Print it now, with no email gate
Start here, because this is the part that usually costs a teacher fifteen minutes. This worksheet prints directly. There is no opt-in form, no newsletter box, no "enter your email to get the PDF" step, and no account to create. The printable version comes as two documents: a blank student sheet with the questions and answer lines, and a separate answer key with the explanations. Both open as an ordinary printable page, so you can send them straight to a printer or choose Save as PDF in your browser's print dialog if you want a file to keep. Print one, print both, print thirty copies. Nothing is held back.
We mention it up front because the friction is real. A teacher looking for a scientific method worksheet at seven in the morning does not want to trade an email address for a page of variable-identification questions, and does not want to discover the gate only after reading two hundred words of preamble. Worksheet sites commonly ask for an address before they release the file. This one does not, and never will.
The blank sheet and the key are generated from the same question set you see below, so they never drift apart. If we correct a question here, the printable corrects with it. That also means you can preview the whole thing on screen before you commit paper to it — scroll through the four rounds, decide which tier fits your group, and print only what you need.
One practical note about the key. It is written to be readable by a substitute teacher or a parent helping with homework, not only by someone who already teaches science. Each explanation names the reason: why a constant is not a variable, why a hypothesis has to be phrased so it could turn out false, why "supported" is the correct word in a conclusion and "proved" is not. Where more than one wording deserves credit, the accepted alternatives are printed in the answer itself rather than buried in a note, so the person holding the key can see them. If you hand the key to a student who got an item wrong, they should be able to work out their mistake without you standing over them.
If you are teaching this remotely or on a cart with no printer, the rounds work as an on-screen quiz too. Read the scenario aloud, take answers on whiteboards, then reveal the explanation. The scenarios are short on purpose so they can be read once and still be understood.
The seven steps of the scientific method
The version most often taught as "the seven steps" runs like this. One: observe something. Two: ask a question about it that could be answered by measuring. Three: do background research to find out what is already known. Four: form a hypothesis, a prediction you could be wrong about. Five: test the hypothesis with an experiment. Six: record and analyze the data. Seven: draw a conclusion and communicate it.
Each step does a specific job, and the jobs are easy to blur. Observation is description without explanation. If a student writes "the plant on the windowsill has more leaves," that is observation. If they write "the plant on the windowsill has more leaves because it gets more light," they have skipped ahead to a hypothesis and skipped the question entirely. Keeping those separate is most of what this worksheet drills.
The question step is where good investigations are won or lost. A testable question names one thing you will change and one thing you will measure. "Does the amount of sunlight change how many green leaves a bean plant grows in four weeks?" is testable: sunlight is the thing changed, the number of green leaves is the thing counted, four weeks bounds it. "Which plant is best?" is not testable, because "best" is not a measurement. Students who cannot get a workable question usually have not decided what they are going to measure.
Background research is the step classrooms cut when time is short, and it is worth defending. It is the step that stops a student spending three weeks rediscovering something that is in the textbook, and it is where they find out what a reasonable range of values looks like. It also gives them a reason to expect a particular result, which is what turns a guess into a hypothesis.
The hypothesis is a prediction with a mechanism behind it, written so that reality could contradict it. The if/then form is a scaffold, not a law: "If a bean plant gets twelve hours of light a day instead of two, then it will produce more green leaves, because photosynthesis needs light." The "because" clause is the part that separates a hypothesis from a coin flip. Note that the measurement has to match the mechanism. Stem height would be a poor choice here, because a seedling kept in the dark actually elongates for a while — it grows tall, pale and spindly on its seed reserves before it collapses — so height alone would point the wrong way.
The experiment tests exactly one thing at a time. Everything else is held constant, and where possible there is a comparison group that gets no treatment. The data step is bookkeeping: units on every column, every trial recorded including the ones that look wrong, and enough repeats that a single odd result cannot carry the conclusion.
The conclusion answers the original question in plain language, quotes the numbers that justify it, and says whether the hypothesis was supported or not supported. Communicating it — a poster, a lab report, a presentation — is treated as part of step seven in some lists and as a separate eighth step in others. Both are defensible. What matters is that the student can hand their method to someone else who could then repeat it and check.
Five steps, six steps, seven steps: why the lists disagree
If you search for the scientific method you will find lists of four, five, six, seven and eight steps, all presented as though they were the official version. There is no official version. The steps are a teaching scaffold that summarizes how experimental science tends to work, not a procedure anyone signed off on, and different curricula chop it at different points.
The five-step version usually runs: question, hypothesis, experiment, data, conclusion. It folds observation into the question and drops background research entirely. That is the version most elementary programs use, and it is the version Round 1 of this worksheet is built on.
The six-step version comes in two incompatible flavors. One adds observation back at the front. The other keeps five names but splits analysis from collecting data. Where this worksheet says six steps — in Round 2 — it means the first of those: observe, question, hypothesis, experiment, data, conclusion. That is the more common of the two variants, but say which one you mean before you grade an ordering question. The seven-step version keeps both observation and research. The eight-step version splits "conclusion" from "communicate results." Some lists add "revise the hypothesis and test again," which is arguably the most honest addition of all, since real investigations loop rather than run in a straight line.
It is worth telling students this plainly, because a student who has learned six steps at one school and seven at the next will otherwise assume one of their teachers was wrong. Neither was. The underlying logic is stable: notice something, ask something answerable, predict, test fairly, look at what you got, decide what it means, tell someone. The numbering is packaging.
There is a deeper disagreement worth mentioning to older students. Philosophers of science have argued for a long time that no single method describes what scientists actually do. Karl Popper, in Logik der Forschung — issued in Vienna in late 1934 but dated 1935 on the title page, so both years appear in citations, and translated as The Logic of Scientific Discovery in 1959 — argued that what makes a claim scientific is that it could be refuted by evidence, not that it was arrived at by a particular recipe. Paul Feyerabend went further in Against Method in 1975 and argued that no fixed set of rules covers the history of science at all. The classroom method is a useful training wheel, and it is fair to say so.
For grading purposes, pick one list and stick to it inside a single unit. Mixing a five-step warm-up with a seven-step assessment in the same week is where students lose points for reasons that have nothing to do with understanding.
A mnemonic that sticks, and one for variables
There is no single traditional mnemonic for the steps, which is why every classroom seems to have a different one. Three that work well are built from the first letters of the steps themselves, so students can reconstruct the list rather than recall a phrase they have to translate.
