Correlations | AQA A-Level Psychology Revision

For 7182 specification, first teach in September 2025

AQA A-Level Psychology | Free Revision Notes

Estimated study time: 45 minutes

These Correlations A-Level Psychology revision notes explain how psychologists analyse relationships between two measured co-variables. You will learn to distinguish positive, negative and zero correlations and understand why a correlation cannot establish cause and effect. Correlations may use measurements collected through written self-report questions, researcher-led self-reports or other techniques. You will also compare correlational research with investigations that manipulate an independent variable.

Learning Objectives 🎯

By the end of this revision page, you should be able to:

• Define a correlation.

• Explain what co-variables are.

• Describe how relationships between co-variables are analysed.

• Distinguish positive, negative and zero correlations.

• Interpret the direction of a relationship from a description or set of scores.

• Explain the difference between a correlation and an experiment.

• Explain why correlations cannot establish cause and effect.

• Apply correlational terminology to unfamiliar research scenarios.

Revision Notes 📚

Correlations A-Level Psychology revision overview

A correlation is a statistical relationship between two measured variables.

In correlational research, psychologists measure two variables for each participant and examine whether the variables are associated.

The two measured variables are called co-variables.

For example, a psychologist could measure:

• The number of hours each student revises.

• Each student’s score on a psychology test.

The researcher could then investigate whether students who revise for more hours also tend to achieve higher test scores.

The basic correlational question is:

As one co-variable changes, does the other co-variable tend to change in a systematic way?

A correlation can be:

• Positive.

• Negative.

• Zero.

Specification boundary

The AQA specification requires students to understand:

• The analysis of relationships between co-variables.

• The difference between correlations and experiments.

• Positive, negative and zero correlations.

Detailed plotting of scattergrams and analysis of correlation coefficients are developed separately in representing and measuring correlational relationships.

This lesson focuses on the nature and direction of relationships rather than detailed numerical calculation.

What is a co-variable?

A co-variable is one of the two variables measured in a correlational study.

The term is used because neither variable is deliberately manipulated by the researcher.

For example, a psychologist investigates the relationship between:

• Hours of sleep.

• Score on an attention test.

The co-variables are:

1. Number of hours slept.

2. Attention-test score.

Both are measured for every participant.

Co-variables are paired measurements

Each participant must provide a score for both co-variables.

For example:

Participant	Hours of sleep	Attention-test score
A	5	12
B	6	15
C	7	18
D	8	21
E	9	24

The researcher examines whether the pairs of scores follow a pattern.

In this example, participants who report more sleep also tend to receive higher attention scores.

This suggests a positive correlation.

Co-variables are not the same as an IV and DV

Experiments use:

• An independent variable, or IV.

• A dependent variable, or DV.

Correlations use:

• Co-variable 1.

• Co-variable 2.

Correlation	Experiment
Two co-variables are measured	An IV is manipulated and a DV is measured
Neither variable is deliberately changed	Researcher creates different IV conditions
Looks for an association	Investigates whether the IV affects the DV
Cannot establish cause and effect	May support a causal conclusion if well controlled

It is incorrect to label one correlational variable the IV simply because it is listed first.

Measurement rather than manipulation

In a correlational study, the researcher does not decide which value of a co-variable each participant will experience.

For example, the researcher may measure:

• Participants’ existing amount of social-media use.

• Participants’ existing anxiety scores.

The researcher does not instruct one group to use social media for one hour and another group to use it for five hours.

If the researcher deliberately created different levels of social-media use, the study would move towards an experiment.

Sources of correlational data

Co-variable measurements may be collected using several techniques.

Questionnaires

A questionnaire might measure:

• Hours of revision.

• Anxiety rating.

• Frequency of exercise.

• Attitude score.

Interviews

An interviewer might ask participants to estimate:

• Average sleep duration.

• Number of social interactions.

• Time spent using a particular service.

Psychological tests

Researchers may use:

• Memory scores.

• Attention scores.

• Language scores.

• Problem-solving scores.

Behavioural records

Researchers might use:

• Attendance.

• Number of absences.

• Examination results.

• Frequency of recorded behaviour.

Existing records

A correlational analysis could use information already collected by:

• Schools.

• Employers.

• Healthcare services.

• Researchers conducting an earlier study.

The use of data that already exist is explored further in Primary and secondary data.

Operationalising co-variables

Both co-variables must be operationalised.

Operationalisation means defining exactly how a variable will be measured.

A vague variable might be:

Academic success.

A clear operationalisation might be:

The participant’s percentage score on their most recent psychology assessment.

Another vague variable might be:

Revision effort.

A clear operationalisation might be:

The number of hours the participant reports spending on psychology revision during the previous seven days.

