Scientific investigation & inquiry skills

What you will learn

how to turn a vague curiosity into an investigable question and a testable hypothesis,
how to identify independent, dependent and controlled variables in an experiment,
how to plan a fair test, recognise risks, and choose suitable equipment,
how to represent data using tables, line graphs, bar graphs and scatter plots,
how to analyse results, identify anomalies, and write an evidence-based conclusion.

Worked example 0 Real-world example: does fertiliser make plants grow taller?

Question: How does the amount of fertiliser added to soil affect the height of bean plants after 4 weeks?

Hypothesis: If more fertiliser is added, the plants will grow taller (up to a point), because plants use minerals from fertiliser to build tissue.
Variables:
- Independent (what I change): amount of fertiliser (0, 5, 10, 20 g per pot).
- Dependent (what I measure): plant height in cm after 4 weeks.
- Controlled (kept the same): same bean variety, same pot size, same soil, same light, same water, same start date.
Method: $4$ pots per fertiliser level (total $16$ pots). Measure weekly. Record carefully.
Data: average height per treatment group — plot on a bar graph or line graph.
Conclusion: if taller plants appeared with more fertiliser up to some level, and no taller with more, this supports the hypothesis within that range.

Key idea: a fair test changes one thing at a time while keeping everything else constant. This is what lets you say “the fertiliser caused the difference.”

1. Questions and hypotheses

A good investigable question is answerable by measurement. Examples:

Weak: “Are plants healthy?” (vague, no variables).
Better: “How does the amount of light affect the height of a sunflower?”

A hypothesis is a testable prediction that links a cause to an effect. A common structure:

If [I change X], then [Y will do Z], because [scientific reason].

2. Variables

Independent variable (IV): the one thing the experimenter deliberately changes. Usually drawn on the x-axis.
Dependent variable (DV): the measurement that may respond. Drawn on the y-axis.
Controlled variables (CVs): everything else that could affect the DV, kept the same.

Worked example 1 Identifying variables

Question: Does the type of drink bottle affect how long water stays cold in the sun?

IV: the type of bottle (stainless steel, plastic, glass).
DV: temperature of the water after $2$ hours (°C).
CVs: starting water temperature, starting volume, position in the sun, time of measurement, room/outdoor conditions, thermometer used.

Key idea: every CV you miss is a possible alternative explanation that weakens your conclusion.

3. Planning a fair test

Key steps before you start:

State the aim, IV, DV and CVs clearly.
Plan multiple readings (replicates) so you can average and spot anomalies.
Choose equipment that gives the right precision (a $30$ cm ruler for plant height, a digital thermometer to $0.1$ °C).
Identify and manage risks: hot liquids, glassware, chemicals, electricity. State the safety measure.
Check for ethics if using living things: no harm, appropriate numbers.

4. Recording data

A clear table has headings with units, independent variable on the left, and space for repeats.

Fertiliser (g)	Height trial 1 (cm)	Height trial 2 (cm)	Height trial 3 (cm)	Mean (cm)
0	12	13	12	12.3
5	18	17	19	18.0
10	24	23	25	24.0
20	22	21	20	21.0

5. Graphs

Line graph: when both variables are numerical and continuous (e.g. time vs temperature).
Bar graph: when the IV is a category (e.g. type of bottle, brand of battery).
Scatter plot: when you have pairs of measurements and want to see correlation.
Pie chart: for proportions of a whole.

Example line graph of plant height vs fertiliser added. Note the anomaly at 20 g.

6. Analysis and conclusion

An anomaly is a data point that does not fit the trend. Possible causes: measurement error, a disturbed specimen, a real effect (e.g. too much fertiliser is harmful). Report anomalies — do not silently drop them.

A conclusion links back to the hypothesis using the evidence.

Worked example 2 Writing a conclusion

Results: the data in the table above.

“As fertiliser increased from $0$ to $10$ g, mean plant height increased from $12.3$ to $24.0$ cm — roughly doubling.”
“At $20$ g, mean height dropped to $21.0$ cm — this is an anomaly relative to the trend.”
“The results support the hypothesis up to $10$ g of fertiliser. Above this, extra fertiliser may damage the plants, so the hypothesis is only partly supported.”
“Limitations: small sample, only one variety of bean, only $4$ weeks — repeat with more plants over longer time.”

Key idea: conclusions state what the evidence shows, what it does not show, and what uncertainties remain.

7. Errors, limitations and evaluating

Random errors: small fluctuations (thermometer reading to nearest 0.5°C). Reduce by taking the mean of repeats.
Systematic errors: consistent bias (a misaligned scale). Reduce by calibrating equipment.
Limitations: things your design could not control (weather changing between test days).

Worked example 3 Spotting an unfair test

A student times how long different ice blocks take to melt. Block A sits in a warm kitchen; block B sits on a shaded bench outside.

IV is supposed to be the ice block itself — but the location is also different.
Temperature, light and air flow are not controlled.
Any difference in melt time could be due to the location, not the block.

To fix: melt all blocks in the same environment, one at a time or side by side.

Practice: Year 7

Fluency

Tier 1: recall and identify

Define: investigable question, hypothesis, independent variable, dependent variable, controlled variable.
Write a hypothesis for: “Does the height of a ramp affect how far a toy car rolls after leaving it?”
In the experiment above, identify IV, DV and two CVs.
State two reasons to repeat a measurement.
Which type of graph should you use for: (a) temperature over time, (b) brand of battery vs total run time, (c) height vs weight of classmates?
What is an anomaly?
What is a risk assessment, and why is it needed?
Name one random error and one systematic error in measuring liquid volume.
What is a control group?
State one reason why sample size matters.

