This is the sense-making phase. You’ve got a graph — now you have to say what it means. Interpreting well is a two-part move: describe the trend in proper scientific language, then explain it using physics. Layer on what the gradient and intercept reveal, how you justify an anomaly, and the precise vocabulary of accuracy and precision, and you’ve got an interpretation that scores.
📘 What you need to know
Interpreting is two steps: describe the trend, then explain it with physics
The gradient often equals a physical quantity; the y-intercept can expose a systematic error
Error bars show precision — a good fit passes through them all
Mark and justify anomalies with a likely experimental cause, never just “it looks odd”
Accuracy = close to the true value; precision = repeats close together — they’re different
Reliability = consistent when repeated; validity = a fair test that answers your question
Describe, then explain the trend
Interpreting a graph is always a two-step process, and students who skip the second step leave marks on the table.
Describe
→ what →
state the trend
Explain
→ why →
use the physics
To describe, state what the graph shows using the right terms: directly proportional, linear positive correlation, inversely proportional, exponential increase. To explain, reach for the physical principle or equation that makes the data behave this way. The explanation is the most important part — it turns a description into science.
Quick recap: “directly proportional” is a promise of a straight line through the origin. If the line has an intercept, it’s a linear correlation, not a proportional one.
Reading the gradient and intercept
Once your line of best fit is drawn, its features carry real physics.
The gradient
The slope frequently is a physical quantity. For a current–voltage graph of an ohmic resistor (I against V), the gradient equals 1 ÷ R — the reciprocal of resistance. In a linearised pendulum graph of T2 against L, the gradient equals 4π2 ÷ g, so measuring it hands you a value for g.
The y-intercept
The intercept tells you the dependent variable’s value when the independent variable is zero — and it’s a brilliant detector of systematic error. If theory says the line should pass through the origin but yours cuts the axis at, say, +0.30 Ω, that offset is evidence of something constant skewing every reading (like contact resistance, or a zero error on a sensor).
🧭 What the graph’s features tell you
Gradient — often a physical quantity (1÷R, 4π2÷g, and so on). Read it off, then solve.
y-intercept — the value at zero; a non-zero one you didn’t expect signals a systematic error.
Area under the curve — can be a total quantity, like displacement from a velocity–time graph.
Error bars — small bars mean precise data; a best-fit line through all of them means a good fit.
WE 1
A student plots T2 against L for a pendulum and measures a gradient of 4.03 s2 m−1. Find g and comment on the accuracy (accepted value 9.81 m s−2).
Step 1 — link gradient to the physics
gradient = 4π² ÷ g, so g = 4π² ÷ gradient
g = (4 × π²) ÷ 4.03 = 9.80 m s⁻²g = 9.80 m s⁻²Step 2 — percentage error vs accepted value
% error = |9.80 − 9.81| ÷ 9.81 × 100
= 0.14%A tiny error → the result is highly accurate. Accuracy is only judgeable because a literature value exists to compare against.
Justifying anomalies
An anomaly (outlier) is a point that clearly breaks the trend. Highlight it on the final graph — never quietly delete it — and in your analysis justify why it’s anomalous by linking it to a likely experimental error.
Circle an anomaly on the graph and exclude it from the line of best fit — then explain its likely cause in your analysis.
A strong justification names a mechanism: “the reading at 60 °C was excluded as it sat far above the others, most likely from a delay in starting the stopwatch — a random error.” That links the point to a specific error, which is exactly what examiners want.
Accuracy, precision, reliability, validity
These four words have exact scientific meanings, and using them correctly signals real understanding. The classic mix-up is accuracy versus precision.
Precision is the spread of your shots; accuracy is how close their centre sits to the bullseye. A group can be one without the other.
🧭 The four terms, precisely
Accuracy — how close the result is to the true value. Improve it by repeating and averaging; it’s harmed by systematic errors. Only judgeable against a literature value.
Precision — how close repeat readings are to each other. A small spread = high precision; it’s harmed by random errors.
Reliability — consistency when the experiment is repeated. Ask: would someone else reach the same conclusion?
Validity — whether the method was a fair test that actually answers the research question, with all other variables controlled.
WE 2
A resistance–length graph for a wire has small error bars, tightly-fitting points, and a line that cuts the y-axis at +0.30 Ω instead of the origin. Interpret the precision, reliability, and accuracy.
Precision & reliability
small error bars → precise readings
points hug the line → reliable, strong relationship
Accuracy — read the intercept
a zero-length wire should have zero resistance
but the line cuts at +0.30 Ω→ a systematic errorLikely contact resistance from the crocodile clips, adding a fixed extra amount to every reading — it shifts the whole line up without changing its slope.
💡 Top tips
Always explain, don’t just describe — link the trend back to a physical principle or equation. That’s where the marks are.
Keep circling back to the research question — the point of interpretation is to answer it, not just narrate the graph.
Handle anomalies transparently — mark them, keep them visible, and justify exclusion with physics, not “it looks odd”.
Use the vocabulary deliberately — accuracy, precision, reliability, and validity each mean something specific; using them right shows you understand your data’s quality.
⚠ Common mistakes
Describing the trend but never explaining it with physics
Swapping accuracy and precision — or commenting on accuracy with no literature value to compare against
Ignoring a non-zero intercept that’s quietly flagging a systematic error
Deleting an anomaly instead of marking and justifying it
That wraps up Inquiry 2: Collecting & Processing Data. You can now take raw measurements all the way to a fully interpreted result — collect clean data, process it with propagated uncertainties, and read the story your graph is telling. Next in the cycle: concluding and evaluating your investigation.
Want this to actually click before the exam?
Book a free meeting and let’s work through the tricky bits together.