Clear operationalisation helps researchers:

• Collect consistent data.

• Replicate the study.

• Interpret the relationship.

• Decide whether the measures are valid.

This process is developed in Variables and operationalisation.

Example of a correlational aim

A suitable aim might be:

To investigate whether there is a relationship between the number of hours students spend revising psychology and their score on a psychology test.

This aim identifies:

• Co-variable 1: hours of revision.

• Co-variable 2: test score.

• The relationship being investigated.

It does not claim that revision necessarily causes higher scores.

Correlational hypotheses

A correlational hypothesis predicts a relationship between two co-variables.

For example:

There will be a positive correlation between the number of hours students spend revising psychology and their psychology-test scores.

A non-directional correlational hypothesis might state:

There will be a correlation between the number of hours students spend revising psychology and their psychology-test scores.

A correlational hypothesis should not predict a difference between conditions, because a correlation does not contain experimental conditions.

The construction of hypotheses is covered in directional and non-directional predictions.

Analysing relationships between co-variables

The researcher collects two scores from each participant.

The process can be summarised as:

1. Operationalise both co-variables.

2. Measure both variables for each participant.

3. Keep each participant’s pair of scores together.

4. examine whether changes in one variable are associated with changes in the other.

5. Identify the direction of the relationship.

6. Avoid treating the association as proof of causation.

The relationship may also be displayed using a scattergram.

Scattergrams

A scattergram, also called a scattergraph, displays the relationship between two co-variables.

Each participant is represented by one point.

• One co-variable is shown on the horizontal axis.

• The other co-variable is shown on the vertical axis.

• The participant’s two scores determine the position of the point.

The overall pattern of points indicates whether the relationship is:

• Positive.

• Negative.

• Zero.

Detailed construction and interpretation are covered in Scattergrams and correlation coefficients.

Correlation describes a pattern

A correlation describes how two sets of scores vary together.

It does not mean that every participant follows the pattern perfectly.

For example, a positive correlation between revision and test performance does not mean that:

• Every student who revises more receives a higher score.

• All students who revise for the same amount of time achieve the same mark.

• Revision is the only factor related to test performance.

Instead, it means there is an overall tendency for higher values of one co-variable to occur with higher values of the other.

Positive correlations

What is a positive correlation?

A positive correlation occurs when the two co-variables change in the same direction.

As one co-variable increases, the other tends to increase.

As one decreases, the other tends to decrease.

The basic pattern is:

Higher values of co-variable 1 are associated with higher values of co-variable 2.

Positive correlation example

A psychologist measures:

• Hours spent revising.

• Psychology-test score.

Suppose the data show that students who revise for more hours generally achieve higher scores.

This is a positive correlation.

Revision time	Test score
1 hour	42%
2 hours	49%
4 hours	61%
6 hours	73%
8 hours	82%

Both co-variables increase together.

Positive does not mean beneficial

The term positive describes the direction of the relationship.

It does not mean that the relationship is desirable, healthy or beneficial.

For example, there could be a positive correlation between:

• Stress and sleep difficulties.

• Number of cigarettes smoked and breathing problems.

• Time spent waiting and customer dissatisfaction.

In each example, higher values of one variable are associated with higher values of the other.

The word positive does not evaluate whether the outcome is good.

Positive does not mean causal

A positive correlation between revision and examination scores does not prove that revision caused the higher scores.

Possible explanations include:

• Revision may improve performance.

• Students who perform well may feel more motivated to revise.

• A third variable may influence both revision and performance.

For example, academic motivation could lead students to revise more and also to work more carefully in lessons.

Positive relationship in words

Look for descriptions such as:

• The more X, the more Y.

• Higher X is associated with higher Y.

• As X increases, Y also tends to increase.

• Lower X is associated with lower Y.

• The two variables move in the same direction.

These descriptions indicate a positive correlation.

Applying positive correlation

Consider this scenario:

A psychologist finds that pupils with higher attendance percentages tend to receive higher assessment scores.

This is a positive correlation because:

• Higher attendance is associated with higher assessment scores.

• The co-variables change in the same direction.

The finding does not show that attendance alone caused the higher scores.

Negative correlations

What is a negative correlation?

A negative correlation occurs when the two co-variables change in opposite directions.

As one co-variable increases, the other tends to decrease.

The basic pattern is:

Higher values of co-variable 1 are associated with lower values of co-variable 2.

Negative correlation example

A psychologist measures:

• Number of hours spent using a phone after going to bed.

• Number of hours slept.

Suppose participants who spend longer using their phones tend to report fewer hours of sleep.

