Scatter Diagrams and Correlation: Finding Connections in Data
Welcome to this guide on Scatter Diagrams and Correlation! Have you ever wondered if there is a relationship between how much time you spend revising and the grades you get? Or whether taller people have bigger shoe sizes? In statistics, we use scatter diagrams to investigate whether two different things (variables) are linked. Don't worry if this topic sounds a bit technical at first — once you see the patterns, it becomes very straightforward and fun to work with!
What you will learn in this chapter:
• What bivariate data and scatter diagrams are
• How to identify different types of correlation (positive, negative, and zero)
• How to describe the strength of a relationship
• How to draw an accurate line of best fit
• How to use your line of best fit to estimate values (interpolation and extrapolation)
• Why correlation does not always mean causation
1. What is a Scatter Diagram?
A scatter diagram (often called a scatter graph) is used to display bivariate data. The word bivariate simply means "two variables" (data that comes in pairs for each subject or item).
Example: If you ask \(10\) students for their height and their shoe size, each student provides a pair of numbers \((x, y)\). Plotting these points on a grid produces a scatter diagram.
Key Features of a Scatter Diagram:
• The horizontal axis (\(x\)-axis) usually shows the independent variable (e.g., temperature).
• The vertical axis (\(y\)-axis) usually shows the dependent variable (e.g., ice cream sales).
• Each data pair is marked with a small cross \((\times)\) or dot.
• Important: Never connect the points with lines like a dot-to-dot puzzle!
Quick Takeaway: A scatter diagram displays pairs of numerical values to help us see if there is any visible pattern or trend between them.
---2. Understanding Correlation
Correlation describes the relationship between two variables. When looking at a scatter diagram, we check whether the points follow a general direction.
Types of Correlation:
1. Positive Correlation:
As the variable on the \(x\)-axis increases, the variable on the \(y\)-axis also increases.
• Visual pattern: The points slope upwards from bottom-left to top-right.
• Real-life example: The number of hours you revise and your test score. More revision generally leads to higher scores.
2. Negative Correlation:
As the variable on the \(x\)-axis increases, the variable on the \(y\)-axis decreases.
• Visual pattern: The points slope downwards from top-left to bottom-right.
• Real-life example: The age of a car and its selling price. As a car gets older, its value usually goes down.
3. Zero / No Correlation:
There is no link between the two variables.
• Visual pattern: The points are scattered randomly across the grid with no clear slope.
• Real-life example: A person's height and their score in a French vocabulary test.
Strength of Correlation:
We can also describe how strong a correlation is by looking at how close the points are to forming a straight line:
• Strong correlation: The points are packed closely together around a straight trend line.
• Weak correlation: The points follow a general trend, but are more widely spread out.
Memory Aid: Think of a hill! Going up the hill from left to right is Positive. Going down the hill from left to right is Negative.
Did You Know? In exams, when asked to describe a relationship, always mention both variables in context! For example: "As the temperature increases, the sales of hot soup decrease."
---3. The Line of Best Fit
A line of best fit is a straight line drawn through the data points to represent the overall trend. It allows us to make reasonable estimates for unknown values.
Rules for Drawing a Good Line of Best Fit:
1. Use a clear ruler and a sharp pencil. It must be a single straight line, not drawn freehand.
2. Follow the trend: The line should follow the general slope of the points.
3. Balance the points: Try to have an approximately equal number of points above and below the line.
4. Pass through the mean point: If you are given or calculate the mean of \(x\) (written as \(\bar{x}\)) and the mean of \(y\) (written as \(\bar{y}\)), your line must pass directly through the coordinate \((\bar{x}, \bar{y})\).
5. Do not force it through the origin \((0, 0)\): Only pass through \((0,0)\) if the trend naturally goes there.
Quick Takeaway: A line of best fit does not need to touch every point (or even any point). Its goal is to capture the central trend of all the data.
---4. Using the Line of Best Fit (Making Predictions)
Once your line of best fit is drawn, you can use it to estimate values by drawing dashed guide lines from one axis to the line and then across to the other axis.
Interpolation vs. Extrapolation:
• Interpolation (Reliable):
Estimating a value inside the range of the given data points.
Example: If your data covers temperatures from \(10^\circ\text{C}\) to \(25^\circ\text{C}\), estimating sales at \(18^\circ\text{C}\) is interpolation. This is generally reliable because we have observed data in this region.
• Extrapolation (Unreliable / Risky):
Estimating a value outside the range of the given data points (extending your line beyond the data).
Example: Using the same data to predict sales at \(40^\circ\text{C}\). This is unreliable because the trend might change (e.g., people might stay indoors because it is too hot).
Exam Tip: If an exam question asks: "Is your estimate reliable? Give a reason," check if the value is within the original data range. If it is outside, write: "No, because it requires extrapolation outside the range of given data."
---5. Correlation vs. Causation
One of the most important concepts in data handling is understanding that correlation does not imply causation.
Just because two variables show a strong correlation does not automatically mean that one variable causes the other to change.
A Classic Example:
There is a strong positive correlation between ice cream sales and the number of sunburn cases.
• Does eating ice cream cause sunburn? No!
• Does having sunburn make you buy ice cream? No!
• There is a third factor (a common cause): hot sunny weather increases both ice cream sales and sunburn cases independently.
Quick Takeaway: Two variables may show a strong statistical link purely by coincidence or because of a hidden third variable. Always think about whether a direct cause makes logical sense!
---6. Common Mistakes to Avoid
• Joining the dots: Never connect plotted points with line segments. Always draw a single straight line of best fit.
• Forcing the line through \((0, 0)\): Unless the data genuinely points to \((0, 0)\), do not drag your ruler to the bottom-left corner.
• Vague descriptions: Avoid writing just "it's positive." Write: "There is a positive correlation: as hours studied increase, test scores increase."
• Ignoring scale intervals: Take your time when reading axes. Check how much each small square represents (e.g., \(1\) square \(= 0.2\), \(1\), \(2\), or \(5\) units).
Quick Summary Checklist
Before moving on, make sure you can:
• Correctly plot points on a scatter diagram using accurate scales
• Identify positive, negative, and zero correlation
• Draw a balanced straight line of best fit with a ruler
• Pass the line through the mean point \((\bar{x}, \bar{y})\) if calculated
• Read values accurately using interpolation
• Explain why extrapolation can lead to unreliable predictions
• Explain clearly why correlation does not always mean one thing causes another