IB Psychology SLTopic 5 — Methods of ResearchPaper 1 & 2Core skill~11 min read
Correlational Studies and Their Limits
A correlation tells you that two things move together. It never tells you that one made the other happen. Almost every mark on this topic comes from holding that line, in both directions: knowing what a correlation can show, and refusing to claim what it cannot.
📘 What you need to know
A correlational study measures two co-variables. Nothing is manipulated, so there is no IV.
Each participant contributes two scores, one for each co-variable.
Scores are plotted on a scattergraph, one point per participant.
Positive: both rise together. Negative: one rises as the other falls. Zero: no pattern.
The correlation coefficient runs from −1 to +1. The sign gives direction, the size gives strength.
A correlation cannot establish cause and effect, because a third variable may be driving both.
It also misses non-linear relationships, where the link is real but curved.
What a correlational study actually does
In an experiment you change one thing and watch another. In a correlational study you change nothing. You take two measurements from every person and ask whether they line up.
Those two measurements are the co-variables, and there is no “first” and “second”. Hours of sleep and exam mark. Screen time and reported loneliness. Age and reaction time. Sometimes both already exist as records; sometimes you collect them yourself with a questionnaire or a test.
The difference in one line
Experiment: manipulate one variable, measure another. Correlation: measure both, manipulate nothing.
Why do it at all? Because for most interesting questions you are not allowed to manipulate anything. You cannot randomly assign teenagers to five hours of sleep a night for a year. A correlation is how psychology studies those questions honestly.
Reading a scattergraph
Each cross is one person: their score on the horizontal variable and their score on the vertical one. The pattern in the crosses is the relationship.
Notice the middle graph slopes downwards. That is still a relationship, and a strong one — “negative” describes direction, never quality.
The correlation coefficient
Eyeballing a scattergraph is fine for direction, but not for strength. The correlation coefficient turns the pattern into a single number between −1 and +1.
The sign tells you the direction. Minus means as one goes up the other goes down.
The size, ignoring the sign, tells you the strength. Closer to 1 means the points sit closer to a straight line.
A value near 0 means no linear relationship at all.
The bands are rough guides used across psychology, not fixed rules. What matters in an exam is that you read the sign and the size separately.
Coefficient
How to describe it
What the scattergraph looks like
+0.85
Strong positive
Points hug an upward line
−0.62
Moderate negative
Clear downward slope, some spread
+0.21
Weak positive
Slight upward tilt in a wide cloud
−0.04
Essentially no relationship
A shapeless scatter
🤔 A mistake worth being careful about
Students often read a value like −0.09 as “a strong negative correlation” because the number looks tidy and the minus sign feels dramatic. It is not. Strip the sign off and you have 0.09, which is almost zero. The rule is simple: cover the sign with your finger to judge strength, then put it back to state direction.
Why a correlation cannot prove cause
Suppose ice cream sales and drowning deaths rise together across a year. Nobody thinks ice cream causes drowning. Something else — hot weather — is pushing both up at the same time. That something else is a third variable, and it is present in every correlation you will ever meet.
There is a second problem. Even where two things really are connected, the correlation cannot tell you which way round it goes. Do lonely people use social media more, or does heavy use make people lonelier? The same scattergraph fits both stories.
🧩 How to evaluate any correlational finding in four lines
State the direction and strength in words, not just the number.
Name a plausible third variable that could be affecting both co-variables.
Point out the direction problem: either variable could be influencing the other.
Say what it is still good for: identifying relationships worth investigating, and making predictions.
Examiners do not want you to dismiss correlations. They want the balance: a correlation is weak evidence for cause and strong evidence that something is worth studying properly. Write both halves.
What else correlations miss
The coefficient measures linear relationships — straight-line ones. Some real relationships are curved. Performance and arousal is the classic case: a bit of pressure helps, too much wrecks it. The relationship is real and strong, but a correlation coefficient reports it as close to zero, because the points rise and then fall.
Always plot the scattergraph. A near-zero coefficient with an obvious arch in the points means the analysis is wrong, not that there is nothing there.
Worked examples
WORKED EXAMPLE
Describe a correlation coefficient
A study of 240 students found a correlation of −0.71 between hours spent on social media per day and self-reported hours of sleep. Describe what this result shows. [3]
Step 1: Read the sign
Negative, so as one goes up the other goes down.
Step 2: Read the size0.71 ignoring the sign = strongStep 3: Put it into the context of the study
Students who spent more hours on social media tended to report fewer hours of sleep.
A strong negative correlation“tended to” is doing real work here — it keeps the claim honest
WORKED EXAMPLE
Challenge a causal conclusion
A newspaper reports the study above with the headline “Social media is stealing teenagers’ sleep”. Explain why this conclusion is not justified. [4]
Step 1: Name the method’s limit
Nothing was manipulated, so no cause and effect can be established.Step 2: Offer a third variable
Anxiety could keep students awake and also drive them to their phones, producing the same pattern.
Step 3: Reverse the arrow
Students who cannot sleep may reach for their phone because they are awake.
Association, not causationfinish with what would be needed: a controlled experiment manipulating screen time
💡 Exam tip
Use the word co-variables, never IV and DV. Writing “the IV was screen time” in a correlational question loses marks.
Describe results in three parts: direction, strength, and what it means for the people in the study.
Always name a specific third variable. “There might be other factors” is vague; “anxiety could raise both” is a proper answer.
Correlations are genuinely useful for prediction. Say so, then say prediction is not explanation.
If given a scattergraph, mention how tightly the points cluster. That is your evidence for strength.
Large samples of existing data are a real strength here — you can study thousands of people cheaply.
⚠️ Common mix-up
Thinking negative means weak. −0.9 is far stronger than +0.2.
Writing “correlation proves”. Correlations support, suggest, or are consistent with. They never prove.
Calling a correlation an experiment. No manipulation means no experiment, whatever the numbers look like.
Confusing “no correlation” with “no relationship”. A curved relationship gives a coefficient near zero.
Assuming a big sample fixes the causation problem. A million people give you a more accurate correlation, and still no cause.
Describing the result without the context. Always say what the two co-variables actually were.
Up next: Case Studies: Depth Over Breadth — what psychology can learn from a single person, and what it cannot.
Want this explained one-to-one?
Book a free session with an experienced IB Psychology tutor and get your trickiest topics made simple.