Phone use after bedtime	Sleep duration
0 hours	8.5 hours
0.5 hours	8 hours
1 hour	7 hours
2 hours	6 hours
3 hours	5 hours

As phone use increases, sleep duration decreases.

This is a negative correlation.

Negative does not mean harmful

The word negative describes the direction of the relationship.

It does not mean that the relationship is necessarily harmful or undesirable.

For example, there might be a negative correlation between:

• Revision time and number of errors.

• Exercise frequency and resting heart rate.

• Practice and time needed to complete a task.

In these examples, higher values of one co-variable are associated with lower values of the other.

The outcome may be helpful even though the correlation is described as negative.

Negative does not mean no relationship

A negative correlation is a systematic relationship.

The co-variables are associated, but they move in opposite directions.

This is different from a zero correlation, where no clear relationship is identified.

Negative does not establish causation

Suppose there is a negative correlation between examination confidence and anxiety.

Possible explanations include:

• Greater confidence may reduce anxiety.

• Greater anxiety may reduce confidence.

• Preparation may increase confidence and reduce anxiety.

• Previous examination experiences may affect both.

The correlation identifies a relationship but cannot determine the causal explanation.

Negative relationship in words

Look for descriptions such as:

• The more X, the less Y.

• Higher X is associated with lower Y.

• As X increases, Y tends to decrease.

• Lower X is associated with higher Y.

• The variables move in opposite directions.

These descriptions indicate a negative correlation.

Applying negative correlation

Consider this scenario:

Students who report more hours of sleep tend to make fewer errors on an attention task.

This is a negative correlation because:

• Sleep duration increases.

• Number of errors decreases.

• The co-variables change in opposite directions.

The result does not prove that increasing sleep will necessarily reduce errors.

Zero correlations

What is a zero correlation?

A zero correlation occurs when there is no clear relationship between the two co-variables.

Knowing a participant’s score on one variable does not allow the researcher to predict whether their score on the other is likely to be high or low.

The scores do not show a consistent positive or negative pattern.

Zero correlation example

A psychologist measures:

• Shoe size.

• Score on a psychology vocabulary test.

Suppose people with larger shoe sizes are not consistently more or less successful on the test.

Shoe size	Vocabulary score
4	18
5	12
6	19
7	14
8	17
9	11

The scores show no clear pattern.

This suggests a zero correlation.

Zero does not mean identical scores

A zero correlation does not mean that every participant receives the same score.

The scores may vary considerably.

The important point is that the variation in one co-variable is not systematically associated with variation in the other.

Zero does not prove that the variables can never be related

A zero correlation in one investigation applies to:

• The sample studied.

• The measures used.

• The time at which the data were collected.

• The particular range of scores.

Another investigation might produce a different result if it used:

• A larger sample.

• A more valid measure.

• A wider range of participants.

• A different context.

Researchers should avoid claiming that a zero correlation proves that no relationship could ever exist.

Zero relationship in words

Look for descriptions such as:

• There is no clear association.

• High and low values of X occur with both high and low values of Y.

• Knowing X does not help predict Y.

• The variables do not follow a consistent pattern.

• There is neither a positive nor a negative relationship.

Applying zero correlation

Consider this scenario:

A researcher finds that the number of siblings a student has is not consistently associated with their score on a visual-memory task.

This is a zero correlation because:

• Students with many siblings may receive high or low scores.

• Students with few siblings may also receive high or low scores.

• No clear direction is present.

Comparing positive, negative and zero correlations

Type of correlation	Pattern
Positive	As one co-variable increases, the other tends to increase
Negative	As one co-variable increases, the other tends to decrease
Zero	There is no clear relationship between the co-variables

Quick identification method

Ask:

What tends to happen to the second co-variable when the first increases?

• It also increases: positive correlation

• It decreases: negative correlation

• It shows no consistent change: zero correlation

Direction is not strength

The direction of a correlation tells the researcher whether the variables move:

• Together.

• In opposite directions.

• Without a consistent relationship.

Direction does not by itself show how closely the scores follow the pattern.

The detailed analysis of strength is covered in Scattergrams and correlation coefficients.

Correlations and experiments

What is an experiment?

An experiment investigates whether manipulation of an independent variable affects a dependent variable.

For example, a researcher might investigate whether sleep duration affects attention.

The researcher could:

1. Create different sleep conditions.

2. Allocate participants to the conditions.

3. Measure attention performance.

4. Control relevant extraneous variables.

5. Compare the conditions.

This differs fundamentally from measuring participants’ usual sleep and attention scores.

Correlational version of a study

A correlational study might investigate sleep and attention by:

• Asking each participant how many hours they slept.

• Giving each participant an attention test.

• Examining the relationship between sleep and attention scores.

The researcher does not manipulate sleep.

