What does a scientist actually do?
By the end of this lesson you will be able to
- Describe the six steps of the scientific method and put them in order.
- Distinguish an observation from an opinion, an inference and an explanation.
- Formulate a hypothesis in the form If… then… because… and explain why the “because” matters.
- State what makes a question testable, and rewrite an untestable question as a testable one.
- Explain why scientists repeat measurements and why results must be shared and checked.
The puzzle
Two glasses of water. In one, a spoon of sugar disappears in seconds. In the other, the same spoon of sugar sits on the bottom for a minute. Nobody stirred either glass. What is different?
You probably have a guess already. Good — a guess is where science starts. But a guess is not where it ends. By the end of this lesson you will know the difference between a guess, a hypothesis and evidence, and you will have done your first proper experiment to find out.
Before reading on, write down
(a) your guess about what is different, and
(b) one thing you would need to measure to find out whether your guess is right.
Compare your answer
Most students guess temperature. Other reasonable guesses: the sugar grains are a different size; one glass has more water; one glass is being warmed by the sun. Any of these could be true — that is exactly why a guess is not enough.
To find out, you must measure something: the time for the sugar to disappear, with a stopwatch, while you change only the thing you guessed. If you wrote “look at it” or “see which is faster”, ask yourself: faster by how much? Science runs on numbers, because numbers can be compared and checked by someone else.
Try it — the kitchen experiment
You need: two identical glasses, tap water, hot water from the tap (not boiling — never handle boiling water for this), sugar, a teaspoon, a clock or phone timer.
| Water | Trial 1 / s | Trial 2 / s | Trial 3 / s | Mean / s |
|---|---|---|---|---|
| Cold | ||||
| Hot |
Notice something about the table: the unit (s, for seconds) is written once, in the heading — never inside the boxes. That is how scientists everywhere write tables, and it is how the exam expects yours.
If you cannot do this at home right now, no problem — use the figures below from a Year 1 class in Nairobi and carry on. Cold: 118 s, 131 s, 124 s. Hot: 21 s, 19 s, 24 s.
The science — structured notes
1.1 Science is a method, not a list of facts
Science is the most successful method humans have ever invented for finding out what is true about the world. It is not a collection of facts to memorise; the facts are the results of the method. Every scientist, from a Year 1 student with two glasses of sugar water to a team operating a telescope in orbit, works round the same loop:
These six statements describe one investigation but are jumbled. Put them in the order of the loop.
A. She concluded that the hot water dissolved the sugar six times faster.
B. She noticed sugar vanishing faster in her tea than in her cold juice.
C. She asked: does water temperature change how fast sugar dissolves?
D. She timed sugar dissolving in cold and hot water, three times each.
E. She predicted hot water would be faster because its particles move faster.
F. She showed her table to her class, who repeated it and got the same result.
Reveal answer
B → C → E → D → A → F. Observe (B), ask (C), hypothesise (E), test (D), conclude (A), share and check (F). If you put E before C, notice that you cannot make a prediction until you have a question to predict about.
1.2 Observation, inference, opinion, explanation
Four kinds of statement look similar and are treated completely differently in science. Getting them apart is the first skill of the course.
| Kind | What it is | Example | Can it be checked by measuring? |
|---|---|---|---|
| Observation | What you detect with senses or instruments, recorded as it is | “The sugar in the hot water vanished in 21 s.” | Yes — anyone can repeat it |
| Inference | A conclusion drawn from observations | “So hot water dissolves sugar faster.” | Yes, if it is based on enough observations |
| Explanation | The reason why — the mechanism | “Because faster water particles knock the sugar apart more often.” | Only indirectly — by testing predictions it makes |
| Opinion | A preference or judgement | “Hot water is better.” | No — “better” is not a measurement |
A student writes in his notebook: “The plant on the windowsill is happier than the one in the cupboard.” Rewrite this as an observation.
Step 1 — Ask: what did he actually see? Probably that one plant is taller, greener, or has more leaves.
Step 2 — Replace the judgement word (“happier”) with something measurable.
Step 3 — Add the numbers: “The windowsill plant is 14 cm tall with 9 green leaves; the cupboard plant is 6 cm tall with 3 yellow leaves.”
Now a second student could check it, and a graph could be drawn from it. “Happier” allowed neither.
Three of these are observations and one is not. Which, and what kind of statement is it?
(a) The candle flame went out after 42 s.
(b) The water in the pan is bubbling.
(c) The metal must be a good conductor.
(d) The thermometer reads 78 °C.
Reveal answer
(c) is an inference — a conclusion drawn from something observed (perhaps the metal felt hot quickly). The other three record what was directly seen or read from an instrument. Notice the word “must”: it is a common sign that a sentence is a conclusion rather than an observation.
