Judgment 4:

1
Judgment 4 Judging Probabilities
I. Kodak in China Scenarios II. NML Pharmaceutica
ls III. Essentials of Probability Theory IV. Heu
ristics Biases A) Representativeness B) Avai
lability V. Fault trees subadditivity VI. Prev
iew session 6
2
Basic Principles of Probability Theory
Definitions S set of all possible events E ?
S are called events P(E) is the probability
of event E
Probability scale 0 ? P(E) ? 1 P(ø) 0 P(S
) 1
Rules of internal consistency
P(A?B) ? P(A), P(B) Inclusion Rule
(if A?B ø) P(A?B) P(A) P(B)
Additivity
Rule for updating belief Bayes Theorem
3
Heuristics Biases
Representativeness we often judge likelihood by
how similar the event in question is to our model
of the world, or how well the story hangs together
1) M ? A paradigm
M
_
adding a likely target to an unlikely basic
target seems to make the conjunction more
likely e.g., P(MSFT down Q1) (MSFT up
1999) P(MSFT down Q1)

A
B
2) A ? B paradigm
M
_
adding a target that suggests the basic
target seems to make the conjunction more
likely e.g., P(MSFT loses to DOJ) (stoc
k falls Q1) P(stock falls Q1)

A
B
4
Heuristics Biases
Availability people assess the frequency of a
class or a probability of an event by the ease
with which instances come to mind.
Problem memory is often biased by media coverage
and recent idiosyncratic experience.
A. American Intl Group, Inc. (26)
B. Dayton Hudson Corporation (24)
C. Dow Chemical Co. (60) D. Eastman Kodak (
91) E. Enron (57) F. Intel (38) G. In
gram Micro (79) H. Motorola, Inc. (29) I. U
SX Corporation (55)
5
Fault Trees Subadditivity
Subadditivity people assign higher probabilities
to more detailed descriptions of events and more
specific events.
P(A?B) Basketball Fans Options Traders This cl
ass!
Problem when assigning probabilities to fault
trees or chance nodes of decision trees, the
distribution of probabilities is biased by the
specific events that you identify.
NML Pharm case