Scatter Graphs and Correlation — The Relationship Test
Part of Fieldwork Presentation and Evaluation Skills — GCSE Geography
This deep dive covers Scatter Graphs and Correlation — The Relationship Test within Fieldwork Presentation and Evaluation Skills for GCSE Geography. Revise Fieldwork Presentation and Evaluation Skills in Geographical Skills for GCSE Geography with 0 exam-style questions and 20 flashcards. This topic shows up very often in GCSE exams, so students should be able to explain it clearly, not just recognise the term. It is section 5 of 16 in this topic. Use this deep dive to connect the idea to the wider topic before moving on to questions and flashcards.
Topic position
Section 5 of 16
Practice
0 questions
Recall
20 flashcards
📉 Scatter Graphs and Correlation — The Relationship Test
Scatter graphs are designed for one purpose: testing whether a relationship (correlation) exists between two continuous variables. They are the standard presentation technique for any hypothesis of the form "as X increases, does Y also change?"
Drawing a Scatter Graph Correctly
Interpreting Correlation
| Pattern of Points | Type of Correlation | What It Means |
|---|---|---|
| Points slope upward left to right; line of best fit rises | Positive correlation | As X increases, Y also increases |
| Points slope downward left to right; line of best fit falls | Negative correlation | As X increases, Y decreases |
| Points scattered randomly; no clear slope to the line | No correlation | X and Y do not appear to be related |
| Points close to the line of best fit | Strong correlation | The relationship is consistent across data points |
| Points widely scattered around the line | Weak correlation | The relationship exists but there is a lot of variability |
Spearman's Rank Correlation Coefficient (rs)
The scatter graph shows you the direction of a correlation visually. Spearman's rank correlation coefficient gives you a numerical measure of its strength, from −1 (perfect negative) through 0 (no correlation) to +1 (perfect positive). An rs value above 0.7 or below −0.7 is generally considered a strong correlation. Most importantly, you can test whether the correlation is statistically significant — meaning it is unlikely to be caused by chance alone — by comparing your rs value against a critical values table at the 0.05 significance level.
When writing up: "The scatter graph shows a strong negative correlation between pebble size and distance from the cliff (rs = −0.87). This is significant at the 0.05 level, meaning there is less than a 5% probability that this relationship occurred by chance. This supports the hypothesis that attrition during transport progressively reduces pebble size with increasing distance."
The Critical Limitation: Correlation Is Not Causation
A scatter graph and Spearman's rank can only tell you that two variables are associated. They cannot prove that one variable causes the other. There may be a third variable (a confounding variable) that actually causes both. This is a standard evaluation point: "A limitation of the scatter graph is that correlation does not prove causation. A confounding variable — such as wave energy at different points along the transect — might explain both the distance and the pebble size patterns independently."
Quick Check: A student calculates Spearman's rank rs = −0.91 for the relationship between distance from cliff and pebble size. What does this tell them, and what limitation should they acknowledge?
An rs value of −0.91 indicates a very strong negative correlation — as distance from the cliff increases, pebble size decreases. This is likely to be statistically significant at the 0.05 level (a critical value table would confirm this), meaning less than 5% probability this pattern occurred by chance. The limitation to acknowledge is that correlation does not prove causation: while attrition during transport is the most plausible explanation, a confounding variable (such as wave energy or sediment source variation at different points) could explain the pattern instead. The student cannot establish causation from this data alone.