For the five-step version — Question, Hypothesis, Experiment, Data, Conclusion — use "Quiet Hens Eat Dry Corn." Five words, five steps, in order. It is concrete enough to picture, which is what makes a mnemonic survive a week.
For the common six-step version — Observe, Question, Hypothesis, Experiment, Data, Conclusion — use "Old Quiet Hens Eat Dry Corn." One word added at the front, and the rest of the phrase is unchanged, so a class that learned the five-step phrase does not have to learn a new one. If your six-step list instead drops observation and splits analysis from data collection, keep the five-step phrase and add "Analyze" between Data and Conclusion.
For the seven-step version — Observe, Question, Research, Hypothesis, Experiment, Data, Conclusion — use "Our Quiet Rabbits Have Eaten Dad's Carrots." Same trick, two extra words at the front and middle.
The more valuable mnemonic is the one for variables, and this one is genuinely standard in science classrooms: DRY MIX. DRY stands for Dependent, Responding, Y-axis. MIX stands for Manipulated, Independent, X-axis. It packs two things at once. First, it gives the alternative names — many textbooks call the independent variable the manipulated variable and the dependent variable the responding variable, so a student who only knows one pair of terms stalls on a worksheet that uses the other. Second, it settles which axis each goes on when the time comes to graph, which is the single most common formatting mistake in lab reports.
Say it out loud with the class the first time: the dependent variable responds, and it goes on the y-axis; the independent variable is the one you manipulate, and it goes on the x-axis. Then have them write DRY MIX in the margin of every graph they draw for two weeks. After that it tends to stay.
One caution. Mnemonics tell you the order and the names. They do not tell you what the variables are in a given scenario, which is the actual skill. That is why Rounds 2 and 3 below are scenarios rather than definitions.
Identifying variables worksheet: independent, dependent and controlled in plain English
Almost every point lost on an identifying variables worksheet comes from four ideas that sound similar and are not. Here they are in the order students meet them.
The independent variable is the one thing the investigator deliberately changes. There should be exactly one. It is independent in the sense that its value does not depend on the experiment — the investigator chose it before anything started. In Maya's tomato study it is the brand of fertilizer. In a ramp experiment it might be the mass of the car. It does not have to be a number: brand, color and species are all perfectly good independent variables, they just call for a bar chart rather than a line graph.
The dependent variable is the thing you measure to see what happened. Its value depends on the independent variable, which is where the name comes from. It should be something with a unit: height in centimeters, time in seconds, number of words recalled, distance in meters. If a student writes "how well the plant does" as their dependent variable, they have not finished the job; they need to say how they will measure doing well.
Controlled variables — also called constants — are everything else you deliberately hold the same. Same soil, same volume of water, same hours of light, same room, same measuring instrument. They are not the thing you are testing; they are the reason you can blame the result on the thing you are testing. If two factors change between groups, the experiment cannot tell you which one caused the difference, and the result is worthless no matter how carefully you measured it.
The control group is a different idea that unfortunately shares a word. It is a comparison group that gets no treatment, or gets the standard treatment, so you have a baseline. Maya's four unfertilized plants are her control group. Jonas's class memorizing in silence is his control group. A student who answers "the control is the soil" has confused a constant with a control group, and this is the single most common error on identifying variables worksheets. Worth saying explicitly: a controlled variable is a thing you keep the same, a control group is a set of subjects you leave alone.
Two more terms come up at high school level. A confounding variable is one that changes along with your independent variable, so its effect cannot be separated from the effect you are studying — if the fertilized plants also happen to sit closer to the window, sunlight is confounded with fertilizer. A lurking variable is one nobody thought about at all, which is what randomization is there to guard against: random assignment stops unknown factors lining up with the treatment. A larger sample does not help here. More data shrinks random variation only, so a confounded design just produces a more precise wrong answer. Textbooks also differ on these two words. Some use lurking and confounding interchangeably; others reserve lurking for a factor that was never measured and confounding for one that was. Check which convention your course uses before marking an answer wrong.
Finally, the operational definition. This is the sentence that turns a vague concept into something measurable: "plant health" becomes "stem height in centimeters plus a count of leaves longer than two centimeters, recorded every Friday at 9am." Insisting on an operational definition for the dependent variable fixes more student work than any other single intervention, because it forces the decision that everything downstream depends on.
Round 1 — Elementary: the five-step method
Twelve questions using the five-step version: question, hypothesis, experiment, data, conclusion. Observation is folded into the question step in this version, and it is treated that way here. The scenarios follow one student, Priya, through a plant investigation, then branch out. Aimed at roughly grades 3 to 5. No variable vocabulary is required; that starts in Round 2. Answers are short enough to write on one line.
Q1.Priya notices that the classroom plant on the windowsill has more leaves than the identical plant in the dim corner by the door. Has she made an observation or a hypothesis?Tap to reveal
Answer: An observation
An observation describes what you notice, using your senses or a tool, and stops there. Priya has not said why the plants are different, only that they are. The moment she adds a reason, she has moved on to a hypothesis. Observation is where an investigation starts: in the five-step list used in this round it is folded into step one, asking a question, and in the six- and seven-step lists it is a numbered step of its own.
Q2.Priya writes: "Does the amount of sunlight change how many green leaves a bean plant grows in four weeks?" What is this step called?Tap to reveal
Answer: Asking a testable question
A testable question names one thing you will change (sunlight) and one thing you will count (green leaves), and sets a time limit. "Which plant is the best?" is not testable, because nobody can measure "best."
Q3.Which of these is written as a hypothesis? (a) Plants like sunlight. (b) If a bean plant gets more sunlight, then it will grow more green leaves. (c) I will count the leaves on my plants every Friday.Tap to reveal
Answer: (b)
A hypothesis predicts a result and could turn out to be wrong. The if/then form connects the change to the measurement. Sentence (a) is an opinion with nothing to measure, and (c) is a plan for the experiment, not a prediction.
Q4.Priya grows six bean plants: three on the sunny windowsill and three on a shelf in the dimmest corner of the room. She gives them all the same soil and the same amount of water. Which step is this?Tap to reveal
Answer: The experiment (the test)
An experiment changes one thing on purpose and keeps everything else the same. Priya changed the light and kept the soil and water the same. Setting the test up is the experiment step; the measuring that comes next is counted separately, which is what the next question is about.