1.3 Testable questions
Not every good question is a scientific question. “Why is the sky beautiful?” is a fine question but no experiment can answer it. A question is testable when it names something you could change and something you could measure, so that an experiment could give an answer.
| Question | Testable? | Why / how to fix it |
|---|---|---|
| Why does sugar dissolve? | Not as written | Too big for a kitchen. Narrow it: “Does water temperature change how fast sugar dissolves?” |
| Does water temperature change how fast sugar dissolves? | Yes | Change: temperature. Measure: time to dissolve. |
| Is chocolate the best snack? | No | “Best” is opinion. A testable version: “Does eating chocolate 30 min before a test change the score?” |
| Do plants grow faster in blue light or red light? | Yes | Change: colour of light. Measure: height gained per week. |
Rewrite “Is exercise good for you?” as a testable question. Name what you would change and what you would measure.
Reveal answer
One good version: “Does the number of minutes of running per day affect resting heart rate?” Change: minutes of running per day (0, 10, 20, 30, 40). Measure: resting heart rate in beats per minute, taken first thing in the morning.
Any answer that names a variable to change and a quantity with a unit to measure is correct. “Is exercise good?” fails because “good” cannot be measured; “does exercise make you healthier?” fails for the same reason unless you say what “healthier” will be measured as.
1.4 The hypothesis: a prediction with a reason
A prediction says what you expect to happen. A hypothesis says what you expect and why. The difference is one word, and it is the most valuable word in this course:
Look at the three parts. If names the thing you change. Then names what you expect to measure. Because gives the scientific mechanism — the reason the change should cause the effect. Without the because, you have made a bet; with it, you have made a claim that can be right or wrong for a reason, which is what an experiment can test.
Write a hypothesis for: “Does the size of the sugar grains affect how fast they dissolve?”
Step 1 — If: name the change. “If the sugar grains are smaller…”
Step 2 — then: name the measured effect and its direction. “…then they will dissolve in less time…”
Step 3 — because: give the mechanism. “…because small grains have more surface touching the water for the same mass of sugar, so water particles can reach and remove sugar particles from more places at once.”
A weaker hypothesis — “smaller sugar dissolves faster because it is smaller” — repeats the prediction as the reason. The because must add a mechanism the prediction did not already contain.
A student writes: “If I use hot water, then the sugar will dissolve faster, because hot water is faster.” What is wrong with this hypothesis and how would you fix it?
Reveal answer
The “because” is circular: it says the effect happens because the effect happens. “Hot water is faster” is not a mechanism. Fix: “…because the particles in hot water have more energy and move faster, so they collide with the sugar more often and with more force, breaking it apart more quickly.” Now the reason is something separate from the prediction — and something a later experiment could test on its own.
“A hypothesis that turns out to be wrong means the experiment failed.”
Reveal answer
False. A hypothesis that the data do not support is a result. The experiment succeeded in testing it. The only failed experiment is one whose method was so poor that it could not have told you either way. Michelson and Morley’s most famous experiment (Lesson 5) gave the opposite result to their hypothesis — and won a Nobel Prize.
1.5 Testing fairly and concluding from numbers
Your sugar experiment changed only one thing: the temperature. Everything else stayed the same — the amount of water, the amount of sugar, no stirring, the same glasses. That is what makes it a fair test. If you had also used more water in the hot glass, you would not know which change caused the result. Lesson 3 is entirely about this, so here is just the rule: change one thing, measure one thing, keep everything else the same.
A conclusion then answers the question using the data, with numbers. Compare these two sentences about the Nairobi class results (cold mean 124 s, hot mean 21 s):
Using the Nairobi class data (cold: 118, 131, 124 s; hot: 21, 19, 24 s), calculate both means and state how many times faster the sugar dissolved in hot water, to the nearest whole number.
Reveal answer
Cold mean = (118 + 131 + 124) ÷ 3 = 373 ÷ 3 = 124.3 s (124 s to 3 significant figures).
Hot mean = (21 + 19 + 24) ÷ 3 = 64 ÷ 3 = 21.3 s.
Ratio = 124.3 ÷ 21.3 = 5.8, so about 6 times faster.
Marks are usually given for showing the addition, the division and the unit. Write the unit on every mean.
1.6 Repeats, and being checked
Your three cold trials gave three different times: 118, 131 and 124 s. None of them is “the” answer. Every measurement has small, unavoidable differences — you pressed the timer a fraction late, one sugar grain was bigger, the water cooled slightly between trials. These are called random errors, and they scatter above and below the true value. Taking the mean of several trials cancels much of the scatter, which is why repeating is part of the method rather than an optional extra.
The last step is the one students most often forget: a result nobody else can repeat is not yet science. Scientists publish their method in enough detail that a stranger on another continent can copy it and see whether they get the same thing. Being checked is not an insult. It is the entire reason science is trusted where opinion is not.
