Scatter diagrams

A scatter diagram is the simplest way to see a relationship. Plot one variable on the X-axis and the other on the Y-axis, and mark a point for each pair of values.

Then look at the pattern of the points:

• Points scattered around a line rising upwards (to the right): positive correlation.
• Points around a line going downwards: negative correlation.
• Points spread with no line at all: no correlation.
• Points lying exactly on a straight line: perfect correlation, positive or negative (its value is 'in unity', +1 or –1).

How close the points are to the line shows the strength. Points close to the line mean high correlation; points widely dispersed around it mean low correlation.

If the points follow a curve rather than a straight line, the relationship is non-linear (the book's Figures 6.6 and 6.7). A scatter diagram shows such patterns, which a single number might miss.

So it is always wise to look at the scatter diagram first, before calculating any coefficient. It shows the type of relationship and roughly how strong it is.