Q5.Priya counts the green leaves on every plant each Friday. After four weeks the windowsill plants averaged 11 green leaves and the dim-corner plants averaged 3. What do we call these numbers?Tap to reveal
Answer: Data (accept: results, measurements)
Data are the measurements and observations you record. Data on its own is not a conclusion — 11 and 3 are just numbers until someone says what they mean.
Q6.Priya writes: "Bean plants grown in sunlight grew more green leaves than bean plants grown in the dim corner, so my hypothesis was supported." Which step is this?Tap to reveal
Answer: The conclusion
A conclusion answers the original question, points at the data, and says whether the hypothesis was supported. Note the word "supported" rather than "proved": one experiment adds evidence, it does not settle a question forever.
Q7.Put these five steps in the correct order: experiment, conclusion, question, data, hypothesis.Tap to reveal
Answer: Question, hypothesis, experiment, data, conclusion
You need a question before you can predict an answer, a prediction before you know what to test, a test before you have numbers, and numbers before you can decide what they mean. Each step depends on the one before it.
Q8.Devin wants to know whether paper airplanes with wider wings fly farther. He throws each plane once, as hard as he can. Name one thing he should keep the same for every plane.Tap to reveal
Answer: Any one of: the same paper, the same throwing force or launcher, the same starting height, the same room with no draft
These are things you keep the same on purpose, which science class calls constants. Throwing "as hard as he can" is not the same force every time, so a rubber-band launcher or a set drop from a fixed height makes the test fairer. If two things change at once, you cannot tell which one changed the distance.
Q9.True or false: if the results do not match your hypothesis, the experiment failed.Tap to reveal
Answer: False
An unsupported hypothesis is a real result. It tells you something about the world that you did not know before, and scientists publish results like this all the time. An experiment only fails if it was set up so badly that it cannot answer the question.
Q10.What makes a test a fair test?Tap to reveal
Answer: Only one thing is changed, and everything else is kept the same
Only one thing is different between the groups, and everything else stays the same. If Priya's windowsill plants had also been given extra water, the extra leaves could have come from the water instead of the light, and her test could not tell the two apart.
Q11.Devin flies each plane once and records the distance. Why should he fly each plane at least three times instead?Tap to reveal
Answer: To check that the result was not a one-off caused by a bad throw or a draft
Repeating trials and taking an average smooths out random variation. One throw can go wrong for reasons that have nothing to do with wing width; three or five throws make the pattern easier to trust.
Q12.A class asks whether colder water dissolves sugar more slowly. Which tool measures what happens as a result: the stopwatch or the thermometer?Tap to reveal
Answer: The stopwatch
The class sets the water temperature themselves and checks it with the thermometer. What happens as a result is how long the sugar takes to dissolve, and that is timed in seconds with the stopwatch. The tool that measures the result is the one that answers the question.
Round 2 — Middle school: a scientific method scenarios worksheet
Seventeen questions on identifying variables, spotting confounds and choosing the right graph, plus one item on the step list itself. Where this round refers to the six-step method it means observe, question, hypothesis, experiment, data, conclusion. Each item is a short experiment scenario; students name the independent variable, the dependent variable, the constants or the control group, and find the design flaws. Aimed at roughly grades 6 to 8. Several questions reuse the same scenario, and each one names the study it belongs to so the items still work out of order.
Q1.Maya tests whether the brand of fertilizer affects tomato plant height. Every plant gets the same soil, the same amount of water and the same hours of sunlight. What is the independent variable?Tap to reveal
Answer: The brand of fertilizer (accept: the manipulated variable — the fertilizer brand)
The independent variable is the one factor Maya deliberately changes between groups. She chose the brands before the experiment started, so their values do not depend on anything that happens during it. Everything else in her setup is deliberately held steady.
Q2.In Maya's fertilizer-brand tomato study, what is the dependent variable, and what unit should it have?Tap to reveal
Answer: The height of the tomato plants, stated in a unit of length — centimeters or inches, as long as the same unit is used for every plant (accept: the responding variable)
The dependent variable is what she measures to see the effect, and its value depends on the fertilizer. Attaching a unit is part of the answer, and which unit matters far less than using one and using it consistently: "how the plants do" is not measurable, "height in centimeters" is.
Q3.Name two controlled variables in Maya's fertilizer-brand tomato study.Tap to reveal
Answer: Any two of: type and amount of soil, amount of water, hours of sunlight (also pot size, tomato variety, room temperature)
Controlled variables, or constants, are the factors kept identical across every group so that any difference in height can only be blamed on the fertilizer. They are the reason her comparison means anything.
Q4.Maya also grows four tomato plants with no fertilizer at all. What is that group called, and why is it not a "controlled variable"?Tap to reveal
Answer: It is the control group. A control group is a set of subjects that gets no treatment; a controlled variable is a factor held the same across all groups.
The two terms share a word and mean different things, which is the most common mistake on variables worksheets. The control group gives Maya a baseline: without it she could compare brands against each other but could not say whether any brand beats no fertilizer at all.
Q5.Jonas assigns three classes to different conditions. Each class memorizes the same 20-word list for five minutes while music plays at 60, 100 or 140 beats per minute, and he then counts how many words each student recalls. What is the independent variable?Tap to reveal
Answer: The tempo of the music, in beats per minute
Jonas set the three tempos in advance and assigned them to classes. Note that an independent variable can have more than two levels — three tempos is still one variable, not three.
Q6.In Jonas's music-tempo memory study, what is the dependent variable?Tap to reveal
Answer: The number of words correctly recalled (accept: recall score, words remembered — the responding variable)
It is a count, it has a clear unit, and it responds to the tempo. On a graph it goes on the y-axis, because DRY MIX: Dependent, Responding, Y-axis.
Q7.In Jonas's music-tempo memory study, what should the control group do?Tap to reveal
Answer: A fourth class should memorize the same word list for the same five minutes in silence
A control group receives no treatment, so a silent class is the baseline. Without it, Jonas can only say which tempo beat which other tempo. He could not tell whether any music at all helps or hurts compared with no music.
Q8.Suppose Jonas had given each class a different 20-word list. What would that do to his results?Tap to reveal
Answer: It would confound them — a difference in recall could be caused by the tempo or by one list being easier than another
A confounding variable changes at the same time as the independent variable, so their effects cannot be separated. The fix is to use the same list for every class, or to rotate the lists so each tempo is tested with each list.
Q9.A class rolls a toy car down a ramp, taping coins to the car to increase its mass, and measures how far past the ramp the car travels. Name the independent and dependent variables.Tap to reveal
Answer: Independent: the mass of the car. Dependent: the distance it travels past the ramp.
