Scatter graphs are graphs that plot two sets of numerical data against each other as points, so you can see whether there is a relationship between them. In this guide I explain how to draw scatter graphs, how to describe correlation, how to add a line of best fit, how to interpolate and extrapolate, and how to spot and explain anomalies.

Table of Contents
What are scatter graphs?
Scatter graphs show whether one variable changes when another variable changes. Each point on the graph is one piece of paired data. For example, one point might be a single site on a river where you measured both the distance downstream and the channel width.
Geographers use scatter graphs because so much of geography is about relationships, such as whether river velocity increases with depth or whether richer countries have a higher life expectancy. A table of numbers can hide a pattern. Scatter graphs make it visible in seconds.
The key idea is that one variable is usually independent and the other is dependent. The independent variable is the one you think causes the change, and it goes on the x axis (horizontal). The dependent variable is the one that responds, and it goes on the y axis (vertical).
Key terms you need to know
- Variable: something that can be measured and can change, such as rainfall, distance or population.
- Independent variable: the variable that is thought to cause the change. Plotted on the x axis.
- Dependent variable: the variable that responds to the change. Plotted on the y axis.
- Correlation: the relationship between two variables. It can be positive, negative or none.
- Line of best fit: a straight line drawn through the middle of the points to show the overall trend.
- Anomaly (outlier): a point that does not fit the general pattern.
- Interpolation: estimating a value that falls inside the range of your data.
- Extrapolation: estimating a value beyond the range of your data by extending the line of best fit.
How to draw and read scatter graphs: 6 essential steps
The six steps below take you from a table of raw data to a full interpretation. I use the same river example as the diagram at the top of this page, where a student measured channel width at ten sites along a small river.
1. Choose the right variables and axes
Before you draw anything, decide which variable is independent. In the river example, distance downstream is independent because channel width changes as you move downstream, not the other way round. Distance goes on the x axis and width on the y axis.
Next, choose a scale. Look at the smallest and largest values in each data set. Your scale must cover the full range, use even steps (for example 0, 2, 4, 6) and fill most of the grid. A graph squashed into one corner is hard to read and loses marks.
- Label both axes with the variable name and the unit, for example “Channel width (m)”.
- Start at zero unless there is a good reason not to.
- Give the graph a clear title.
2. Plot the points accurately
Plot each pair of values as a small cross or dot. Use a sharp pencil. For the river data, site 1 is at 1 km with a width of 1.8 m, so you go across to 1 on the x axis and up to 1.8 on the y axis. Never join the points up with a line. Scatter graphs are about the overall spread of points, not a sequence.
In the exam you are often asked to complete scatter graphs by plotting two or three missing points. These are usually worth 1 or 2 marks and need to be accurate, so double check each point against the table.
3. Describe the correlation
Correlation is the pattern made by the points. There are three main types, and you should also comment on how strong the relationship looks.
| Type of correlation | What the points do | Geography example |
|---|---|---|
| Positive | Rise from bottom left to top right. As x increases, y increases. | Distance downstream and channel width |
| Negative | Fall from top left to bottom right. As x increases, y decreases. | Distance from the CBD and pedestrian count |
| No correlation | Spread randomly, with no clear pattern. | Height of students and their house number |
| Strong | Points lie close to a straight line. | Altitude and air temperature on a mountain (strong negative) |
| Weak | A general trend, but points are widely scattered. | GDP per capita and car ownership in a small sample |
When you describe scatter graphs in an answer, use the language “as … increases, … increases (or decreases)”. Then add data. For example: “There is a strong positive correlation. As distance downstream increases from 1 km to 10 km, channel width increases from 1.8 m to 8.8 m.”
4. Draw a line of best fit
A line of best fit is a single straight line that follows the trend of the points. It does not have to pass through the origin, and it does not have to touch any particular point. Aim for roughly the same number of points above and below the line, and ignore any obvious anomaly when you decide where it goes.
Use a ruler. A freehand wobbly line is not accepted.
The steeper the line, the faster the dependent variable changes. In the river example the width rises by about 0.8 m for every kilometre downstream.
5. Interpolate and extrapolate
Once you have a line of best fit, you can use it to make estimates.
- Interpolation means estimating a value inside your data range. In the river example, there was no site at 5.5 km. Draw a line up from 5.5 km to the line of best fit, then across to the y axis. It gives a width of about 5.4 m.
- Extrapolation means extending the line beyond your data to estimate a value outside the range, such as the width at 15 km.
Interpolation is fairly reliable because it sits between real measurements. Extrapolation is much less reliable. The river might join a tributary, flow through a gorge or reach a reservoir, and then the trend would change. Always say this if a question asks you how confident you are in an estimate.