Experimental version of a study

An experimental study might:

• Assign participants to different sleep conditions.

• Keep other procedures as consistent as possible.

• Measure attention after the manipulation.

• Compare attention scores between conditions.

The researcher creates the difference in sleep duration.

Direct comparison

Feature	Correlation	Experiment
Variables	Two co-variables	IV and DV
Manipulation	No variable is manipulated	Researcher manipulates the IV
Conditions	No experimental conditions are required	At least two IV conditions
Allocation	Participants are not allocated to levels of measured co-variables	Participants may be allocated to conditions
Main purpose	Identify an association	Investigate a possible causal effect
Conclusion	Variables are related or unrelated	IV may have affected the DV
Cause and effect	Cannot be established	May be established with sufficient control
Third variables	Often difficult to eliminate	Can be controlled more effectively
Directionality	Causal direction remains unclear	Manipulation helps establish direction

Correlations use co-variables

A correlational description should use terms such as:

• Co-variable.

• Relationship.

• Association.

• Positive.

• Negative.

• Zero.

An experimental description should use terms such as:

• Independent variable.

• Dependent variable.

• Manipulation.

• Condition.

• Cause and effect.

Using the correct terminology shows that the research methods have been distinguished.

Why correlations cannot establish cause and effect

A correlation shows that two variables are associated.

It does not show:

• Which variable causes the other.

• Whether either variable causes the other.

• Whether a third variable produces the relationship.

There are two central problems:

1. The directionality problem.

2. The third-variable problem.

The directionality problem

What is directionality?

The directionality problem occurs because a correlation does not show which co-variable influences the other.

Suppose there is a positive correlation between:

• Examination anxiety.

• Time spent worrying about examinations.

Possible explanations include:

• Anxiety causes more worrying.

• Worrying increases anxiety.

• The two influence each other.

• Neither causes the other directly.

The correlation itself cannot decide between these explanations.

Direction of correlation is different from causal direction

This distinction is crucial.

Direction of the correlation

Positive, negative or zero.

Direction of causation

Whether X causes Y, Y causes X, both influence each other or another variable affects both.

A negative correlation still has an uncertain causal direction.

For example, if exercise is negatively correlated with stress:

• Exercise might reduce stress.

• People with lower stress might exercise more.

• Another factor might influence both.

Example of directionality

A researcher finds a negative correlation between loneliness and number of social activities.

Possible causal explanations include:

• Loneliness causes people to avoid activities.

• Fewer social activities increase loneliness.

• Both processes occur.

• A third variable affects both.

The correlational result cannot establish which explanation is correct.

The third-variable problem

What is a third variable?

A third variable is an unmeasured factor that may influence both co-variables.

Suppose there is a positive correlation between:

• Number of hours spent revising.

• Examination score.

Possible third variables include:

• Motivation.

• Previous attainment.

• Attendance.

• Access to resources.

• Teacher support.

• Interest in the subject.

Motivation, for example, might cause students to revise more and also to work more effectively during lessons.

The observed relationship could therefore be partly or entirely explained by motivation.

Another third-variable example

A researcher finds a positive correlation between ice-cream sales and reports of sunburn.

It would be inaccurate to conclude that buying ice cream causes sunburn.

A likely third variable is hot, sunny weather.

Hot weather may:

• Increase ice-cream purchases.

• Increase exposure to sunlight.

• Increase the likelihood of sunburn.

The correlation arises because both measured variables are associated with a third factor.

Controlling third variables

Researchers may attempt to measure or control possible third variables.

For example, in research on revision and test scores, they might also measure:

• Previous test performance.

• Attendance.

• Motivation.

• Access to revision resources.

However, it may be impossible to identify or control every relevant factor.

The inability to eliminate alternative explanations is why correlation alone cannot prove causation.

Strengths of correlations

Strength: variables can be studied without manipulation

Correlations allow researchers to investigate variables that would be difficult, impossible or unethical to manipulate.

Examples include:

• Age.

• Existing anxiety.

• Long-term life experience.

• Naturally occurring sleep habits.

• Existing brain damage.

• Family circumstances.

A researcher should not deliberately create serious anxiety or harmful deprivation simply to test its effects.

Correlational measurement allows the relationship to be studied without imposing the variable.

Strength: useful when an experiment is impractical

Some variables develop over long periods.

For example, a psychologist might investigate the relationship between:

• Years of musical practice.

• Performance on an auditory task.

Randomly allocating participants to complete different amounts of musical practice for many years would be impractical.

A correlation allows existing variation to be studied.

Strength: identifies relationships

A correlation can show whether variables are associated.

This may help psychologists:

• Identify patterns.

• Make predictions.

• Develop hypotheses.

