In AQA A-Level Psychology (Paper 2: Research Methods, Section C), you must know the difference between a correlation (association) and a difference (experiment).
If you confuse correlation with difference — or mix up association vs correlation — you risk losing marks in hypothesis questions and when choosing the correct statistical test.
AQA Specification Requirements
This comes directly from the specification, which requires students to:
- “Understand the differences between experiments and correlations”
- “Be able to write hypotheses (null, directional, non-directional)”
- “Understand the use of inferential statistical tests, including tests for association/correlation and difference”
1. Correlation
A correlation investigates whether two co-variables are related.
- Both are measured — neither is manipulated
- Produces a correlation coefficient (r) between –1 and +1
- Positive: as one increases, so does the other
- Negative: as one increases, the other decreases
- Zero: no consistent relationship
“There will be a significant positive correlation between the number of hours students revise per week and their exam score (percentage out of 100).”
2. Association vs Correlation
These terms are related but not identical. AQA sometimes swaps them in exam papers — this can trap students.
| Term | Data Type | Statistical Test | Example |
|---|---|---|---|
| Association | Nominal (categorical) | Chi-squared | Gender and subject choice |
| Correlation | Ordinal/Interval (numeric) | Spearman’s rho / Pearson’s r | Hours of exercise and stress ratings |
| Relationship | Any (safe catch-all) | Depends on data type | Often credited but may limit marks |
Top Tip: A Subtle But Important Nuance
AQA examiners sometimes interchange these terms, which can catch students off-guard. Understanding this nuance gives you an edge:
- Association: A general term for a link between two variables. Most commonly used with nominal (categorical) data such as gender, subject choice, or yes/no responses. Tested with Chi-squared tests. AQA often requires this specific term when dealing with any categorical data.
- Correlation: The statistical test of a relationship between two ordinal or interval variables (numeric data). This includes ratings, scores, measurements, or rankings. Tested with Spearman’s rho or Pearson’s r. AQA expects this term when both variables are numerical.
- Relationship: A safe, catch-all word that AQA frequently credits in mark schemes regardless of data type. However, using the precise term (association/correlation) often secures full marks while “relationship” may limit you to 2/3 marks.
Example: In Specimen Paper 2, full marks were awarded for using “correlation” between map-reading scores and driver ratings (both numerical variables). Using “association” instead would have lost marks because both variables were numeric.
In other cases (e.g., Chi-squared tests with nominal data), the mark scheme required “association”.
Pro Strategy: Nominal/categorical data → “association” | Numeric data → “correlation” | Uncertain → “relationship” (safer but may limit marks)
3. Correlation Strength Visualization
Understanding correlation strength is crucial for AQA exams. Here are visual examples:
Strong Positive (r = +0.8)
Moderate Positive (r = +0.5)
Weak/No Correlation (r = +0.1)
4. Correlation Strength Matrix
AQA often shows scatterplot images in multiple choice questions and asks “How strong is this correlation?” Use this guide:
r = ±0.7 to ±1.0
Points tightly clustered around line. Clear predictable pattern.
r = ±0.3 to ±0.7
Clear trend but more scattered points. Useful relationship.
r = 0.0 to ±0.3
Highly scattered or random pattern. Little predictive value.
Perfect Negative
Strong Negative
Moderate Negative
Weak Negative
No Correlation
Weak Positive
Moderate Positive
Strong Positive
Perfect Positive
5. Difference (Experiment)
An experiment tests for a difference between conditions.
- Involves manipulating an independent variable (IV)
- Measuring a dependent variable (DV)
- Hypotheses use the word “difference”
“There will be a significant difference in recall scores (out of 20 words) between students who revise with music and those who revise in silence.”
6. Co-variables vs Variables: Crucial Distinction
This terminology difference shows examiners you truly understand research design:
Correlation Studies: Use “co-variables” because both are measured (neither manipulated)
Experimental Studies: Use “variables” – specifically IV (manipulated) and DV (measured)
Getting this right can secure borderline marks and demonstrates methodological precision.
7. Quick Reference
| Feature | Correlation/Association | Difference (Experiment) |
|---|---|---|
| Variables | Two co-variables (both measured) | IV manipulated, DV measured |
| Aim | Test for a relationship/association | Test for an effect/difference |
| Hypothesis | “There will be a significant correlation/association…” | “There will be a significant difference…” |
| Example | Hours revised vs exam score | Music vs silence recall test |
8. Practice Scenarios
Try to identify whether each scenario is a correlation, association, or difference study. Click the cards to reveal answers!
Scenario 1: (Example)
A researcher measures students’ self-esteem scores (1-50) and their number of social media followers.
Answer: CORRELATION — both variables measured, numeric data → “There will be a significant correlation between self-esteem scores and number of social media followers.”
Scenario 2:
A researcher compares memory performance between participants who drink coffee vs water before a test.
Scenario 3:
A researcher investigates the relationship between personality type (introvert/extrovert) and preferred study location (library/home).
Scenario 4:
A researcher measures anxiety levels (0-100) and exam performance (percentage) in students.
Question 5: Correlation Analysis
Look at this scatterplot showing the relationship between study hours per week and exam scores:
What is the strength and direction of this correlation?
Key Takeaways for AQA Exams
- Correlation = numeric data (ordinal/interval), gives r value
- Association = categorical (nominal) data, tested with Chi-squared
- Relationship = safe wording but may limit marks
- Difference = experimental design (IV → DV)
- Always operationalize variables (e.g., “recall out of 20 words”)
- Mirror the question wording in your hypothesis
Remember: This distinction affects both hypotheses (3/3 vs 2/3 marks) and choosing the right statistical test — easy marks if you get it right!