…the Nairobi class had done only one trial for each glass and got cold 131 s, hot 19 s? Would their conclusion be wrong? What would they have lost?
Reveal answer
The conclusion (hot is much faster) would still be right in this case — the effect is large. But they would have no way of knowing how reliable their numbers were: a single reading of 131 s could have been 118 s on another day. With three trials they can see the scatter (118–131 s) and know that the six-fold difference between hot and cold is far bigger than the scatter, so it is real. With one trial they would be trusting luck. When the effect is small — say hot 60 s and cold 70 s — a single trial can easily give the wrong answer.
Explore it live
Try it: Change one thing about glass B at a time, watch the timers, and decide whether your test is fair.
Common mistakes to avoid
- Writing the prediction twice. “…because hot water dissolves faster” is not a reason. The because must name a mechanism (particles, energy, surface area…).
- Mixing what you saw with what you think. Record “21 s”, not “it was quick”. Keep inferences for the conclusion.
- Saying “proved”. Scientists say the evidence supports a hypothesis. One surprising result can overturn any idea, so the door is kept open.
- Treating different trial results as failure. Scatter is normal; that is why you take a mean.
- Skipping the numbers in a conclusion. “Faster” is worth little; “from 124 s to 21 s” is worth the marks.
How we found out — the man who refused to believe
In 1847, in a hospital in Vienna, a young doctor named Ignaz Semmelweis noticed something terrible. In the ward where medical students delivered babies, about one mother in ten died of fever. In the ward next door, run by midwives, far fewer died. Same hospital, same city, same year. That was his observation — and he wrote it down as numbers, month by month, not as a feeling.
Everyone had explanations. Bad air. The mothers’ fear of the students. The priest’s bell. Semmelweis did not accept any explanation without evidence. He looked for the one thing that was different between the wards — and found it: the students came to the ward straight from dissecting dead bodies. The midwives did not. His hypothesis: something invisible was being carried on the students’ hands from the dead to the living.
His test: he ordered every student to wash their hands in chlorinated water before touching a patient. Within months the death rate in the students’ ward fell to the same level as the midwives’. The data were unmistakable — a fair test, because nothing else about the ward had changed.
Here is the part that makes it a real science story rather than a fairy tale: most doctors refused to believe him. Nobody had seen a bacterium yet; the idea that a gentleman’s hands could carry death was insulting. Semmelweis was right, he had the evidence, and he was still ignored for twenty years, until Louis Pasteur and Robert Koch showed the microbes under a microscope in the 1860s and 1870s. Evidence wins in the end — but the end can take a long time, and being right is not the same as being believed.
Semmelweis’s hypothesis was correct, but he could not say what was on the students’ hands. Does that make his hypothesis unscientific? Explain.
Reveal answer
No. A hypothesis must be testable, not complete. “Something on the hands carries the disease, and washing removes it” predicts that washing will reduce deaths — a prediction that can be tested and could have failed. It passed. The mechanism (bacteria) was filled in later by Pasteur and Koch, exactly as the loop predicts: one answer opens the next question. A hypothesis that explains everything but predicts nothing would be the unscientific one.
Where it is used now
- Botswana and Kenya — malaria control. Deciding whether a new type of bed net works is done exactly like your sugar experiment: villages with the new net and villages with the old net, everything else the same, and the number of malaria cases counted for a year. Only the numbers decide, and the trial is published so other countries can check it before buying millions of nets.
- Denmark — wind turbines. Engineers test blade shapes in wind tunnels, changing one measurement at a time and recording the power produced. A turbine blade is a hypothesis that survived hundreds of tests.
- Everywhere — your phone’s battery. Every improvement in how long a battery lasts came from someone changing one ingredient, keeping everything else the same, and measuring. There is no other way to find out.
Key terms
| Observation | Information gathered with the senses or an instrument and recorded as it is, ideally as a number with a unit. |
| Inference | A conclusion drawn from observations. |
| Testable question | A question that names something to change and something to measure, so an experiment can answer it. |
| Hypothesis | A prediction with a scientific reason, written as If… then… because…. |
| Fair test | An experiment in which only one variable is changed and everything else is kept the same. |
| Mean | The sum of a set of readings divided by how many there are. |
| Random error | Small, unavoidable scatter in readings above and below the true value; reduced by repeating and averaging. |
| Evidence | Observations and data that support or do not support a claim. |
Say it like a scientist
Science now — and the scientist you need to be
What is happening in 2026
The loop you learned today has not changed since Galileo, but the speed of it has. In 2020 an artificial-intelligence system called AlphaFold worked out the three-dimensional shape of almost every protein known to biology — a problem that had taken human scientists fifty years to solve for a few hundred proteins. The AI did not replace the loop; it ran the “hypothesise” step millions of times a second, and human scientists still had to test its predictions in the laboratory before anyone trusted them. In 2023 another system, GNoME, proposed 2.2 million new crystal materials; laboratories are now checking which of them are real. The bottleneck in science today is no longer having ideas. It is testing them properly and knowing which results to believe — exactly the two skills in this lesson.