The coins are added on purpose, so mass is the manipulated variable. Distance is what gets measured in response. Ramp angle, release point and floor surface are constants.
Q10.In the ramp-and-coins car experiment, the class raises the ramp higher for the heavier cars "so they get a fair start." What is wrong with this?Tap to reveal
Answer: They have changed two things at once, so the results cannot show what mass alone does
Ramp height sets how fast the car is moving when it reaches the floor, whatever its mass. Raising the ramp for the heavier cars means any change in distance could come from the extra height or from the extra mass, and the experiment cannot separate them. Fairness in an experiment means holding everything else identical, not compensating for the thing you are testing.
Q11.When the class graphs the ramp-and-coins car experiment, which variable goes on the x-axis?Tap to reveal
Answer: The independent variable, mass, goes on the x-axis; distance goes on the y-axis
DRY MIX: Dependent, Responding, Y-axis; Manipulated, Independent, X-axis. The convention exists so that any reader can look at a graph and immediately tell which factor was being changed.
Q12.The class runs five trials at each mass and uses the average. Why is averaging better than a single trial?Tap to reveal
Answer: It reduces the effect of random variation, so one unusual run cannot decide the result
Random errors push measurements up on some trials and down on others. Averaging lets them partly cancel out. Averaging does not fix a systematic problem, such as a ramp that always releases the car crookedly.
Q13.Petra runs a two-week windowsill study comparing tap water with fertilizer solution on identical seedlings, all on the same sill. A heat wave arrives in week two and she cannot control the room temperature. What should she do, and does the heat wave invalidate her comparison?Tap to reveal
Answer: Record the temperature, report it as a limitation, and continue — because the heat affected both groups equally it is not confounded with the treatment, though it may have changed the size of the effect
Some variables cannot be controlled in a school setting. A factor that hits every group alike cannot explain a difference between them, so it is a limitation rather than a confound; a factor that hits one group only is a confound. The honest response is to measure what you cannot control, note when it changed, and say in the conclusion that the result may have been affected. Hiding an uncontrolled factor is worse than admitting it.
Q14.Ana tests three detergent brands on identical cotton squares, each stained with the same amount of grass stain and washed for ten minutes at 30°C. Why is the amount of stain a constant rather than the independent variable?Tap to reveal
Answer: Because it is the same for every square; only the brand changes between groups
What makes a factor the independent variable is not what it is, but whether the investigator varies it. Stain amount could be an independent variable in a different experiment — one testing whether heavier stains are harder to remove — but here it is held steady so brand is the only thing that differs.
Q15.In Ana's detergent-brand test the independent variable is detergent brand. Does an independent variable have to be a number, and how does that change the graph?Tap to reveal
Answer: No. Brand is a categorical variable, so the results are shown as a bar chart rather than a line graph.
Line graphs imply that the values between the plotted points mean something, which is true for temperature or mass but meaningless between Brand A and Brand B. Categorical independent variables get bars; numerical ones that vary continuously get lines.
Q16.A gardener tests whether soaking bean seeds before planting changes how quickly they sprout. He soaks half the seeds for twelve hours and plants the rest dry, then records how many days each seed takes to sprout. Name the independent variable, the dependent variable and the control group.Tap to reveal
Answer: Independent: whether the seeds were soaked. Dependent: the number of days until sprouting. Control group: the seeds planted dry.
Soaking is the one thing the gardener changes, days-to-sprout is what he measures in response, and the dry seeds are the untreated comparison. This scenario is worth studying because the control group and one level of the independent variable are the same set of seeds — that happens whenever the treatment is simply present or absent.
Q17.Put the six steps used in this round in order: data, experiment, observation, conclusion, question, hypothesis.Tap to reveal
Answer: Observation, question, hypothesis, experiment, data, conclusion
This is the five-step list with observation restored at the front, which is the more common of the two six-step versions. The other six-step version keeps observation folded into the question and instead splits analyzing the data from collecting it. Both are in use, so a class should be told which one it is being graded against.
Round 3 — High school: controls, error and experimental design
Fifteen questions on the parts of experimental design that separate a school project from a defensible one: positive and negative controls, blinding, randomization, sample size, error types, and the difference between a hypothesis and a prediction. Aimed at roughly grades 9 to 12 or an introductory college lab course. Two questions use real historical experiments, because they are the cleanest illustrations there are.
Q1.A student measures how fast catalase breaks down hydrogen peroxide at five temperatures. She also sets up one tube with catalase that was boiled for ten minutes before use. What is that tube, and what does it show?Tap to reveal
Answer: A negative control (accept: a control). It shows that the bubbling in the other tubes comes from active enzyme, not from something else in the mixture.
Boiling denatures the enzyme, so this tube contains everything the others contain except a working catalase. If it bubbles anyway, the experiment has a problem — perhaps the peroxide is decomposing on its own or the glassware is contaminated. A negative control is the tube that should show nothing.
Q2.In an iodine test for starch, a student includes a tube of a solution known to contain starch. What is this called, and what problem does it catch?Tap to reveal
Answer: A positive control. It catches a failed or expired reagent.
If the known-starch tube does not turn blue-black, the iodine is not working, and every negative result in the experiment is meaningless. Positive controls confirm that the method can detect the thing it is looking for. Negative controls confirm it does not detect things that are not there.
Q3.In a drug trial, one group receives a tablet with no active ingredient. What is that tablet called, and what effect is it there to separate out?Tap to reveal
Answer: A placebo. It separates the effect of the drug from the effect of believing you are being treated.
People who think they are being treated often report improvement, and some measurable outcomes shift too. Without a placebo group, any improvement in the treated group could be the drug, the expectation, or simply time passing.
Q4.In a trial of a new asthma inhaler, the patients do not know whether they have been given the active inhaler or a dummy, and neither does the nurse who hands them out and records how the patients are doing. What is this design called, and what does hiding the assignment from the nurse prevent?Tap to reveal
Answer: Double-blind. It prevents the person delivering the treatment and recording the outcome from encouraging, measuring or interpreting the two groups differently.
Single-blind hides the assignment from participants only. Double-blind also hides it from staff, because a researcher who knows which group a subject is in can bias the result without intending to — a longer look at a borderline reading, a more optimistic note in the record. The allocation is held by someone not involved in data collection and opened once the data is in.
