A typical correlation study in Psychology uses a non-experimental source (e.g. a survey) to obtain data which show numerical values on two variables (e.g. IQ and extraversion). Correlation means analysing the
relationship between two such variables (which are called a study's
'co-variables').
A scattergram displays the relationship: for each participant, a point or cross is marked at the point on the graph where their scores on the co-variables meet:
The above example shows scores on two tasks. If a person scores 14 on the first task and 6 on the second task, a dot is marked where the two points meet. This is repeated for every participant, so that a pattern of dots emerges.
The
line of best fit is a line drawn through this pattern to summarise the data. A computer statistics programme will produce a line of best fit - it is very difficult to do it accurately by hand. If the line is going upwards from left to right, there is a positive correlation. If it is going downwards from left to right, there is a negative correlation.
A postive or negative correlation means that the variables do have a relationship, but cause and effect cannot be assumed, i.e. just because two things seem linked, it doesn't mean that one is causeing the other to change. A classic example is that the time your alarm clock goes off is correlated with the time the sun comes up, but this doesn't mean that the sun is making your alarm clock go off, or that your alarm clock is making the sun rise!
Video clip on why correlation does not equal causation...
The strength of a correlation shows how closely linked the two variables are. If there is a strong correlation, then the two co-variables are very closely linked. The IQ of identical twins is strongly correlated. A weak correlation means that there is some relationship but not a strong one. Many things in Psychology are weakly correlated, including the number of major life stressors we experience and illness (Rahe
et al., 1970). Strength is shown with a number, a correlation coefficient between 1 and 0 (or betwen -1 and 0 for a negative correlation). The further the number is from zero, the stronger the relationship.
Above all, remember that two things being correlated does not mean that there is a cause and effect relationship. The media often misunderstand correlation data, mistaking a relationship for proof of cause and effect. Often, strange relationships are shown, such as
a study which found that using more abbreviations in text messages tend to be better readers (Plester
et al., 2009)! More often than not, there are additional variables playing a role.
Plester, B., Wood, C. and Joshi, P. (2009). Exploring the relationship between children's knowledge of text message abbreviations and school literacy outcomes. British Journal of Developmental Psychology, 27 (1), 145-161.
Rahe, R.H., Mahan, J. and Arthur, R. (1970). Prediction of near-future health-changes from subjects' preceding life changes. Journal of Psychosomatic Research, 14, 401-406.