• Select questions for later research.

• Decide whether an experimental investigation is worthwhile.

For example, a relationship between sleep and concentration could encourage researchers to investigate the issue experimentally.

Strength: can use naturally occurring data

Correlational studies can use information that already exists.

For example:

• Attendance records.

• Assessment scores.

• Medical records.

• Behavioural frequencies.

• Previously collected questionnaire data.

This may reduce the time and cost required to collect new data.

Researchers must still consider whether the measures are valid and whether using the records is ethically acceptable.

Strength: may have real-world relevance

Participants can be measured as they naturally differ rather than being placed into artificial conditions.

For example, researchers may investigate people’s actual:

• Sleep duration.

• Exercise frequency.

• Study habits.

• Social-media use.

This may increase the relevance of the findings to everyday behaviour.

However, natural measurements may be less controlled and may rely on inaccurate self-reports.

Strength: starting point for further research

Correlational findings can guide later investigations.

The sequence may be:

1. A correlation identifies a relationship.

2. Researchers propose possible explanations.

3. A hypothesis is developed.

4. An experiment tests one causal explanation where ethical and practical.

5. Findings from several methods are compared.

A correlation is therefore valuable even though it does not prove causation.

Limitations of correlations

Limitation: no cause and effect

The central limitation is that correlation does not establish causation.

Even a clear relationship does not show:

• Which variable is the cause.

• Whether a third variable is responsible.

• Whether the relationship is accidental.

• Whether the same pattern would occur after intervention.

Researchers should use cautious language such as:

• “Is associated with.”

• “Is related to.”

• “Tends to occur alongside.”

They should avoid causal claims such as:

• “Produces.”

• “Leads to.”

• “Results in.”

unless supported by additional evidence.

Limitation: third variables

Uncontrolled factors may produce or influence the relationship.

This weakens the ability to explain why the variables are associated.

The researcher may measure several possible third variables, but an unmeasured factor may remain.

Limitation: directionality

Researchers cannot determine which co-variable influences the other.

This makes it difficult to use the relationship as evidence for one particular psychological explanation.

Limitation: inaccurate measurement

A correlation is only as valid as the measures used.

For example, a researcher may ask:

How many hours do you usually revise?

Participants may:

• Misremember.

• Exaggerate.

• Interpret “revise” differently.

• Give socially desirable answers.

• Estimate rather than record their time.

An apparently clear correlation may be misleading if one or both co-variables were measured poorly.

Limitation: restricted range

If participants have very similar scores on a co-variable, it may be difficult to identify a relationship.

For example, a study of sleep and attention might include only people who sleep between seven and eight hours.

The small amount of variation in sleep may prevent the wider relationship from becoming visible.

Researchers need an appropriate range of scores.

Limitation: unusual scores

An unusually high or low score may affect the overall pattern.

For example, one participant might:

• Report 15 hours of revision.

• Receive a very low test score.

Researchers should inspect the data carefully rather than relying on a general impression.

The detailed treatment of scattergrams and unusual scores belongs to Scattergrams and correlation coefficients.

Limitation: oversimplification

Human behaviour is often influenced by many interacting factors.

A correlation between two variables may create an overly simple picture.

For example, examination performance may be related to:

• Revision.

• Attendance.

• Prior knowledge.

• Sleep.

• Motivation.

• Anxiety.

• Teaching.

• Resources.

• Examination technique.

Focusing on only two co-variables may overlook this complexity.

Correlation does not mean causation

Safe interpretation

Suppose a study finds a positive correlation between exercise and wellbeing.

A safe conclusion is:

Higher exercise frequency was associated with higher wellbeing scores in this sample.

This conclusion describes the relationship found.

Unsafe interpretation

An unsafe conclusion is:

Exercise caused participants to have higher wellbeing.

This conclusion is not justified because:

• Wellbeing might influence exercise.

• A third variable might influence both.

• Exercise was not manipulated.

• Alternative explanations were not controlled experimentally.

Strong correlation does not solve causation

A relationship may appear very consistent, but it still does not prove causation.

The directionality and third-variable problems remain.

The clarity of an association and the ability to explain its cause are separate issues.

Classifying research scenarios

Scenario 1: correlation

A psychologist records each student’s weekly revision time and examination score, then examines whether the scores are related.

This is a correlation because:

• Two variables are measured.

• Neither is manipulated.

• Each student provides paired scores.

• The researcher analyses a relationship.

Scenario 2: experiment

A psychologist allocates students to complete either one hour or three hours of revision and then gives all students the same test.

This is an experiment because:

• The researcher manipulates revision time.

• There are different IV conditions.

• Test performance is the DV.

• The researcher investigates the effect of the IV.