Q5.Instead of letting students pick which study method to use, a researcher assigns them by drawing names. Why does random assignment matter?Tap to reveal
Answer: It spreads unknown differences between people evenly across the groups instead of letting them cluster in one
If students choose, the motivated ones may all pick the same method, and motivation rather than method explains the result. Randomization does not remove confounding factors, but it stops them lining up systematically with the treatment.
Q6.Two studies find the same effect. One tested 3 subjects, the other tested 300. Why does the second carry more weight?Tap to reveal
Answer: Larger samples make it far less likely that the result came from chance variation between individuals
With three subjects, one unusually fast learner can move the average enough to create an effect that is not there. Sample size does not fix a biased design — 300 badly chosen subjects are still badly chosen — but for the same design, more data means a more reliable estimate.
Q7.A student writes that her dependent variable is "how healthy the plants are." Rewrite this as an operational definition.Tap to reveal
Answer: Something measurable, such as: stem height in centimeters plus the number of leaves longer than 2 cm, recorded every Friday at 9am
An operational definition states exactly what will be measured, with what instrument, in what units, and when. It is what allows another person to repeat the experiment and get comparable numbers. Vague dependent variables are the most common reason school data cannot be analyzed.
Q8.A student writes: "Light intensity limits the rate of photosynthesis in this pondweed." Her lab partner writes: "If the lamp is moved from 40 cm to 20 cm, then the bubble count per minute will increase." Which sentence is the hypothesis and which is the prediction?Tap to reveal
Answer: The first is the hypothesis; the second is the prediction
A hypothesis is a proposed explanation — it names a mechanism. A prediction is the specific observable result you expect at the bench if that explanation is correct. Many school worksheets merge the two, which is why a single if/then sentence often carries both. One design note on this particular setup: moving a lamp closer also warms the water, so a real version needs a heat shield or a water bath between lamp and beaker, or temperature is confounded with light intensity. Leave the prediction directional rather than putting a multiplier on it, because photosynthetic rate saturates as light rises.
Q9.State the null hypothesis for an experiment testing whether fertilizer brand affects tomato height.Tap to reveal
Answer: Fertilizer brand has no effect on tomato plant height — any difference between the groups is due to chance
The null hypothesis is the position of no effect. Statistical tests work by asking how unlikely the observed data would be if the null were true. You never prove the null; you either reject it or fail to reject it.
Q10.Which of these is falsifiable? (a) Tomato plants spoken to for ten minutes a day grow taller over six weeks than tomato plants left in silence. (b) There is an invisible force that helps plants but leaves no measurable trace.Tap to reveal
Answer: (a)
Statement (a) forbids a specific outcome. If the spoken-to plants are no taller after six weeks, the claim is refuted, and that is what makes it testable. Statement (b) is built so that no observation could ever count against it, which places it outside science. Karl Popper made falsifiability the center of his account of science in Logik der Forschung, published in German in 1934, dated 1935 on the title page, and translated as The Logic of Scientific Discovery in 1959. Note that a claim of the form "some plants respond" would not work here: no finite set of observations can refute an existence claim, which is why the statement has to be written as a general one.
Q11.A balance reads 2 g too high on every single measurement. Is this random error or systematic error, and will repeating trials fix it?Tap to reveal
Answer: Systematic error. Repeating will not fix it.
Systematic error shifts every reading in the same direction by a similar amount, so averaging more readings just gives a more precise wrong answer. It is fixed by calibrating or taring the instrument. Random error scatters readings both ways and does shrink with repeats.
Q12.A school tests a new reading program on the ninth grade and compares scores with the eighth grade, who used the old program. Name one confounding variable in this comparison, and say why it cannot be separated from the effect of the program.Tap to reveal
Answer: Any one of: age or grade level, different teachers, a different prior curriculum, different testing conditions — the two groups differ in more than the program, so no score difference can be pinned on it
There is no single right answer here, and that is the point: a comparison between two whole year groups is confounded several times over. Ninth graders are a year further along, taught by different staff, and were taught something different last year. Comparison groups should differ only in the treatment, so splitting one grade at random into two groups would fix it.
Q13.A district notices that on days when more students wear coats, more students are absent. Why does this not show that coats cause absence?Tap to reveal
Answer: Both rise with cold weather, which is a third factor driving each of them independently
Correlation shows that two things move together. Causation requires ruling out other explanations, usually by controlling or randomizing them. Here the shared cause is temperature: cold days put coats on students and also spread the illnesses that keep them home. Whenever two things rise together, the first question is what third thing could be pushing both.
Q14.In work published in 1668, Francesco Redi compared meat left in open jars with meat in sealed jars. When critics objected that the sealed meat had simply been deprived of air, he ran the comparison again with jars covered in fine gauze. What did the gauze jars settle?Tap to reveal
Answer: They let air reach the meat while keeping flies out, so the absence of maggots could no longer be blamed on a lack of air
From the critics' point of view the sealed jar changed two things at once: access for flies and access for air. The gauze version changed only one of them. Maggots appeared on the gauze itself, not on the meat beneath it. Both experiments appear in Esperienze Intorno alla Generazione degl'Insetti in 1668, and the sequence is the lesson: the second experiment exists because someone raised an alternative explanation for the first.
Q15.In 1747 James Lind gave six different treatments to six pairs of sailors with scurvy aboard the same ship, keeping their diet otherwise identical. Which feature of his design made the comparison meaningful?Tap to reveal
Answer: Every pair shared the same diet, conditions and stage of illness, so the treatment was the only difference between them
Lind held the setting constant and varied only the remedy. After six days — the point at which the fruit ran out — one of the pair given oranges and lemons was fit for duty and the other had improved enough to nurse the rest; no other pair came close. The trial was stopped by supply rather than by design, which is a limitation Lind reported himself and a useful lesson about trial length. He published the work in his Treatise of the Scurvy in 1753, and it is routinely cited as one of the earliest controlled clinical trials.
Round 4 — Hypotheses, conclusions and reporting
Twelve questions on the writing end of the method, most of them built on a short piece of student work you have to fix: a vague claim to turn into a testable if/then hypothesis, a conclusion that says too little, a result that does not cooperate, a table with a problem in it. These items work at middle or high school level and pair well with a lab report assignment.
Q1.Rewrite this as a testable if/then hypothesis: "Sugar affects heart rate."Tap to reveal
Answer: Something like: If a person drinks 200 ml of a sugary drink, then their resting heart rate will increase within 30 minutes compared with drinking 200 ml of water.
