Correlation

Correlation reports how tightly the points in a scatterplot hug a straight line and which way that line tilts, and it carries no units, so switching from inches to centimeters leaves it unchanged. The value $r$ is computed as $r = \frac{1}{n-1}\sum \left(\frac{x_i - \bar{x}}{s_x}\right)\left(\frac{y_i - \bar{y}}{s_y}\right)$, where $\bar{x}$ (read x-bar) and $s_x$ are the mean and standard deviation of $x$. The catch is that $r$ sees straight-line patterns only: the five points $(-2, 4)$, $(-1, 1)$, $(0, 0)$, $(1, 1)$, $(2, 4)$ follow $y = x^2$ perfectly, yet $r = 0$. Correlation is also not resistant, so one far-out point can drag $r$ up or down on its own.