Scenario 3: correlation using questionnaire data

Participants complete one scale measuring loneliness and another scale measuring social confidence. The researcher examines whether the two scores are related.

This is a correlation because both co-variables are measured.

The use of questionnaires does not change the overall correlational design.

Scenario 4: experiment using questionnaire data

Participants complete either an individual or group task. They then complete a questionnaire measuring enjoyment.

This is an experiment because:

• Task type is manipulated.

• Enjoyment is measured as the DV.

• The questionnaire is the technique used to collect the DV data.

Scenario 5: quasi-experiment rather than correlation

A researcher compares the memory scores of younger and older adults.

This may be a quasi-experiment because:

• Participants are divided into pre-existing age groups.

• Memory scores are compared between groups.

A correlational version would measure each participant’s exact age and memory score and analyse the relationship between the two co-variables.

This distinction is explored in Natural and quasi-experiments.

Correlation or experiment decision method

Question 1: Did the researcher manipulate a variable?

• Yes: likely an experiment.

• No: continue.

Question 2: Were two variables measured for each participant?

• Yes: the study may be correlational.

• No: consider another method.

Question 3: Was the relationship between the paired scores analysed?

• Yes: correlation.

• No: it may be a group comparison or descriptive study.

Question 4: Does the conclusion use causal language?

If the study is correlational, causal language is not justified.

Designing a correlational study

Step 1: identify the co-variables

Both co-variables should be clearly stated.

Example:

• Daily social-media use in minutes.

• Score on a loneliness scale.

Step 2: operationalise both co-variables

State precisely how each will be measured.

For example:

• Social-media use: average daily minutes recorded by the participant’s phone during the previous seven days.

• Loneliness: total score on a specified questionnaire scale.

Step 3: select an appropriate sample

The sample should include enough variation in both co-variables.

A very restricted sample may make the relationship difficult to detect.

Sampling and generalisation are examined in Populations and samples.

Step 4: collect paired scores

Each participant must provide both measurements.

A participant with only one score cannot contribute a complete pair to the correlational analysis.

Step 5: standardise the procedure

Researchers should keep constant:

• Instructions.

• Testing conditions.

• Time limits.

• Measurement tools.

• Scoring procedures.

• Information given to participants.

This improves consistency.

Step 6: consider validity

Ask whether the measures represent the intended variables.

For example:

• Does a single test accurately measure attention?

• Does self-reported phone use reflect actual usage?

• Does the questionnaire measure loneliness rather than general unhappiness?

Measurement validity is explored in Validity.

Step 7: consider ethics

Researchers should address:

• Informed consent.

• Confidentiality.

• Right to withdraw.

• Protection from harm.

• Sensitive personal information.

• Secure storage of paired data.

The association between two records could reveal information participants consider private.

Step 8: interpret cautiously

The final conclusion should describe:

• The direction of the relationship.

• The variables measured.

• The sample studied.

• The limits on causal interpretation.

Ethical issues in correlational research

Sensitive co-variables

Researchers may investigate relationships involving:

• Mental health.

• Family income.

• Academic performance.

• Health behaviours.

• Diagnoses.

• Personal relationships.

Combining two measurements can make participants more identifiable or reveal sensitive patterns.

Confidentiality

Researchers should:

• Use participant codes.

• Remove names from data files.

• Store identifying information separately.

• Report group-level findings.

• Avoid publishing identifiable combinations of details.

• Limit access to raw records.

Informed consent

Participants should understand:

• What information will be collected.

• Whether existing records will be accessed.

• How the variables will be linked.

• How confidentiality will be protected.

• Their right to withdraw where possible.

Consent to provide one measure does not automatically mean consent for it to be linked with another dataset.

Protection from harm

Questions or results may cause distress.

For example, a participant may become concerned after receiving a high anxiety score.

Researchers should decide:

• Whether individual results will be shared.

• How potentially upsetting questions will be handled.

• What support information is appropriate.

• Whether the investigation creates unnecessary risk.

Reliability in correlational research

A correlational finding depends on reliable measurement of both co-variables.

Suppose:

• Revision time is measured inconsistently.

• Test performance is measured reliably.

The relationship may still be unreliable because one part of each score pair is unstable.

Researchers can improve reliability by:

• Standardising instructions.

• Using clear questions.

• Using consistent scoring.

• Checking measurement procedures.

• Piloting the study.

• Repeating measurements where appropriate.

The wider principles are covered in Reliability.

Validity in correlational research

A correlation may be calculated accurately but still have low validity.

For example, the researcher may claim to measure stress using:

The number of times a participant checks their phone.

Phone checking may be influenced by:

• Boredom.

• Notifications.

• Habit.

• Work demands.

• Social expectations.