The rewrite names the change, the measurement, the timeframe and the comparison. The original sentence cannot be tested as written because it does not say how much sugar, measured in whom, over what period, or compared with what.
Q2.Which of these cannot be tested by experiment? (a) Classical music helps students concentrate. (b) Classical music is the most beautiful music. (c) Students recall more words in silence than with music.Tap to reveal
Answer: (b)
Beauty is a value judgement with no agreed measurement, so no result could show it false. Statements (a) and (c) both name something you could measure and could fail to find.
Q3.A student's conclusion reads: "This experiment proved that Brand B fertilizer makes tomatoes grow taller." Which word should be changed, and why?Tap to reveal
Answer: "Proved" should be "supported" — one experiment adds evidence but cannot rule out every alternative explanation or every future contradicting result
Proof belongs to mathematics, where conclusions follow from axioms. Experimental science accumulates support, and any well-established finding remains open to revision if better evidence arrives. Teaching this early prevents the mistaken idea that science deals in certainties, and it costs one word to fix.
Q4.A student's whole conclusion reads: "The fertilizer worked best." Name two things missing.Tap to reveal
Answer: Any two of: which fertilizer performed best, the numbers from the data that show it, and whether the hypothesis was supported
A conclusion must be checkable by a reader who has only the report. "Brand B plants averaged 31 cm against 24 cm for the unfertilized control, supporting the hypothesis" gives the reader the evidence and lets them disagree with the interpretation if they want to.
Q5.In five trials, a student records 4.2 s, 4.4 s, 4.1 s, 9.8 s and 4.3 s. What should she do with the 9.8 s reading?Tap to reveal
Answer: Keep it in the record, flag it as an anomaly, investigate the likely cause, and say in the report if it was excluded from the average and why
Deleting inconvenient data without saying so is falsification. Identifying an anomaly, looking for a reason such as a mistimed stopwatch, and stating clearly what you did with it is honest practice and is usually rewarded in grading.
Q6.Plants averaged 20 cm without fertilizer and 25 cm with it. What is the percentage increase?Tap to reveal
Answer: 25 percent
The increase is 5 cm. Divided by the original 20 cm that is 0.25, or 25 percent. Percentage change is always measured against the starting value, not the final one — dividing 5 by 25 would give 20 percent and would be the wrong comparison.
Q7.An experiment compares four detergent brands. Bar chart or line graph?Tap to reveal
Answer: Bar chart
The independent variable is categorical. A line graph would draw a line between Brand A and Brand B, implying that points in between exist and mean something. Line graphs are for independent variables that vary continuously, such as temperature, time or mass.
Q8.What does the spread between the highest and lowest of a set of repeat readings tell you?Tap to reveal
Answer: How consistent the measurement is — a wide spread means the value is less reliable and small differences between groups mean less
This spread, the range, is the simplest measure of variability. If the fertilized group averages 25 cm with a range of 22 to 28, and the control averages 24 cm with a range of 19 to 30, the two overlap heavily and the 1 cm difference in averages is not convincing.
Q9.Before a pondweed study is published, two researchers who did not run it read the method and report that the light meter readings are missing. What is this process called, and what is it checking?Tap to reveal
Answer: Peer review. It checks whether the method actually supports the conclusions.
Reviewers with relevant expertise look for whether the controls were present, whether the measurements reported are enough to judge the claim, whether the analysis fits the data, and whether the conclusions go further than the evidence allows. It is a filter, not a guarantee — reviewed work is sometimes later found to be wrong — but it removes a large amount of weak work before it reaches print.
Q10.Why does it matter that another lab can repeat an experiment and get the same result?Tap to reveal
Answer: Because a result that only appears in one lab may come from a local quirk, a mistake or chance rather than from the effect being studied
Replication is the check that turns a single finding into established knowledge. It is also why methods sections must be detailed enough for a stranger to follow — a method nobody can repeat cannot be confirmed or refuted.
Q11.Petra's plant study ran for two weeks through a heat wave, used four plants per group, and measured height with a 30 cm ruler marked in millimeters. Name three things that belong in her limitations section.Tap to reveal
Answer: The uncontrolled room temperature, the small number of plants per group, and the short timeframe — the ruler's precision is worth a line too if she reported heights to the nearest millimeter
Limitations are not an apology, they are part of the evidence. Anything that could have affected the result and was not controlled belongs here: sample size, uncontrolled conditions, instrument precision, a timeframe cut short. They tell a reader how much weight the conclusion can bear, and they usually point directly at what the next experiment should fix.
Q12.A data table has columns headed "Trial," "Mass," and "Distance." What is missing from the column headings?Tap to reveal
Answer: The units — the headings should read Mass (g) and Distance (cm)
Without units the numbers are ambiguous and cannot be compared with anyone else's. The convention is to put the unit once in the column heading rather than after every value, which keeps the table readable.
How to write a hypothesis students can actually test
More class time is lost to bad hypotheses than to anything else in the method, and it is nearly always the same three faults. Here is what to look for and how to fix each one on the spot.
The first fault is the untestable claim. "Music helps you study" has no measurement in it. Ask a single question: what would you count? If the student cannot answer, they do not yet have a hypothesis. Push them to a number — words recalled, problems completed in ten minutes, score out of twenty. The moment a countable thing appears, the sentence usually rewrites itself.
The second fault is the missing comparison. "If I use fertilizer, the plants will grow" is not a prediction, because the plants would grow anyway. A hypothesis needs a comparison built in: taller than what? Faster than what? Compared with the control group, compared with the untreated squares, compared with the same task in silence. Teaching the phrase "compared with" as a required part of the sentence removes this problem almost entirely.
The third fault is more than one variable in the same sentence. "If plants get more light and more water, then they will grow taller" describes an experiment that cannot answer either question. Split it. Two hypotheses, two experiments, or one experiment that varies light while holding water constant. Students often resist this because varying two things feels more efficient. The counter-argument is concrete: if the plants grow taller, you will not know what to tell the next gardener.
A fourth fault is quieter and shows up mostly in biology: measuring something that does not track the mechanism you named. A student who reasons that light drives photosynthesis and then measures stem height has picked a measurement that can move the wrong way, because seedlings in the dark stretch upward on their stored reserves before they die. Ask students to say why the thing they are measuring should respond to the thing they are changing. If they cannot, the measurement is the problem, not the hypothesis.
