. The independent variable is the one thing you deliberately change -- the amount of light, the mass hung on a spring, the temperature of the water. The dependent variable is what you measure to see what happened: the height of the seedling in millimetres, the stretch of the spring in centimetres, the time to dissolve in seconds. The controlled variables are everything you hold steady on purpose, and there are usually many more of them than students expect, which is why listing them is part of the plan rather than an afterthought.
Take a simple case. You want to know whether seedlings grow taller with more light, so you put one tray under a lamp for 12 hours a day and another for 4 hours. To make that a fair test the trays need the same soil, the same seeds from the same packet, the same 50 mL of water each day, the same room temperature and the same size pots. If the bright tray also sits on a warm windowsill, you have changed two things at once, and when it grows 30 mm taller you have no way of saying which change did it. The result is not wrong, it is simply unusable. A good habit is to write the controlled variables down as a numbered list before you touch anything, because the ones you forget are never the obvious ones.
Pause and explain
What goes wrong if two variables change at the same time?
Check your reasoning after answering
The result stops being attributable. If the brighter tray is also the warmer one, a 30 mm difference in growth could have come from either factor or from both together, and nothing in the data separates them. The measurement itself may be perfectly accurate; what has been lost is any ability to say what caused the difference.
Part 2 · The group you did nothing to
A control group is the comparison you set up by deliberately not applying the treatment. It sounds wasteful and it is the most important part of the design, because many things change on their own. Polio case numbers swung widely from year to year before any vaccine existed, so a vaccinated group with few cases would have proved nothing without a group measured over the same months in the same towns. The 57 per 100,000 figure is what tells you that 16 per 100,000 is low.
Two refinements in the 1954 trial are worth copying. The comparison group received a placebo, an injection of salt water indistinguishable from the real one, so that both groups were treated the same in every respect except the vaccine itself. And the trial was double-blind: neither the child nor the doctor diagnosing them knew which injection had been given. That matters because a doctor who knows a child was vaccinated may, without meaning to, record a borderline case differently, and the decision about whether a case counts is exactly where expectation leaks in. The same leak appears in school investigations. If you know which tray is supposed to grow faster, you will measure that tray slightly more generously without ever deciding to, which is why it helps to label trays A and B and have someone else hold the key until the measuring is finished.
Pause and explain
Why was the comparison group given an injection of salt water rather than nothing at all?
Check your reasoning after answering
So that the two groups differed only in the vaccine. Being injected is itself an experience, and children who knew they had received nothing could have behaved differently or been assessed differently. Matching everything except the active ingredient means any difference in outcome has only one candidate cause left.
Part 3 · Doing it again
A single trial gives you a number with no way of judging it. Repeat the measurement and you learn the spread, which is what tells you whether a difference is real. If five measurements of a seedling give 62, 65, 64, 66 and 63 mm, the readings scatter over about 4 mm, and a 2 mm difference between two trays is smaller than your own noise. The same comparison with a 30 mm difference is on completely different footing, and you only know which situation you are in because you repeated.
Repeating also exposes the occasional reading that does not belong -- a seedling that was knocked over, a timer started late. The honest response is to record it, say what happened, and explain why you excluded it, rather than quietly dropping it because it spoiled the pattern. In the polio trial the equivalent protection was sheer size: with 623,972 children, a run of bad luck large enough to produce a gap between 16 and 57 per 100,000 is so unlikely that it can be set aside, which is what large numbers buy you. There is a limit worth knowing: repeating a measurement reduces random scatter but does nothing at all about a fault that pushes every reading the same way. A ruler with a worn end gives you five consistent, consistently wrong numbers, and no amount of repetition reveals it. That is the subject of the next lesson.
Pause and explain
What does repeating a measurement tell you that a single reading cannot?
Check your reasoning after answering
How much the number moves on its own. Five readings that scatter across 4 mm establish that any difference smaller than that is within ordinary variation and has not really been detected. One reading supplies a value with no indication of its reliability, so there is no way to judge whether a comparison means anything.
