Y e xy e x y e x y x y e xy x y x y x y e xy x y

Info icon This preview shows pages 56–63. Sign up to view the full content.

View Full Document Right Arrow Icon
Y  E XY E X Y E X Y X Y E XY X Y X Y X Y E XY X Y where the last equality is useful for calculations. Note that Cov X , X Var X . By applying the Cauchy-Schwarz inequality to the deviations from means we get | Cov X , Y | SD X   SD Y 56
Image of page 56

Info icon This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
(cov1) For random variables X and Y , Cov X , Y Cov Y , X . (cov2) For constants a 1 , b 1 , a 2 , and b 2 , Cov a 1 X b 1 , a 2 Y b 2 a 1 a 2 Cov X , Y (cov2) is liability for a measure of “association” because it implies that the strength of the association can be increased simply by multiplying one or both of the random variables by a large number. In particular, the covariance depends on units of measurement (such as dollars versus thousands of dollars or years versus months). 57
Image of page 57
As a measure of association, it makes more sense to define the covariance between the standardized RVs, X X X and Y Y Y The covariance between the standardized variables is simply E  X X  Y Y  X Y which leads us to define the correlation between X and Y : Corr X , Y XY XY X Y Cov X , Y SD X SD Y 58
Image of page 58

Info icon This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
(corr1) By the Cauchy-Schwarz inequality, 1 Corr X , Y 1 (corr2) For constants a 1 , b 1 , a 2 , and b 2 with a 1 a 2 0, Corr a 1 X b 1 , a 2 Y b 2 Corr X , Y These properties allow the magnitude of the correlation coefficient to have meaning. 59
Image of page 59
If Corr X , Y 0 or, equivalently, Cov X , Y 0, we say X and Y are uncorrelated . Determining whether two variables are correlated – and the direction of the correlation – is the starting point of many studies. It remains to determine whether there is a a causal relationship, including whether other “confounding” factors affect both variables. 60
Image of page 60

Info icon This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
EXAMPLE : It is easily established that amount of schooling ( X ) and annual income ( Y ) are positively correlated. Does going to school longer “cause” people to have higher earning power, or are education and income partly determined by a third factor such as “cognitive ability.” 61
Image of page 61