178 REGRESSION AND CORRELATION
Y
r=-l
**
I -*
1 *a*
X
r=+~ t
*
f
Y
X
r= +1
a*
I ** 0.
Figure 12.4 Scatter diagrams with T = -1 and T = +1.
Y
r=O
...
. ..*._ --.
- .', a, '. a
a.. ,.:.* ,
..... .... '.:,'....... .....
. -... -.
r=O
% t
**'
X
Figure 12.5 Scatter diagrams with T = 0.
that the correlation coefficient can be 0 and still there may possibly be a nonlinear
relation between the two variables. The correlation coefficient is a measure of the
linear relationship between X and Y.
Examining the magnitude of T cannot replace a look at the scatter diagram, but
it is a useful measure of the degree of linear relationship between two variables.
A value of T less than -.7 or greater than +.7 in a large set of data might be considered
to indicate a high degree of relationship. It can be seen from comparing the formulas
for T and b that they always have the same sign. A negative value of T indicates that
the regression line slopes downward. However, the size of r does not indicate the
size of the slope coefficient.