It may not be a valid measure of stress.

The relationship identified would then be difficult to interpret.

Both co-variables must represent the concepts the researcher claims to study.

Writing an effective explanation of correlation

A strong explanation might state:

A correlation analyses the relationship between two co-variables. Both variables are measured for each participant, and the paired scores are examined to identify whether they change together. A positive correlation occurs when higher values of one co-variable are associated with higher values of the other. A negative correlation occurs when higher values of one are associated with lower values of the other. A zero correlation occurs when there is no clear relationship.

This answer:

• Defines correlation.

• Uses the correct term co-variable.

• Explains paired scores.

• Distinguishes all three directions.

Writing an effective comparison with experiments

A strong comparison might state:

In a correlation, the researcher measures two co-variables and analyses whether they are associated. Neither co-variable is deliberately manipulated. In an experiment, the researcher manipulates an independent variable and measures its effect on a dependent variable. Experimental control may allow cause-and-effect conclusions, whereas a correlation cannot establish causation because the causal direction is unclear and a third variable may explain the relationship.

This answer:

• Compares the variables used.

• Compares manipulation.

• Compares the purpose.

• Explains the difference in causal conclusions.

Writing an effective application answer

Scenario:

A researcher finds that students who report spending more time on social media tend to report greater difficulty concentrating.

A developed answer might state:

This is a positive correlation because higher social-media use is associated with greater reported concentration difficulty. However, the finding does not establish that social-media use caused the difficulty. Students who find concentration difficult may use social media more frequently, or a third variable such as poor sleep may increase both social-media use and concentration problems.

The answer:

• Identifies the direction.

• Applies both co-variables.

• Avoids causal language.

• Explains directionality.

• Identifies a plausible third variable.

Overall summary

A correlation examines the relationship between two co-variables.

The researcher:

• Measures both variables.

• Collects paired scores.

• Analyses whether the scores change together.

• Does not manipulate an IV.

• Cannot establish cause and effect.

A positive correlation means:

• Higher values of one co-variable are associated with higher values of the other.

A negative correlation means:

• Higher values of one co-variable are associated with lower values of the other.

A zero correlation means:

• No clear relationship is present.

Correlations are useful because they:

• Investigate variables that cannot be manipulated.

• Identify meaningful relationships.

• Use naturally occurring measurements.

• Generate hypotheses for later research.

Their central limitations are:

• The directionality problem.

• The third-variable problem.

• Dependence on valid and reliable measurement.

• Inability to establish cause and effect.

Key Words 🔑

Key word	Student-friendly definition	How it may be used in an exam
Correlation	A statistical relationship between two measured variables.	Define the method or identify a correlational investigation.
Co-variable	One of the two variables measured in a correlation.	Identify the paired measurements used in a study.
Paired scores	The two measurements collected from the same participant.	Explain how correlational data are organised.
Positive correlation	A relationship in which the co-variables tend to change in the same direction.	Interpret a pattern where both variables increase together.
Negative correlation	A relationship in which one co-variable tends to increase as the other decreases.	Interpret a pattern where variables change in opposite directions.
Zero correlation	A result showing no clear relationship between the co-variables.	Identify a pattern with no consistent direction.
Scattergram	A graph displaying paired scores for two co-variables.	Explain how a correlational relationship may be presented.
Independent variable	The variable manipulated by a researcher in an experiment.	Distinguish experimental variables from co-variables.
Dependent variable	The outcome measured after an IV is manipulated.	Distinguish experiments from correlations.
Cause and effect	A relationship in which changing one variable directly produces a change in another.	Explain what a correlation cannot establish.
Directionality problem	Uncertainty about which co-variable, if either, influences the other.	Evaluate causal claims based on correlation.
Third variable	An unmeasured factor that may influence both co-variables.	Suggest an alternative explanation for a correlation.
Operationalisation	Defining exactly how a variable will be measured.	Design valid and replicable co-variable measures.
Correlation hypothesis	A prediction about a relationship between two co-variables.	Write a directional or non-directional hypothesis.
Association	A pattern in which two variables are related.	Describe findings without implying causation.
Quantitative data	Numerical data that can be measured or counted.	Identify the form of data required for correlation.
Reliability	The consistency of a measurement or finding.	Evaluate the quality of co-variable measurements.
Validity	The extent to which a measure represents what it claims to measure.	Evaluate the interpretation of a correlation.
Restricted range	A limited spread of scores on a co-variable.	Explain why a relationship may be difficult to detect.
Generalisability	The extent to which findings apply beyond the sample studied.	Evaluate conclusions drawn from a correlational sample.

Common Mistakes ⚠️

Mistake: Calling one correlational variable the independent variable.

