You are on page 1of 2

1) Calculate the means of x and y, 

x and y.

2) Add up the sum of all (x - x)(y - y) and divide by a scaling factor

Why does this work?

If there is a positive association all points will be either above and to the right of the
centre of the distribution or below and to the left.

When (x - x) is positive, so is (y - y).

When (x - x) is negative, so is (y – y).

Therefore all the values of (x - x)(y - y) will be positive.

If there is a negative association all points will be either above and to the left of the
centre of the distribution or below and to the right.

When (x - x) is positive, (y - y) will be


negative.

When (x - x) is negative, (y - y) will be


positive.

Therefore all the values of (x - x)(y - y) will


be negative.
If there is no association, points will be randomly scattered all around the centre of the
distribution.

Positive and negative values of (x - x)(y - y) will cancel each other out.

An expression for r which is easier to work out is the following.

Many calculators have special functions allowing you to calculate the correlation
coefficient r.

1. Get into the correct mode, usually the one called regression or stat x,y.
2. Enter the data, putting in each pair of observations, with a comma between them
into the data button.
3. Use the r button to calculate the coefficient.

It can be very easy!

You might also like