A further analysis of the decision problem requires the numerical assessment of the
preferences of the decision maker. It is assumed that the decision maker prefers B to , C
to
A
A, and B to. This statement of preferences may be expressed by any function u such
that:
C
uB uC uA() () ()&& (3.1)
The task is to find a particular function u namely the utility function such that it is logically
consistent to decide between and by comparing a 1 a 2 uC
with the expected value of the
utility of the action , namely: a 1
puA() (1 )()puB (3.2)
where p is the probability that the state of nature is : 1.
Assuming that and have been given appropriate values the question is - what
value should have in order to make the expected value a valid decision criterion? If the
probability
uA()
uC
uB()
p of : 1 being the state of nature is equal to 0 the decision maker would choose
over because she prefers
a 1
a 2 B to C. On the other hand if the probability of : 1 being the state
of nature is equal to 1 she would choose a 2 over. For a value of a 1 p somewhere between 0
and 1 the decision maker will be indifferent to choosing over. This value a 1 a 2 p* may be
determined and uC is assigned as:
uC()pu A**() (1p uB)() (3.3)
From Equation (3.3) it is seen that uC
will lie between uA() and uB() for all choices of
p* and therefore the utility function is consistent with the stated preferences. Furthermore it
is seen that the decision maker should choose the action a to only if the expected utility
given this action
12 a
Eua 1 is greater than Eua 2. This is realized by noting that for all p
greater than p* and with uC
given by Equation (3.3) there is:
**
*
() () (1 )()
() (1 )() () (1 )()
( ) (( ) ( )) ( ) (( ) ( ))
uC pu A p uB
puApuBpuApuB
uB uA uB p uB uA uB p
&
&
&
(3.4)
This means that if is properly assigned in consistency with the decision makers stated
preferences i.e.
uC
B preferred to preferred to and the indifference probability C A p*, the
ranking of the expected values of the utility determines the ranking of actions.