What a correlation is

1. A correlation looks at whether two things you have measured tend to change together.
2. Neither variable is manipulated, which is why a correlation is not an experiment and has no independent variable.
3. You take a pair of scores from every participant and then look at the pattern across the whole set.

Positive and negative correlations

1. Positive correlation means that as one co-variable goes up, the other tends to go up as well.
2. Negative correlation means that as one co-variable goes up, the other tends to go down.
3. No correlation means the two co-variables show no consistent pattern in either direction.

Scatter diagrams

1. A scatter diagram shows a correlation by putting one co-variable on each axis.
2. Each participant contributes exactly one point, placed where their two scores meet.
3. Points that climb from left to right show a positive correlation, and points that fall from left to right show a negative one.
4. Points scattered with no slope show no correlation, and the closer the points sit to a straight line the stronger the relationship.

Correlation does not show causation

1. A correlation tells you that two variables move together and it tells you nothing else.
2. The relationship may run the opposite way round, so poor sleep could be driving the late phone use rather than the other way about.
3. A third variable you never measured could be producing both, such as worry keeping a pupil awake and on their phone at the same time.
4. Only an experiment, where you set the independent variable yourself, can properly test for cause.

Strengths

1. It reaches relationships you could not create, since variables such as sleep and exam performance can be studied without interfering with anyone's life.
2. It is a good first step, because it is quick and often uses data that already exists, so it shows whether a relationship is worth a full experiment.

Weaknesses

1. Cause stays out of reach, because however convincing the pattern looks the design cannot tell you which variable acted on which, or whether either did.
2. Both co-variables have to become numbers, so something like stress must be scored before it can be plotted, and that score may capture the real thing poorly.

When to choose a correlation

1. Choose it when manipulating either variable would be impossible, impractical or unethical.
2. Choose it when you want to check cheaply whether two variables are related at all before designing something larger.
3. Choose an experiment instead the moment your question is actually about cause.

Turning this into marks

1. Reading a scatter diagram, name the direction together with the two variables.
2. A hypothesis here predicts a relationship between two named, measurable variables, not a difference between two groups.
3. The mark most often lost comes from writing that a correlation proves one thing causes another.