THE USE OF COUNTER EXAMPLES IN LEARNING
PROBABILITY AND STATISTICS
(2nd International Conference on Teaching Statistics, Victoria (Canada))
1. For Random events we consider the notions of a mutual independence and pairwise independence.
Question: Are there sets of random events which are pairwise independent but not mutually independent?
2. Suppose A, B, C are random events satisfying the relation:
P(ABC)=P(A)P(B)P(C).
Does it follow that A, B, C are pairwise independent?
SOLUTION 1. (BERNSTEIN, 1928)
SUPPOSE A BOX CONTAINS 4 TICKETS LABELLED BY
112 121 211 222
LET US CHOOSE ONE TICKET AT RANDOM, AND CONSIDER THE RANDOM EVENTS
A1={1 OCCURS AT THE FIRST PLACE}
A2={1 OCCURS AT THE SECOND PLACE}
A3={1 OCCURS AT THE THIRD PLACE}
P(A1)=1/2 P(A2)=1/2 P(A3)=1/2
A1A2={112} A1A3={121} A2A3={211}
P(A1A2)=P(A1A3)=P(A2A3)=1/4.
So we conclude that the three events A1, A2, A3 are pairwise independent.
However
A1A2A3=
fP(A1A2A3)=0
¹P(A1)P(A2)P(A3)=(1/2)3CONCLUSION: Pairwise independence of a given set of random events does not imply that these events are mutually independent.
SOLUTION 2.
Suppose that
P(A1A2A3)=P(A1)P(A2)P(A3)
Are the events A1, A2, and A3 pairwise independent?
Toss two different standard dice, white and black.
The sample space S of the outcomes consists of all oredred pairs
ij, i,j=1, ..., 6
S={(1,1), (1,2), ..., (6,6)}
Each point in S has probability 1/36.
A1={first die=1,2 or 3}
A2={first die=3,4 or 5}
A3={sum of faces is 9}
A1A2={31,32,33,34,35,36}
A1A3={36}
A2A3={36,45,54}
A1A2A3={36}
P(A1)=1/2 P(A2)=1/2 P(A3)=1/9
P(A1A2A3)=1/36=(1/2)(1/2)(1/9)=P(A1)P(A2)P(A3)
P(A1A2)=1/6
¹1/4=P(A1)P(A2)P(A1A3)=1/36
¹1/18=P(A1)P(A3)P(A2A3)=1/12
¹1/18=P(A2)P(A3)