The mindshift
What you need to be able to do
Employers and universities in every country now list the same three things above any list of facts: ask a testable question, judge whether evidence is good enough, and explain your reasoning to someone who disagrees. Every lesson in this course ends with a task that practises them. Facts you can look up in two seconds; these you cannot.
See an example
Claim: “Listening to music while studying helps you remember more.” Testable question: Does listening to music while learning a list of 20 words change the number recalled after 10 minutes? Hypothesis: If music is playing while learning, then fewer words will be recalled, because attention is shared between the music and the words, so each word is processed less deeply. Fair test: same 20-word list difficulty (two lists checked to be equally hard), same 5-minute learning time, same quiet room, same person tested on different days at the same time; music vs silence is the only change; repeat with five people; compare mean words recalled. The design could prove the claim wrong — which is what makes it a test rather than an argument.
Think like a scientist
Criterion B A student says: “I want to find out whether sugar dissolves faster if I use more sugar.” Write a hypothesis for her in the If… then… because… form. Then say one thing she must keep the same for the test to be fair, and how she would check it.
Model answer
Hypothesis: If more sugar is added to the same amount of water, then it will take longer for all of it to dissolve, because there are more sugar particles for the same number of water particles to pull apart, so each water particle has more work to do.
Keep the same: the temperature of the water — checked with a thermometer before each trial and kept at 20 °C. (Also acceptable: the volume of water measured with a measuring cylinder; whether it is stirred; the type of sugar.) Notice the answer names a measurable thing and the instrument — “the water” alone is not enough.
Probe deeper — exam-style questions
Attempt every question before opening the answer. Marks in brackets show what a full answer is worth.
State the difference between an observation and an inference, and give one example of each from an experiment of your choice.
Reveal answer
An observation is what is directly detected and recorded (e.g. “the ice cube melted in 8 minutes”); an inference is a conclusion drawn from observations (e.g. “the room must be warm”). [1 for the distinction, 1 for a matched pair of examples]
A student wants to know whether the colour of a cup affects how quickly hot tea cools. Write a testable research question and a hypothesis in the If–then–because form.
Reveal answer
Research question: “How does the colour of the cup affect the temperature drop of tea in 10 minutes?” [1]
Hypothesis: “If the cup is black, then the tea will cool by more degrees in 10 minutes than in a white cup, because dark surfaces emit (radiate) heat faster than light surfaces.” [1 for a correct if–then, 1 for a mechanism in the because]
“My results were 45 s, 47 s and 61 s, so my experiment did not work.” Explain what the student should do instead of giving up.
Reveal answer
The 61 s reading is far from the other two and is probably anomalous (a random error such as a late stop of the timer). The student should repeat that trial [1]; if the repeat is close to 45–47 s, use the repeat, calculate the mean of the consistent readings and state that one anomalous result was excluded and why [1].
Explain why Semmelweis’s colleagues were wrong to reject his result even though nobody had seen a bacterium.
Reveal answer
In science a claim is judged on the evidence, not on whether the mechanism is known [1]. His data showed that hand-washing reduced deaths in a fair comparison — the effect was real regardless of what caused it, and the correct response was to repeat his test, not to dismiss it because it clashed with their beliefs [1].
Three trials of sugar dissolving in warm water gave 48 s, 52 s and 50 s.
(a) Calculate the mean.
(b) A fourth trial gives 49 s. Without recalculating, state whether the mean will go up or down and explain how you know.
Reveal answer
(a) (48 + 52 + 50) ÷ 3 = 150 ÷ 3 = 50 s [1, with unit].
(b) It will go down slightly [1], because the new reading (49 s) is below the current mean (50 s), so it pulls the average towards itself [1]. (New mean = 199 ÷ 4 = 49.75 s.)
True or false, with a reason: “If a scientist publishes a result and another laboratory cannot repeat it, the second laboratory must have made a mistake.”
Reveal answer
False [1]. A result that cannot be repeated is not yet accepted; either laboratory could be wrong, and the disagreement itself is the signal to look for a hidden variable or error in both methods. Reproducibility is the test, not the reputation of whoever published first [1].
Summary
- Science is a loop: observe → ask a testable question → hypothesise (with a because) → test fairly → conclude from the numbers → share so others can check.
- An observation is what you detect; an inference is what you conclude; an opinion cannot be measured at all.
- A question is testable when it names something to change and something to measure.
- The “because” in a hypothesis must be a mechanism, not the prediction repeated.
- Repeat and take a mean to reduce random error; quote the numbers in every conclusion; say “supports”, not “proves”.
Next time → Measuring: how do you know your number is right?