A working template that survives all of these: "If [the one thing I change], then [the thing I measure] will [increase, decrease, stay the same] compared with [the comparison group], because [the reason I expect this]." The final clause is optional at elementary level and should be compulsory by high school, because it is what distinguishes a reasoned prediction from a guess.
One more habit worth building. Ask students to say, before they start, what result would show their hypothesis is wrong. If they cannot describe such a result, the hypothesis is not falsifiable and needs rewriting. This takes thirty seconds per student and catches problems that would otherwise surface three weeks later when the data will not analyze.
Running the three tiers in one classroom
The four rounds above are graded, which lets one worksheet cover a mixed group. Round 1 is a five-step elementary set with no variable vocabulary in it. Round 2 adds variables and confounds at middle school level. Round 3 covers controls, blinding and error for high school. Round 4 is about writing up, and works for either of the upper two bands.
For a single-grade class, the simplest approach is to print the round that matches and use the round below it as a warm-up. Ten minutes on Round 1 before starting Round 2 costs almost nothing and reveals which students still confuse observation with hypothesis. That confusion is invisible on a variables worksheet, but it explains a surprising amount of later trouble.
For a mixed or differentiated group, print Rounds 1 and 2 as one sheet and Rounds 2 and 3 as another. Round 2 overlaps both, so the class is doing shared work in the middle and everyone can contribute to the same discussion afterwards. Nobody is holding a visibly easier sheet, which matters more in some rooms than others.
As a diagnostic at the start of a unit, use five items: question 3 from Round 1 (recognizing a hypothesis), questions 1 and 4 from Round 2 (independent variable, control group versus constant), question 7 from Round 3 (operational definition) and question 1 from Round 4 (rewriting an if/then). Those five predict how the rest of the unit will go better than a twenty-item quiz does, and they take six minutes.
As a lab-report scaffold, hand out Round 4 alongside the students' own experiment write-up. Each question maps to a section of the report: the hypothesis item to their hypothesis, the anomaly item to their results, the limitations item to their discussion. Students who have just answered a question about what belongs in a limitations section write a better limitations section ten minutes later.
For a substitute lesson, the scenarios are self-contained. Nothing requires equipment, a lab, or prior teaching from that week, and the answer key explains each item well enough that a non-specialist can lead the review. That is the situation this worksheet was designed to survive.
And as a homework set, split by round across a week: Monday Round 1, the five-step scenarios; Wednesday Round 2, the variables scenarios; Thursday Round 3, the design questions; Friday Round 4, the write-up questions. Short daily sets retain better than one long sheet, and the recurring scenarios — Maya's tomatoes, Jonas's music, the ramp — mean students are not learning a new setup every night.
Grading guide: what to accept and what to push back on
The answer key gives one correct answer per item, and prints the common alternatives alongside it, but student wording still varies and some variations deserve full credit. Here is where to be generous and where to hold the line.
Accept alternative names for the variables. Manipulated variable is the independent variable. Responding variable is the dependent variable. Constants and controlled variables are the same thing. Textbooks split on which pair they use, and a student who has learned one set should not lose points to a worksheet that uses the other.
Accept a dependent variable stated without a unit at elementary level, and require the unit from middle school upward. "Height" is fine in Round 1. By Round 2 it should be "height in centimeters" — or inches, as long as every plant in the study is measured the same way. The point of the middle-school tier is that measurement gets specific, not that one country's units win.
Accept any reasonable constant when the question asks for one. In the paper airplane item, students name paper type, throwing force, launch height, room, and even the person throwing. All are correct. The item is testing whether they understand what a constant is, not whether they guessed the one in the key.
Push back when a student names a control group as a controlled variable, or the reverse. This is the error worth spending time on, because it is conceptual rather than verbal. The quick diagnostic is to ask: is this a thing you keep the same, or a group of subjects you leave alone? Constants are things, control groups are subjects.
Push back when the dependent and independent variables are swapped. The test is direction of dependence: does fertilizer brand depend on plant height, or does plant height depend on fertilizer brand? Students who get this wrong usually chose whichever variable was mentioned first in the scenario, so vary the order in your own examples.
Push back on hypotheses that contain no prediction. "I think the fertilizer will do something" and "my hypothesis is that I will measure the plants" both fail, the first for having no direction and the second for describing the method. Ask for the missing half rather than marking it wrong outright — most students can supply it once they see what is absent.
Hold the line on "proved." A conclusion that says the experiment proved the hypothesis should be corrected every time, at every level. It is a one-word fix and the underlying idea — that evidence supports rather than settles — is one of the few things from a school science course that stays useful for life.
Finally, give credit for a well-argued wrong answer on the design questions in Round 3. If a student names a different confounding variable than the key does, and their reasoning holds, they have demonstrated exactly the skill the item tests. The key is a reference, not a ceiling.
Where students go wrong, and what fixes it
Six failure modes account for most of the points lost on scientific method work. Each has a specific fix that takes less than a lesson.
Confusing the control group with controlled variables. Fix: teach the two terms in the same breath, deliberately, and never introduce one without the other. Say "a controlled variable is a thing you keep the same; a control group is a set of subjects you leave alone" and have the class write both down. The error persists mainly because the terms are usually taught weeks apart.
Choosing a dependent variable that cannot be measured. Fix: insist on the operational definition before any data is collected. If the student cannot say what instrument they will use and what unit will come out of it, the experiment is not ready to run. This single requirement prevents most of the projects that collapse at the analysis stage.
Changing two things at once. Fix: have students write the sentence "the only thing different between my groups is ___" and fill in the blank. If more than one thing goes in the blank, the design needs splitting. Students who resist this are usually trying to save time, and the honest answer is that they are trading time for a result nobody can interpret.
Treating an unsupported hypothesis as failure. Fix: say out loud, early and more than once, that a hypothesis which turns out to be wrong is a successful experiment. Then grade accordingly, so students learn from the grade rather than from the speech. A report with clean method, honest data and a conclusion of "not supported" should score higher than one with a supported hypothesis and sloppy controls.
Running one trial. Fix: make repeats a structural part of the data table rather than an instruction. If the table you hand out has columns for Trial 1, Trial 2, Trial 3 and Mean, students fill them in. If it has one results column, they run one trial. The format does the teaching.
Writing a conclusion that restates the hypothesis instead of the result. Fix: require the conclusion to contain at least two numbers from the student's own data. This is a mechanical rule and it works, because a conclusion that cites the actual figures cannot help but answer the question. It also makes it obvious, to the student and to you, when the data does not support what they wanted to say.
