2.

18 Independent RVs
2.18 Independent Random Variables
Intro / Denition
Consequences of Independence
Covariance and Correlation
Anti-University of Georgia Example
Theorems Involving Covariance
Random Samples
1
2.18 Independent RVs
Intro / Denition
Recall that two events are independent if Pr(AB) =
Pr(A)Pr(B).
Then
Pr(A|B) =
Pr(A B)
Pr(B)
=
Pr(A)Pr(B)
Pr(B)
= Pr(A).
And similarly, Pr(B|A) = Pr(B).
Now want to dene independence for RVs, i.e., the
outcome of X doesnt inuence the outcome of Y .
2
2.18 Independent RVs
Denition: X and Y are independent RVs if, for all
x and y,
f(x, y) = f
X
(x)f
Y
(y).
Equivalent denitions:
F(x, y) = F
X
(x)F
Y
(y), x, y
or
Pr(X x, Y y) = Pr(X x)Pr(Y y), x, y
If X and Y arent indep, then theyre dependent.
3
2.18 Independent RVs
Theorem: If X and Y are indep, then f(y|x) = f
Y
(y).
Proof:
f(y|x) =
f(x, y)
f
X
(x)
=
f
X
(x)f
Y
(y)
f
X
(x)
= f
Y
(y).
Similarly, X and Y indep implies f(x|y) = f
X
(x).
4
2.18 Independent RVs
Example (discrete): f(x, y) = Pr(X = x, Y = y).
X = 1 X = 2 f
Y
(y)
Y = 2 .12 .28 .4
Y = 3 .18 .42 .6
f
X
(x) .3 .7 1
X and Y are indep since f(x, y) = f
X
(x)f
Y
(y), x, y.
5
2.18 Independent RVs
Example (cts): Suppose f(x, y) = 6xy
2
, 0 x 1,
0 y 1.
After some work (which can be avoided by the next
theorem), we can derive
f
X
(x) = 2x, if 0 x 1, and
f
Y
(y) = 3y
2
, if 0 y 1.
X and Y are indep since f(x, y) = f
X
(x)f
Y
(y), x, y.
6
2.18 Independent RVs
Easy way to tell if X and Y are indep. . .
Theorem: X and Y are indep i f(x, y) = a(x)b(y),
x, y, for some functions a(x) and b(y) (not necessarily
pdfs).
So if f(x, y) factors into separate functions of x and
y, then X and Y are indep.
7
2.18 Independent RVs
Example: f(x, y) = 6xy
2
, 0 x 1, 0 y 1. Take
a(x) = 6x, 0 x 1, and b(y) = y
2
, 0 y 1.
Thus, X and Y are indep (as above).
Example: f(x, y) =
21
4
x
2
y, x
2
y 1. Funny (non-
rectangular) limits make factoring into marginals im-
possible. Thus, X and Y are not indep.
8
2.18 Independent RVs
Example: f(x, y) =
c
x+y
, 1 x 2, 1 y 3.
Cant factor f(x, y) into fns of x and y separately.
Thus, X and Y are not indep.
Now that we can gure out if X and Y are indep,
what can we do with that knowledge?
9
2.18 Independent RVs
Consequences of Independence
Denition/Theorem (another Unconscious Statistician):
Let h(X, Y ) be a fn of the RVs X and Y . Then
E[h(X, Y )] =
_
_
_

y
h(x, y)f(x, y) discrete
_

h(x, y)f(x, y) dxdy continuous

(x +y)f(x, y) dxdy
=
_

xf(x, y) dxdy +
_

yf(x, y) dxdy
=
_

x
_

f(x, y) dy dx +
_

y
_

f(x, y) dxdy
=
_

xf
X
(x) dx +
_

yf
Y
(y) dy
= E[X] +E[Y ].
11
2.18 Independent RVs
Can generalize this result to more than two RVs.
Theorem: If X
1
, X
2
, . . . , X
n
are RVs, then
E
_
n

i=1
X
i
_
=
n

i=1
E[X
i
].
Proof: Induction.
12
2.18 Independent RVs
Theorem: If X and Y are indep, then E[XY ] = E[X]E[Y ].
Proof (cts case):
E[XY ] =
_