Why this is incorrect:Neither variable is manipulated in a correlation. Both are co-variables.

How to improve:Use the terms co-variable 1 and co-variable 2.

Mistake: Saying that a positive correlation is a beneficial relationship.

Why this is incorrect:Positive describes variables changing in the same direction, not whether the outcome is desirable.

How to improve:Describe what happens to one variable as the other increases.

Mistake: Saying that a negative correlation means there is no relationship.

Why this is incorrect:A negative correlation is a systematic relationship in which the variables change in opposite directions.

How to improve:Distinguish negative correlation from zero correlation.

Mistake: Saying that a zero correlation means every participant obtained the same score.

Why this is incorrect:Scores may vary widely, but they do not show a consistent relationship.

How to improve:Focus on the lack of an overall positive or negative pattern.

Mistake: Using the eye-catching word “negative” to claim that the outcome is harmful.

Why this is incorrect:The term describes statistical direction rather than value or desirability.

How to improve:Explain whether the co-variables move together or in opposite directions.

Mistake: Claiming that a correlation proves one variable caused another.

Why this is incorrect:The direction of causation is unknown, and a third variable may explain the relationship.

How to improve:Use “associated with” rather than “caused”.

Mistake: Confusing the direction of a correlation with the direction of causation.

Why this is incorrect:Positive or negative describes the pattern of scores, not which variable causes the other.

How to improve:Discuss directionality separately from positive and negative relationships.

Mistake: Saying that a strong-looking relationship establishes causation.

Why this is incorrect:Directionality and third-variable problems remain even when the pattern is consistent.

How to improve:Separate the clarity of the association from evidence of its cause.

Mistake: Describing two groups as co-variables.

Why this is incorrect:A correlation requires two measurements for each participant rather than simply comparing group averages.

How to improve:Identify the two paired scores recorded for every participant.

Mistake: Calling any study with two variables a correlation.

Why this is incorrect:An experiment also contains two variables, but one is manipulated and the other is measured.

How to improve:Ask whether the researcher manipulated an IV or measured two co-variables.

Mistake: Assuming questionnaire research must be correlational.

Why this is incorrect:A questionnaire is a data-collection technique and can be used in correlations, experiments or descriptive studies.

How to improve:Identify how the variables are used rather than focusing only on how data were collected.

Mistake: Writing a difference hypothesis for a correlation.

Why this is incorrect:Correlational hypotheses predict a relationship between co-variables, not a difference between conditions.

How to improve:Use wording such as “There will be a positive correlation between…”

Mistake: Ignoring operationalisation.

Why this is incorrect:Vague variables such as “success” or “stress” do not explain how scores will be obtained.

How to improve:State the exact questionnaire score, behavioural measure or numerical record used.

Mistake: Giving a third variable without explaining it.

Why this is incomplete:The examiner needs to understand how the third variable could affect both measured variables.

How to improve:Link the proposed factor separately to each co-variable.

Mistake: Interpreting a zero correlation as proof that no relationship could ever exist.

Why this is incorrect:The result applies to the particular sample, measures and range of scores studied.

How to improve:State that no clear relationship was found in that investigation.

Exam-Style Questions ✍️

Question 1

Which one of the following best describes a correlation?

A. A comparison between two experimentally created conditions

B. An analysis of the relationship between two measured co-variables

C. A study in which participants are randomly allocated to groups

D. An observation of behaviour in a natural setting

[1 mark]

Question 2

Define the term co-variable.

[2 marks]

Question 3

A psychologist measures the number of hours students revise and their scores on a psychology test.

Identify the two co-variables.

[2 marks]

Question 4

Explain the difference between a positive correlation and a negative correlation.

[4 marks]

Question 5

A researcher finds that students who sleep for more hours tend to make fewer errors on an attention task.

Identify the direction of the correlation and explain your answer.

[3 marks]

Question 6

A psychologist investigates the relationship between shoe size and score on a memory test. The results show no consistent pattern.

Explain what is meant by a zero correlation in this investigation.

[3 marks]

Question 7

A researcher finds a positive correlation between weekly revision time and examination performance.

Explain why the researcher cannot conclude that revision time caused higher examination performance.

[4 marks]

Question 8

A psychologist investigates the relationship between social-media use and anxiety.

Write a suitable directional correlational hypothesis for this investigation.

[3 marks]

Question 9

A researcher records participants’ usual sleep duration and attention scores. Another researcher allocates participants to different sleep-duration conditions and then measures attention.

Explain one difference between the two investigations.

[4 marks]

Question 10

Explain how correlations are used to analyse relationships between co-variables.

Refer to positive, negative and zero correlations and distinguish correlations from experiments in your answer.

[8 marks]