One pattern underlies all six: students treat the scientific method as a form to fill in rather than as a set of decisions with consequences. The scenario format in this worksheet exists to push against that, because a scenario forces a judgement — you cannot identify Maya's control group by recalling a definition, only by looking at what she actually set up.
Where the method came from
The scientific method is often presented as though it arrived complete, but it was assembled over centuries, and knowing a few of the landmarks makes the steps less arbitrary to students.
Ibn al-Haytham, working in Cairo in the early eleventh century, wrote the Book of Optics, in which he tested claims about how vision works using darkened chambers and controlled light sources rather than accepting the reasoning of earlier authorities. His insistence that a claim about nature be checked against observation, and that the observer distrust their own preconceptions, is one of the earliest clear statements of the idea.
Francis Bacon published Novum Organum in 1620, arguing that knowledge should be built up from systematic observation and experiment rather than deduced from first principles. He was more interested in the logic of gathering evidence than in any particular experiment, and much of what school textbooks call the method traces back to this program.
Francesco Redi's jars of meat, published in 1668, gave the method its first famous control. Sealed jars produced no maggots and open jars did — and when critics objected that the sealed meat had simply been deprived of air, he repeated the test with gauze-covered jars, which admitted air but not flies. Maggots appeared on the gauze rather than on the meat. The gauze jar is the whole idea of a control in one image, and the fact that it was built to answer an objection is the part worth telling students.
James Lind's shipboard scurvy trial in 1747 applied the same thinking to people. Six pairs of sick sailors, an otherwise identical diet and daily routine, six different remedies. After six days — the point at which the fruit ran out — one of the pair given oranges and lemons was fit for duty and the other had improved enough to nurse the rest. No other pair came close. Lind published the work in 1753, and it is routinely cited as one of the earliest controlled clinical trials, even though it took decades for the finding to change naval practice.
Claude Bernard's Introduction to the Study of Experimental Medicine, published in 1865, set out the reasoning behind comparison groups in physiology and argued that the experimenter must be willing to have their hypothesis destroyed by the result. Around 1860, Louis Pasteur's swan-necked flasks showed that broth stayed sterile while exposed to air, provided dust could not reach it — another design where the control does the persuading.
In the twentieth century the emphasis shifted from how to gather evidence to what makes a claim scientific at all. Karl Popper argued that the mark of a scientific statement is that it forbids something — that it could be shown false. Others, including Paul Feyerabend in Against Method in 1975, argued that no single procedure describes the history of science accurately, and that the tidy numbered list is a reconstruction after the fact.
Both things can be true at once, and it is worth telling students so. The numbered method is a reliable way to run a fair test and to write it up so someone else can check it. It is not a description of how every discovery has ever been made. Fleming's 1928 observation of a contaminated culture plate did not begin with a hypothesis. It began with noticing something odd — which is, in the end, step one.
Frequently Asked Questions
What are the 7 steps of the scientific method?
The seven-step version is: observe something, ask a testable question, do background research, form a hypothesis, test it with an experiment, record and analyze the data, then draw a conclusion and communicate it. Some lists split the last step in two, making eight. Others drop observation and research, making five. There is no official version — the numbering is a teaching scaffold, and different curricula cut it at different points.
What are the 5 steps of the scientific method?
Question, hypothesis, experiment, data, conclusion. This is the version most elementary programs use. It folds observation into the question step and leaves out background research. Round 1 of this worksheet is built on the five-step version, so it matches what most third to fifth grade classes are taught. Round 2 uses the six-step version, which is the same list with observation restored at the front.
What is a good mnemonic for the scientific method?
A mnemonic for the scientific method has to match the step list your class uses. For the five steps — Question, Hypothesis, Experiment, Data, Conclusion — use "Quiet Hens Eat Dry Corn." For the six-step list that puts observation back at the front, use "Old Quiet Hens Eat Dry Corn." For the seven steps — Observe, Question, Research, Hypothesis, Experiment, Data, Conclusion — use "Our Quiet Rabbits Have Eaten Dad's Carrots." For variables, use the standard classroom mnemonic DRY MIX: Dependent, Responding, Y-axis; Manipulated, Independent, X-axis. That one also gives you the alternative textbook names for each variable.
Is there a PDF version of this scientific method worksheet?
Yes, and you make it yourself in one step. The blank sheet and the answer key print directly from this page, and your browser's print dialog will also save either of them as a PDF if you want a file to keep or to send to students. That gives you a scientific method worksheet PDF with no email address, no account and no download gate. The two documents are separate, so students never see the key.
Is this scientific method worksheet free to print, and is there an answer key?
Yes to both, with no email address required and no account to create. The printable version comes as two documents: a blank student sheet and a separate answer key. The key includes the explanation under every answer, not just the answer, plus the alternative wordings that should be given credit, so it can be handed to a student or a substitute teacher directly.
What is the difference between a control and a controlled variable?
A controlled variable, also called a constant, is a factor you keep the same across every group — the same soil, the same volume of water, the same room temperature. A control group is a set of subjects that receives no treatment, so you have a baseline to compare against. Constants are things you hold steady; a control group is subjects you leave alone. Mixing these up is the most common error on identifying variables worksheets.
How do you identify the independent and dependent variable in a scenario?
Ask two questions. What did the investigator deliberately change? That is the independent variable. What did they measure to see the effect? That is the dependent variable. A quick check on direction: plant height depends on fertilizer brand, but fertilizer brand does not depend on plant height, so height is the dependent one. If the answer to "what did they measure" has no unit attached, the scenario is not ready to run yet.
What grade level is this worksheet for?
All three bands, on one page. Round 1 is elementary, using the five-step method and no variable vocabulary, roughly grades 3 to 5. Round 2 is middle school, covering variables and confounds, roughly grades 6 to 8. Round 3 is high school, covering controls, blinding, error types and design. Round 4 covers writing up hypotheses and conclusions and suits either upper band. You can print a single round or combine two for a mixed group.
How do you write a hypothesis in if/then form?
Use this template: "If [the one thing I change], then [the thing I measure] will [increase, decrease or stay the same] compared with [the comparison group], because [the reason I expect this]." The comparison clause is the part students most often leave out, and it is what stops a prediction like "the plants will grow" — which would be true regardless. From middle school upward, require the "because" clause too, since it is what separates a reasoned hypothesis from a guess, and check that the thing being measured actually responds to the mechanism named in it.