6. Identify and explain anomalies
An anomaly is a point that sits well away from the line of best fit. In the diagram, site 6 is only 2.0 m wide even though the trend suggests it should be about 5.8 m. Geographers do not simply delete anomalies. Instead, you should try to explain them.
There are two kinds of explanation:
- A real geographical reason. The river may cut through a band of hard rock, so the channel is narrow and deep. A footbridge or weir might also alter the channel.
- A mistake in data collection. The tape measure may have sagged, the wrong bank may have been used, or a reading was written down incorrectly.
Spotting and explaining anomalies is a common 2 or 3 mark question, and it links directly to evaluating your fieldwork.
Scatter graphs in geography: worked examples
Here are three examples that match topics you study.
River fieldwork: the Bradshaw model
The Bradshaw model predicts how a river changes downstream. Width, depth, velocity and discharge should increase, while gradient, load particle size and angularity should decrease. Scatter graphs are the best way to test these predictions with your own data. Plot distance downstream against width, depth or mean pebble size and see whether the correlation matches the model. You can read more in my guide to the Bradshaw model and in the post on the long profile of a river.
Development: GDP and life expectancy
One of the most famous scatter graphs in geography plots GDP per capita on the x axis and life expectancy on the y axis for every country in the world. It shows a clear positive correlation. Richer countries such as Japan and Norway have life expectancies above 80 years, while many low-income countries in sub-Saharan Africa have life expectancies in the 50s or 60s. You can explore an interactive version on Our World in Data.
The pattern is not a straight line, though. Life expectancy rises steeply at low incomes and then levels off. This is why the chart is often drawn with a logarithmic x axis. The topic links to measuring development and the development gap.
Urban fieldwork: distance from the CBD
In an urban enquiry you might count pedestrians, measure building height or record an environmental quality survey score at points along a transect from the central business district. Scatter graphs of distance against pedestrian count usually show a negative correlation, because footfall is highest in the main shopping streets and falls away towards the suburbs.
Strengths and limitations of scatter graphs
| Strengths | Limitations |
|---|---|
| Show clearly whether a relationship exists | Do not prove that one variable causes the other |
| Make anomalies easy to spot | Need enough data points (at least 8 to 10) to be meaningful |
| Allow interpolation and extrapolation | Lines of best fit drawn by eye can vary between students |
| Quick to draw and easy to read | Only compare two variables at a time |
The most important limitation is that correlation is not the same as causation. Ice cream sales and drowning both rise in summer, but one does not cause the other. Both are linked to a third factor, hot weather. When you explain scatter graphs, use geographical reasons to show why one variable might affect the other.
Using scatter graphs in fieldwork
If your enquiry asks “How does X change with Y?”, scatter graphs are almost always the right choice of presentation. Plan this before you collect data. You need paired measurements, so every site must record both variables, and you need enough sites for a pattern to appear.
- Write a hypothesis, for example “Channel width will increase with distance downstream”.
- Use systematic sampling to choose sites at equal intervals along the river.
- Measure both variables carefully at every site and record them in a table.
- Plot scatter graphs, add a line of best fit and describe the correlation.
- Explain any anomalies and decide whether the hypothesis is supported.
Statistical tests such as Spearman’s rank correlation coefficient are more typical of A level, but they are based on exactly the same idea.
Scatter graphs: frequently asked questions
Should you join the points on scatter graphs?
No. Never join the points dot to dot. Draw a single line of best fit instead, using a ruler, to show the overall trend.
Does a line of best fit have to go through zero?
No. A line of best fit follows the trend of the data. It only passes through the origin if the data suggest it should. In the river example, it crosses the y axis at about 1 m.
What is the difference between interpolation and extrapolation on scatter graphs?
Interpolation estimates a value inside the range of your data. Extrapolation estimates a value outside that range by extending the line. Interpolation is more reliable.
How many points do scatter graphs need?
There is no fixed rule, but scatter graphs with fewer than about eight points rarely show a reliable pattern. More sites give you more confidence in your conclusion.
Scatter graphs: key takeaways
- Scatter graphs show the relationship between two sets of numerical data.
- Put the independent variable on the x axis and the dependent variable on the y axis.
- Correlation can be positive, negative or none, and strong or weak.
- A line of best fit is ruled through the middle of the points, ignoring anomalies.
- Interpolation is more reliable than extrapolation.
- Explain anomalies using geography or errors in data collection.
- Correlation does not prove causation.
To conclude
Scatter graphs are simple to draw but powerful to use. If you can plot points accurately, name the correlation, rule a sensible line of best fit and explain the odd point that does not fit, you will pick up easy marks in the skills and fieldwork questions.
