Correlation
The correlation of two variables is a measure of how the
variables conform to a linear relationship. r is often used as the symbol for correlation.
The measure ranges from -1 to 1.
Possible values
- A positive correlation indicates that a
high value for one variable predicts a high value for the other
variable and a low value predicts a low value for the other
variable. Here the pattern in the scatter plot shows an increasing
line (positive slope).
- A negative correlation indicates that a
high value for one variable predicts a low value for the other
variable and a low value predicts a high value for the other
variable. Here the pattern in the scatter plot shows a decreasing
line (negative slope).
- A correlation of 0 shows no pattern in
the form of a line.
Textbook process for calculating correlation
- Calculate the standard scores for the values of both variables.
- For each observation, calculate the product of the standard scores.
- Sum the products and divide by n - 1.
- The result of the division is the correlation.
Here's an Excel spreadsheet that demonstrates this process of
calculating correlation.
Note: r2 indicates the fraction that the regression
equation accounts for the variance of y. It can be
calculated as the ratio of the variance the predicted y values over
the variance of the actual y values (see section 2.3 in text).
Last modified: Fri Sep 30 19:19:17 Central Daylight Time 2005