Part 4 · The variable nobody listed
A confounding variable is one that changes alongside your independent variable without being on your list, and it is the commonest reason an investigation reaches a wrong conclusion. Drowning deaths and ice cream sales both rise in July; neither causes the other, and the hidden variable is hot weather putting more people in the water and more people at the ice cream counter. Spotting the pattern is easy once it is pointed out, and genuinely hard in the middle of your own investigation.
The defence is to list controlled variables before you start, and then to ask what else might differ between your groups that you have not written down. Where you cannot hold something steady, randomising helps: the polio trial assigned children to groups by chance precisely so that unlisted differences -- diet, neighbourhood, how often they swam in public pools -- would be spread roughly evenly between the groups instead of piling up in one. Randomising does not remove confounding variables, it stops them lining up with the treatment. For a classroom investigation the same principle is easy to apply. Rather than putting the first ten seedlings in the bright tray and the rest in the dim one, shuffle them and deal them out, because the first ten out of the packet may well be the largest seeds and you would never know.
Pause and explain
How does assigning subjects at random help when you cannot control every variable?
Check your reasoning after answering
It stops unlisted differences from lining up with the treatment. Diet, neighbourhood and habits cannot all be held steady, but if people are allocated by chance those factors should appear in roughly equal measure in both groups, so they no longer offer an alternative explanation for a difference between them.
Apply the whole unit · Design a fair test
A claim says that adding 5 g of sugar to the water in a vase makes cut flowers last longer. You have 20 identical flowers, 10 vases, water, sugar and a ruler.
Your task
Design an investigation that could show the claim is false, and say how you would judge the result.
What your response should show
- Names all three kinds of variable and makes the dependent one countable
- Includes a control group and explains why the comparison is necessary
- Specifies the number of repeats and uses the spread to judge the result
- Lists controlled variables with quantities rather than in general terms
Response guidance — after your attempt
Name the independent, dependent and controlled variables, and say how you will measure the dependent one.
Describe your control group and explain why it is needed here.
State how many times you will repeat and what size of difference you would accept as real.
The independent variable is whether 5 g of sugar is added; the dependent variable is how long the flowers last, measured as the number of days until the stem bends past horizontal, since “last longer” has to be turned into something countable. Controlled variables: the same species and size of flower, 200 mL of water in every vase, the same room at about 20 °C, the same light, stems cut to the same length on the same day, and the water changed on the same schedule. The control group is five vases with plain water, needed because flowers wilt at different rates anyway and without that comparison a six-day average means nothing. I would use five vases in each group, calculate the average for each, and look at the spread within each group first: if individual flowers in the plain-water group vary by three days, then a one-day difference between the group averages is inside the noise and I have shown nothing. A difference of four or more days, with tight spread, would be worth taking seriously.
Scientific references for this unit
- US National Library of Medicine: the 1954 Salk polio vaccine field trial
- Centers for Disease Control and Prevention: polio history and vaccine data
- Understanding Science, University of California Museum of Paleontology: fair tests
- Royal Statistical Society: why randomisation and control groups matter
Parent / teacher teaching notes
Begin with the literal observation, quantity, or diagram relationship. Ask the learner to explain why the evidence supports the claim, check a relevant unit or boundary, then discuss an alternative. The supplied numbers are illustrative, not real observations. A parent can read aloud and allow a spoken rehearsal before writing. Judge science and evidence rather than political, faith, or family agreement. Use the complete data option whenever an actual procedure is unavailable or unsuitable.
The key idea
A fair test changes one thing and holds everything else steady, and compares the result against a group you deliberately did nothing to. Without that second group you cannot tell whether your change caused the difference or whether it would have happened anyway.
Worked examples
Example 1 · A fair bridge comparison
- Identify both geometric differences.
- Explain how another changed factor could affect load.
- Specify matched conditions and repeated fresh samples.
Example 2 · Transfer to a different case
- Identify the distinction and conditions in this new case.
- Use the relevant mechanism or quantity; preserve units and the chosen boundary.
- Explain why the answer fits and what further evidence would be needed.