xyf(x, y) dxdy
=
_

xyf
X
(x)f
Y
(y) dxdy (X and Y are indep)
=
_
_

xf
X
(x) dx
__
_

yf
Y
(y) dy
_
= E[X]E[Y ].
13
2.18 Independent RVs
Remark: The above theorem is not necessarily true if
X and Y are dependent. See the upcoming discussion
on covariance.
Theorem: If X and Y are indep, then
Var(X +Y ) = Var(X) +Var(Y ).
Remark: The assumption of independence really is
important here.
14
2.18 Independent RVs
Proof:
Var(X +Y )
= E[(X +Y )
2
] (E[X +Y ])
2
= E[X
2
+2XY +Y
2
] (E[X] +E[Y ])
2
= E[X
2
] +2E[XY ] +E[Y
2
]
(E[X])
2
2E[X]E[Y ] (E[Y ])
2
= E[X
2
] +2E[X]E[Y ] +E[Y
2
]
(E[X])
2
2E[X]E[Y ] (E[Y ])
2
= E[X
2
] (E[X])
2
+E[Y
2
] (E[Y ])
2
= Var(X) +Var(Y ).
15
2.18 Independent RVs
Covariance and Correlation
These are measures used to dene the degree of as-
sociation between X and Y if they dont happen to
be indep.
Denition: The covariance between X and Y is
Cov(X, Y )
XY
E[(X E[X])(Y E[Y ])].
Remark: Cov(X, X) = E[(X E[X])
2
] = Var(X).
16
2.18 Independent RVs
If X and Y have positive covariance, then X and
Y move in the same direction. Think height and
weight.
If X and Y have negative covariance, then X and Y
move in opposite directions. Think snowfall and
temperature.
17
2.18 Independent RVs
Theorem (easier way to calculate Cov):
Cov(X, Y ) = E[XY ] E[X]E[Y ].
Proof:
Cov(X, Y ) = E[(X E[X])(Y E[Y ])]
= E
_
XY XE[Y ] Y E[X] +E[X]E[Y ]
_
= E[XY ] E[X]E[Y ] E[Y ]E[X] +E[X]E[Y ]
= E[XY ] E[X]E[Y ].
18
2.18 Independent RVs
Theorem: X and Y indep implies Cov(X, Y ) = 0.
Proof:
Cov(X, Y ) = E[XY ] E[X]E[Y ]
= E[X]E[Y ] E[X]E[Y ] (X, Y indep)
= 0.
19
2.18 Independent RVs
Danger Will Robinson: Cov(X, Y ) = 0 does not imply
X and Y are indep!!
Example: Suppose X U(1, 1) and Y = X
2
(so X
and Y are clearly dependent).
But
E[X] =
_
1
1
x
1
2
dx = 0 and
E[XY ] = E[X
3
] =
_
1
1
x
3

1
2
dx = 0,
so Cov(X, Y ) = E[XY ] E[X]E[Y ] = 0.
20
2.18 Independent RVs
Denition: The correlation between X and Y is
= Corr(X, Y )
Cov(X, Y )
_
Var(X)Var(Y )
=

XY

Y
.
Remark: Cov has square units; corr is unitless.
Corollary: X, Y indep implies = 0.
21
2.18 Independent RVs
Theorem: It can be shown that 1 1.
1 is high corr
0 is low corr
1 is high negative corr.
Example: Height is highly correlated with weight.
Temperature on Mars has low corr with IBM stock
price.
22
2.18 Independent RVs
Anti-UGA Example: Suppose X is the avg yards/carry
that a UGA fullback gains, and Y is his grade on an
astrophysics test. Heres the joint pmf f(x, y).
X = 2 X = 3 X = 4 f
Y
(y)
Y = 40 .0 .2 .1 .3
Y = 50 .15 .1 .05 .3
Y = 60 .3 .0 .1 .4
f
X
(x) .45 .3 .25 1
23
2.18 Independent RVs
E[X] =

x
xf
X
(x) = 2.8
E[X
2
] =

x
x
2
f
X
(x) = 8.5
Var(X) = E[X
2
] (E[X])
2
= 0.66
Similarly, E[Y ] = 51, E[Y
2
] = 2670, and Var(Y ) = 60.
E[XY ] =

