ABSTRACT
J[I::(:ri-;1:) 2 L(!li-y)2] (2.2) This quantity lies in the r<lJlge [ -1, 1] and measures the strength of the linear association bet.>veen the two variables. It can easily be shown that the value does not depend on the units in which the two variables are measured. The correlation is negative if 'high' values of x tend to go \vith 'lmv' values of y. If the t\vo variables are independent, t.hen the true correlation is zero. Here, we apply an analogous formula to time-series data to measure \vhether sucr·essive observations are correlated.