Example 3 · Keep other important conditions comparable · a focused example
- State the literal condition or observation in the example.
- Apply the distinction explained in this teaching part.
- Check whether the claim stays within the given evidence.
Avoid these mistakes
- Misleading claim: Repeating a confounded test makes its causal claim valid.
- Misleading claim: A control treatment automatically keeps every other condition equal.
- Omitting units, measurement timing, or a model boundary makes the evidence hard to interpret.
- Treating illustrative classroom data as real field measurements or copying a model is not an independent explanation.
Check your understanding
Which variable is the outcome?
Discuss the answer
The number of identical counters held before collapse, using the stated placement and failure rule.
What two changes confound this comparison?
Discuss the answer
Strip width and material or thickness. Use the same material and thickness while varying width if width is the intended factor.
Why keep the count of 8 rather than only 10?
Discuss the answer
Keeping all planned trials represents the observed variation and prevents selecting only the result that favors a claim.
Does an unwrapped reference guarantee a fair test?
Discuss the answer
No. It supplies a reference treatment, but the other influential conditions and measurement procedure must still be comparable.
Sustained scientific explanation and revision
These activities are scheduled in the parent plan. Save a first response, a correction, and the reason for the change in the science notebook.
Activity 1 · Variables, Controls & Repeated Trials · Model and explain
Read A Neat Result Is Not Yet an Explanation and the linked unit teaching; use its complete supplied case and diagrams.
Your task
For Variables, Controls & Repeated Trials, study Parts 3–4 and attempt the worked case: Explain why the comparison does not isolate width and propose one improvement. Explain the steps before opening the model.
Success criteria
- An accurate scientific claim answering the actual question.
- Relevant evidence or a labeled model with quantities and units when applicable.
- Reasoning that connects evidence with the mechanism rather than only repeats it.
- A specific method, boundary, scale, or alternative-explanation limit.
Hint
Begin with the actual observation, quantity, or diagram relationship. Explain how it bears on the question before making a broad claim.
Response guidance — discuss after your attempt
Width and length both change. Length can affect bending, so the larger load does not isolate width. Keep length, material, support spacing, and counter placement the same, vary width, and use several fresh strips for each treatment. Compare all recorded loads under the same failure rule.
Activity 2 · Variables, Controls & Repeated Trials · Practice, revise and exit
Read A Neat Result Is Not Yet an Explanation and the linked unit teaching; use its complete supplied case and diagrams.
Your task
For Variables, Controls & Repeated Trials, complete practice questions 1–6 and both choice checks, then answer independently: Three repeated trials all use different counter sizes. Does repetition fix the method? Save a correction explaining one changed claim or method.
Success criteria
- An accurate scientific claim answering the actual question.
- Relevant evidence or a labeled model with quantities and units when applicable.
- Reasoning that connects evidence with the mechanism rather than only repeats it.
- A specific method, boundary, scale, or alternative-explanation limit.
Hint
Begin with the actual observation, quantity, or diagram relationship. Explain how it bears on the question before making a broad claim.
Response guidance — discuss after your attempt
No. Counter count is not comparable when counter mass differs. Use identical counters or measure load mass, then repeat the corrected method.
Words to know
- independent variable
- The factor deliberately changed for the intended comparison.
- dependent variable
- The response measured under the stated method.
- evidence
- A relevant observation, measurement, or record used to evaluate a claim.
- system boundary
- The chosen limit specifying what is included and which exchanges cross it.
- uncertainty
- A stated limit in measurement or knowledge, distinct from a careless mistake.
In the real world
Before a new medicine is approved, it is tested in a clinical trial where one group gets the real drug and a control group gets a look-alike placebo with no medicine — everything else kept the same. Comparing the two groups is the only way to know whether the drug itself actually works.
Exit check
Three repeated trials all use different counter sizes. Does repetition fix the method?
Check your explanation after answering
No. Counter count is not comparable when counter mass differs. Use identical counters or measure load mass, then repeat the corrected method.
Explain your reasoning without notes. If a key idea is missing, revisit the lesson and try again.