Reasoning

Tier 2: explain and reason

Explain the difference between an observation and an inference.
Explain why controlling variables is essential for drawing cause-and-effect conclusions.
Why is a hypothesis written before data is collected?
A student presents only their three “best” data points. Explain why this is poor science.
Explain why a larger sample size makes a conclusion more reliable.
A friend says, “ice cream causes shark attacks, because both rise in summer.” Identify the logical error.

Problem solving

Tier 3: apply to a novel context

Design a fair test to answer: “Does the colour of a drink bottle affect how quickly water inside heats in the sun?” State IV, DV, at least four CVs, and the number of replicates.
A student records the following lengths of plant shoots (cm): $12, 13, 14, 13, 28, 14, 12$ . Identify the anomaly and suggest two possible causes.
Sketch (in words, not on paper) the shape of a graph you would expect for water temperature over 20 minutes as an ice cube melts and then warms in a beaker. State what the y-axis and x-axis show.
A class tests four brands of paper towel to see which absorbs most water. Describe a procedure with a clear IV, DV, three CVs and a replicate count.

Challenge

Reasoning

Harder reasoning

A study finds that students who eat breakfast perform better on tests. Does this prove that eating breakfast causes better performance? Suggest two alternative explanations and describe how a better study could tell them apart.
A class measures the boiling point of water on five different hotplates and gets readings of $99$ , $101$ , $100$ , $100$ , $102$ °C. Which is most likely anomalous? Calculate the mean with and without it, and decide which is a better estimate of the true value.
You want to know if a new fertiliser really works. Explain why you need a control group (no fertiliser) and why simply comparing “before” and “after” on the same plants is not enough.
Design a simple investigation to test whether the length of a pendulum affects its swing period. State IV, DV, CVs, procedure, and what a graph of length vs period would look like.

Answers

Answer key

Attempt the practice first. When you're ready to check, expand the answers below.

Show the full answer key

Year 7 answers

Fluency

Tier 1: recall and identify

Investigable question: one answerable by measurement. Hypothesis: a testable prediction linking cause to effect. IV: what the experimenter changes. DV: what is measured. CV: anything else kept the same.
Example: “If the ramp is raised higher, the car will roll a greater distance after leaving the ramp, because the car starts with more gravitational potential energy converted to kinetic energy.”
IV: ramp height. DV: distance rolled. CVs: same car, same surface, same starting point on ramp, same release (no pushing), same ramp material.
To reduce random error and identify anomalies by taking the mean of several readings.
(a) Line graph. (b) Bar graph. (c) Scatter plot.
A data point that does not fit the overall pattern of the results.
A list of possible hazards and how to control them, done before the experiment to keep the experimenter, others, and the environment safe.
Random: parallax when reading a meniscus. Systematic: a measuring cylinder miscalibrated so it reads $1$ mL too high.
A group treated exactly like the experimental groups except for the IV. It shows what happens without the “treatment” for comparison.
Larger samples average out random variation, making results more reliable and any real effect easier to detect.

Reasoning

Tier 2: explain and reason

An observation is what you directly see or measure (the leaf is yellow). An inference is an explanation you infer from the observation (the plant lacks nitrogen).
If several variables change at once, you cannot tell which caused the effect. Controlling variables isolates the IV as the only possible cause.
Writing it first prevents you from unconsciously shaping the design or interpretation to match the data — this is called confirmation bias.
Selecting “best” data misrepresents the experiment. Science requires honest reporting of all data, including outliers and disagreements with the hypothesis.
With more data, random variation cancels out more completely and any real pattern stands out more clearly from chance fluctuations.
Correlation without causation — both rise in summer because hot weather drives both independently. One does not cause the other.

Reasoning

Tier 3: apply to a novel context

IV: bottle colour (e.g. black, white, blue, red). DV: water temperature after $1$ hour. CVs: same bottle material and volume, same starting water temperature, same position in sun, same weather conditions, same thermometer, same time of day. Replicates: at least $3$ per colour.
Anomaly: $28$ . Possible causes: typo for $13$ or $14$ ; wrong shoot measured (a different species); a genetic variant in that plant; measurement taken at a different date.
Temperature vs time (time on x-axis, temperature on y-axis). Roughly flat at $0$ °C while the ice melts, then rising as the water warms. A line graph with a flat region then a rise.
IV: paper-towel brand. DV: mass of water absorbed (g). CVs: same piece size, same water volume, same dipping time, same squeezing. Replicates: $3$ strips per brand.

Reasoning

Challenge

No. Alternative explanations: (i) students who eat breakfast may also sleep more or live in wealthier homes, and those factors drive both behaviours. (ii) students feeling good study more regularly and eat breakfast more regularly. A controlled experiment (randomly assigning students to eat or skip breakfast, with other conditions matched) would isolate the cause.
$102$ looks anomalous — pure water boils at $100$ °C at normal pressure. Mean with it: $(99+101+100+100+102)/5 = 100.4$ . Mean without: $(99+101+100+100)/4 = 100.0$ . The $100.0$ value is a better estimate if we suspect $102$ was a misreading or a calibration fault.
Without a control, you cannot tell if changes in the plants are due to the fertiliser or to other changes over time (weather, age, light). Some growth would happen anyway. A control group receiving no fertiliser lets you separate the fertiliser’s effect from natural growth.
IV: length of pendulum. DV: time for $10$ complete swings (divide by $10$ for period). CVs: same mass, same release angle, same location, same stopwatch. Procedure: tie string of known length, release from small angle, time $10$ swings, repeat $3$ times, calculate mean period. Vary length (e.g. $20, 40, 60, 80, 100$ cm). Graph: period increases with length, as a curve (square-root relationship).

Prefer paper? Print the answer key as a separate booklet: open print view ->