x

y
xyf(x, y)
= 2(40)(.0) + +4(60)(.1) = 140
Cov(X, Y ) = E[XY ] E[X]E[Y ] = 2.8
=
Cov(X, Y )
_
Var(X)Var(Y )
= 0.415.
24
2.18 Independent RVs
Cts Example: Suppose f(x, y) = 10x
2
y, 0 y x 1.
f
X
(x) =
_
x
0
10x
2
y dy = 5x
4
, 0 x 1
E[X] =
_
1
0
5x
5
dx = 5/6
E[X
2
] =
_
1
0
5x
6
dx = 5/7
Var(X) = E[X
2
] (E[X])
2
= 0.01984
25
2.18 Independent RVs
Similarly,
f
Y
(y) =
_
1
y
10x
2
y dx =
10
3
y(1 y
3
), 0 y 1
E[Y ] = 5/9, Var(Y ) = 0.04850
E[XY ] =
_
1
0
_
x
0
10x
3
y
2
dy dx = 10/21
Cov(X, Y ) = E[XY ] E[X]E[Y ] = 0.1323
=
Cov(X, Y )
_
Var(X)Var(Y )
= 0.4265
26
2.18 Independent RVs
Theorems Involving Covariance
Theorem: Var(X+Y ) = Var(X)+Var(Y )+2Cov(X, Y ),
whether or not X and Y are indep.
Remark: If X, Y are indep, the Cov term goes away.
Proof: By the work we did on a previous proof,
Var(X +Y ) = E[X
2
] (E[X])
2
+E[Y
2
] (E[Y ])
2
+2(E[XY ] E[X]E[Y ])
= Var(X) +Var(Y ) +2Cov(X, Y ).
27
2.18 Independent RVs
Theorem:
Var(
n

i=1
X
i
) =
n

i=1
Var(X
i
) +2

i<j
Cov(X
i
, X
j
).
Proof: Induction.
Remark: If all X
i
s are indep, then
Var(
n

i=1
X
i
) =
n

i=1
Var(X
i
).
28
2.18 Independent RVs
Theorem: Cov(aX, bY ) = abCov(X, Y ).
Proof:
Cov(aX, bY ) = E[aX bY ] E[aX]E[bY ]
= abE[XY ] abE[X]E[Y ]
= abCov(X, Y ).
29
2.18 Independent RVs
Theorem:
Var(
n

i=1
a
i
X
i
)
=
n

i=1
a
2
i
Var(X
i
) +2

i<j
a
i
a
j
Cov(X
i
, X
j
).
Proof: Put above two results together.
30
2.18 Independent RVs
Example: Var(XY ) = Var(X)+Var(Y )2Cov(X, Y ).
Example:
Var(X 2Y +3Z)
= Var(X) +4Var(Y ) +9Var(Z)
4Cov(X, Y ) +6Cov(X, Z) 12Cov(Y, Z).
31
2.18 Independent RVs
Random Samples
Denition: X
1
, X
2
, . . . , X
n
form a random sample if
X
i
s are all independent.
Each X
i
has the same pmf/pdf f(x).
Notation: X
1
, . . . , X
n
iid
f(x) (indep and identically
distributed)
32
2.18 Independent RVs
Example/Theorem: Suppose X
1
, . . . , X
n
iid
f(x) with
E[X
i
] = and Var(X
i
) =
2
. Dene the sample
mean as

X
1
n
n

i=1
X
i
.
Then
E[

X] = E
_
1
n
n

i=1
X
i
_
=
1
n
n

i=1
E[X
i
] =
1
n
n

i=1
= .
So the mean of

X is the same as the mean of X
i
.
33
2.18 Independent RVs
Meanwhile,. . .
Var(

X) = Var
_
1
n
n

i=1
X
i
_
=
1
n
2
Var
_
n

i=1
X
i
_
=
1
n
2
n

i=1
Var(X
i
) (X
i
s indep)
=
1
n
2
n

i=1

2
=
2
/n.
So the mean of

X is the same as the mean of X
i
, but
the variance decreases!
34