Correlations: analysis of the relationship between co-variables. The difference between correlations and experiments. Positive, negative and zero correlations.

What this dot point is asking

AQA wants you to describe correlations, co-variables, types of correlation, scattergrams and how correlations differ from experiments. The exam skill is to interpret a correlation coefficient correctly and to explain, every time, why a correlation cannot establish causation.

Co-variables and types

A correlation differs fundamentally from an experiment in that it does not manipulate anything. Instead of an independent variable that the researcher changes and a dependent variable that they measure, a correlation simply measures two variables (the co-variables) as they naturally occur and looks at how they vary together. The relationship can take three forms. In a positive correlation, the two co-variables increase together, so as one rises the other tends to rise (for example, hours of revision and exam scores). In a negative correlation, the co-variables move in opposite directions, so as one rises the other tends to fall (for example, hours of sleep deprivation and reaction speed). In a zero correlation, there is no consistent relationship between the variables at all. It is important not to confuse a negative correlation (a clear inverse relationship) with a zero correlation (no relationship).

Scattergrams and the difference from experiments

Correlational data are displayed on a scattergram, where each point represents one participant's pair of scores, and the pattern of points shows the direction and strength of the relationship. The relationship is summarised numerically by the correlation coefficient, a value from $-1$ to $+1$. The sign shows the direction (positive or negative) and the size shows the strength: a coefficient close to $+1$ or $-1$ is a strong correlation, one close to $0$ is weak, and exactly $0$ means no correlation. So $+0.9$ is a strong positive relationship, $-0.8$ is a strong negative one, and $+0.1$ is a weak positive one. The single most important evaluative point is that a correlation cannot establish cause and effect. Because no variable is manipulated and extraneous variables are not controlled, a significant correlation does not tell us that one co-variable causes the other; an unmeasured third (intervening) variable may be responsible for both. A correlation between ice-cream sales and drownings, for instance, is explained by a third variable, hot weather. The strengths of correlations are that they can study variables that could not be ethically or practically manipulated, and that they are a useful first step to identify relationships worth testing experimentally.