transpcr.gif (812 bytes) Correlation Coefficient Formula

If the two variables are related positively, we have observed that if the score on one variable is above the mean, then the score on the other variable will also tend to be above the mean. Earlier in Chapter 6, we learned about z-scores. A z-score tells us how far above (or below) the mean a score is in standardized units. Thus, in a positive relationship, the z-scores for the two variables should tend to agree; in particular, they both should be positive. In the scatterplot below, the two z-scores (in red, one for the horizontal variable and one for the vertical variable) are displayed for each point when you move your mouse near the point. Observe how the z-scores do or do not agree for a point for different values of the correlation coefficient. The green numbers count the number of points in each quadrant defined by the means.

 
Move your mouse over the graph. The z-scores for the horizontal variable and the vertical variable are displayed for the point closest to the mouse.

Move the slider and observe how the z-scores agree as the correlation moves towards either -1 or +1.

 
help.gif (460 bytes)
discover.gif (528 bytes)
Be sure to do the Discover suggestions and questions before reading on. You will have discovered that
  • when the z-scores agree, the correlation coefficient is near 1,
  • when the z-scores agree, but with opposite signs, the correlation coefficient is near -1, and
  • when the z-scores disagree, the correlation coefficient is near 0.

File:
© 1999, Duxbury Press.