1.

February 20th, 2016 - Lecture:

Independence
1. Topics
Multiplication rule for conditional probability.

Independent events: two equivalent definitions and several examples with cards and dice.

People v. Collins and the prosecutor's fallacy.

Mutual independence.

2. Multiplication Rule for Conditional Probability

Example: Hat contains 3 cards, RR, RB, BB. Pick a card, put it on table, see a
R side, what's the probability the other side is B?

o Wrong: 1/2

o Right: 1/3

If E_1 through E_2 are events, all P(E_i) do not equal zero,

P(E1E2...En)=P(E1E2...En)=

P(E1)P(E2|E1)...P(En|E1...En1)P(E1)P(E2|E1)...P(En|E1...En1)

o This might be easier to express as the intersection of smaller events.

o This is really easy thing to prove.

o "Proves itself."

o Proof:
. P(E1)P(E2E1)P(E1)...P(E1...En)P(E1...En1)P(E1)P(E2E1)P(E1)...P(
E1...En)P(E1...En1)

. Cancel. Beautiful cancellation.

o This allows you to say, what happens when the first once happens?
And now, what about the second one? So on so forth.

o Example: What is the probability of a five card hand not containing a

pair?

. E_i = "first i cards in hard do not contain a pair"

. Claim: What we want is E1E2...E5E1E2...E5

. What's funny about this is that we don't want E_1 through E_4, we are actually
only interested in E_5.

. The reason we do this, however, is because it makes this calculation easier.

. Possibilities:

P(E1)=1P(E1)=1

The probaility of the first hand not being a pair is 1, because it's
card.

P(E2|E1)=4851P(E2|E1)=4851

P(E3|E1E2)=4450P(E3|E1E2)=4450

P(E4|E1...E3)=4044P(E4|E1...E3)=4044

P(E5|E1...E4)=3648P(E5|E1...E4)=3648

. Multiply these values together to get 50.7%.

 3. Independence
Definition: Say events E and F are independant if $P(E \cap F)

o P(E) \times P(F) $. + Equivalent definition: E and F are independent i(P(F) = 0 or

P(E|F) = P(E)$.

o Proof:

. Take E and F set P(EF)=P(E)P(F)P(EF)=P(E)P(F).

. If P(F)=0P(F)=0, then done.

. If not, then P(E|F)=P(EF)P(F)P(E|F)=P(EF)P(F)

Exercise: Toss two coins

o E = "1st coin was H"

o F = "2nd coin was H"

o P(E) = P(F) = 1/2

Exercise: Tossing two dice

o E = "sum of dice is 6"

o F = "first die was 4"

o P(E)=P({(1,5),(2,4),(3,3),(4,2),(5,1)})=536P(E)=P({(1,5),(2,4),(3,3),(4,2),
(5,1)})=536

o P(F) = 1/6

o P(EF)=P({(4,2)})=136P(EF)=P({(4,2)})=136

o P(E)P(F)=53616P(E)P(F)=53616
 Exercise: Suppose a family has 3 kids.

o E = "family has at least 1 boy and 1 girl"

o F = "family has at most 1 boy"

o Are these indepedant?

. S={BBB,BBG,BGB,GBB,BGG,GBG,GGB,GGG}S={BBB,BBG,BGB,
GBB,BGG,GBG,GGB,GGG}

. E=S{BBB,GGG}E=S{BBB,GGG}

. F={BGG,GBG,GGB,GGG}F={BGG,GBG,GGB,GGG}

. EF={BGG,GBG,GGB}EF={BGG,GBG,GGB}

. P(EF)=38P(EF)=38

. P(E)P(F)=3412=38P(E)P(F)=3412=38

. Therefore, independent.

1. "Independent" and "Mutually Exclusive" are not the

same

Example: 2 dice

o E = "sum was 6"

o F = "first die was 6"

o E and F are not mutually exclusive, but not independent.

2. Prosecutor's Fallacy: People v. Collins (1968)

The prosectution came up with these numbers:

o The probability of a black man with a beard is 1 in 10

o The probability of a man with a mustache is 1 in 4

o White woman with ponytail is 1 in 10

o So on so forth, 1 in 3, 1 in 10, 1 in 1000

o So he made this calculation:

110141101101311000=112106110141101101311000=112106

o Say that the population of LA was 2410624106

o Then, you'd "expect" to have 2 couples fitting this evidence.

o P(evidence | innocent) P(innocent | evidence)

3. With 3 events

E is indepedant of F and G

F is indepedant of G

Then is E indepedant of F intersect G?

Example: rolling two die

o E = "sum to 7"

o F = "first die was 4"

o G = "second die was 3"

P(E|FG)=1P(E|FG)=1

P(E)=16P(E)=16

o Definition: E, F, and G are independant (alternatively "mutually independant") if:

 P(EF)=P(E)P(F)P(EF)=P(E)P(F)

P(EG)=P(E)P(G)P(EG)=P(E)P(G)

P(FG)=P(F)P(G)P(FG)=P(F)P(G)

P(EFG)=P(E)P(F)P(G)P(EFG)=P(E)P(F)P(G)

4. With n events

Definition: E1,...,EnE1,...,En are indepedant if for every

subset I{1,2,...,n}I{1,2,...,n} of at least size 2

Example: Flip n coins