You are on page 1of 15

complex

management decisions are often


because they involve :
conflicting objectives
->
moke things complex
1 Multiple objectives -

2 Risk and uncertority -> Deterministic -> if everything is certain

3 Complex structure -> or


or Not single shet event

4 -
Multiple Stakeholders

↳ Integrate all
of then

Multiple there direction


obj -> is a
Complex Structure
-> max profit ->
Many decision alternatives exist

-> min cost


-> Not all alternatives might be known or recognized
-> max satisfaction

-> max security ->


Sometimes decisions are sequential

Risk and uncertanity


Multiple Stakeholders
for certain
-> future is not knwn

-
->
Yourself , colleagues,
investors , society at
large

-> the attitudes towers risk - ->


risk averse
-> Conflicting obj may exist
Or
risk seek

that ? bounded
ded with rationality
↳ How to
↳ Herbert Simon
-
?
have problems coping with this
complexity
↳ Why people -> nobel price

has
(economy)
-> The human mind

a limited in processing capability

"memory resources
C
limits
-> The cope
with complexity
a
Simplify problems

2 Use heuristic strategies

and biases
This can keed to inconsistency
->
-
↳ Autarsialik
Oct 10
->
Andysis
down into its constituent parts
breaking something

-> Therefore decision analysis involves

decomposing the decision problems into a set of smaller problems

These
->
smaller problems are
generated so--

All & orientation


->
divide big problems ->
pieces
Andysis : -divide
e
and conquer

↓ Defensible rationale : "audit trail'


-

2 Raised conciousness about issues ->


different
- ideas
investment
problem
d Allows
participation ment
:

> carbide bein de > invelment


2 Insights :

creative
thinking thaum olsum
↳ may be conflict
2
but
2 Guidance on inf needs
commitment
dif perfective + dif solution
↳s horize perceptive

recegration
heuristic -
two option -> car
O
emercedes

I
-> tate
a Expected value theory
↳ If not enough inf(Herber Simal
& Herber Siment heuristic uncertain
- world

a
video watch - take notes
probability theory ->

a
Benjamin Franklin->+/- strategies
↳ moral algebra
solve decision problems
->
Decision find optimal solution
aralysis net on

providing
↳ main purpose - insight

understanding
enhanced perceptives

Keeney
- - Clecision with multiple dbs

C
May application decision analysis may address
only part of the problem This
·

n
partial
decision andysis can
the
concentrate on aspect where insights will be
most voluable

-> think - limited source


slaytte
"Conditionally Prescriptive" ->
tanm

↳ the

·
will to

⑤ The analysis provide guidence


11

2
- decision maker based on the judgements
that
O
- him/her during the
have been from
gethered
different judgement
course
of the Molysis

-T focus pieces
- vebu
wee
Pressen
Big

A Caxioms - rule)
of decision analysis rational
Rationality concept be
main sensible the to
↳ people regard
as
,
-> that
-
assumption is rationality
-
The basic

↳ Look at definition

and intuition ->


Segisl
ebtw analysis

decision maker's intuitive feelings
course of action and the
7 Sme conflict arises btw the prescribed

Perhaps :

failed to capture some aspect of problem Or


-

Analysis futorsie
-
formed inconsistend
-
Intuitive preferences were only partly or were

about the decision problem ·

lead to insights and better understanding


deeper
a
Exploring this conflict has

Video Notes
:

Rationality
-
2 driver ↑average

consistency
hertive
->
a lows of logic
-

it

alternatives men
ecology e is
C
lows of probability careful
Inconsitency

↳erzg e

-

driving better
fast
intere rationality
-

cogriteen
less accident
heuristic ·
ecological wottadity
-
different then median

-> mea

Vision of rationality performance


-> good
Decison A B
and
Decision rash decision perfect autcome
I E
-> bod ->

8%
-> good
decision
- -
Outcome
- bod
a carefully considered
successful operation died
the mon
↳ bod autcome

then A -> some potential benefits


9 B is better decision
across many decisions
lower risk -> Outcomes
C A is successful outcome
provide a better guide

-tober
Decision Tree - Influence Diagrams
uncetonity
Preference/value/Utility nation
↳ If
everything is certain

book ->
page
11-17 -> we look at some concept such as SMART , influence diagram , Utility function

I different ores - decision analysis


e
areas use this concepts
concept read and see how different
↳ how they apply After learning
->
->

involving multiple objectives


->
Examples of decisions
-ronking
> Choosing a holiday ?
a Choosing a supplier There is free lunch
& no

service
best after-soles is best ?
2 Which
-

one

-
fastest delivery time

-lowest prices

-best reputation for reliability

We will first focus


mistic
-> on

i a

with decision problems


"approximate methods to deal most ·

2 The limitation of the human mind mean that people use

than
As a result they seek to identify satisfactory ,
rather optimall . . . . -> Slaytten bok

best
a
Best University ? -> research

-> lot
producing a

when time
quick ways of moking decisions ,
which
people
use ,
especially
2
Gigerenzer -> fast & frugal heuristic ->
is limited ...

o
-> -
gorg sigerenzer -> heuristic
less time -> but accuricy ?
spend
needless effort ->
acrossy 11 trade off -> not always tru (in the real word - not
truel
↳ in the small world
just
robust
-

- fast & quick


-> heuristic - > complex heuristic - certain environment

↳ bosed on care
capacity recognition
TV show -> $1 , 000 , 000 question
-> time is gale ...

us -> not work

↳ ple Supplier cose -> not knew


your supplier
& recognition heuristic here
America -> more complex to answer
] ↳ its work certain situations
↳ you can apply heuristic
Germany -> more
easily answer

↳ Nerede , he zoma ,
hongi heuristic ?
3
~
~ -> Compensatory us Non Compensatory Strategies

objective

Compensatory Strategy >
others
↳ Poor
attributes is compensated by good performance or

performance on some

effort
a involve more cognitive .

Compensatory strategies

Making tradeoffs on
attributes is difficult

=>
Non-compensatory

23
19 10
.

-
.

Heuristic
The Recognition Think US example
-
btw the options
to choose
have
-> used where people option is choose
-
De

is not ,
the recognized -

and other
-> If one is recognized
-
for us citizent think
two products example stayt recognize I
.

>
bew
:
Choose
->
Amenager both for German citibent

↳ recognize one

2 Peanut butter example


tests others (941)
people never

X brond
others

The Minimalist Strategy

heuristic
-> first apply recognition
which is the best option
simply
:

-> If neither option is recognized -> guess evia) a) ribes


--

-> Ponder

->
: <
2 Canon

Both are
and

well
Nikon

recognized
Is it

-
compensatory

noncompensatory
or noncompensatory?

attribute -> es . The megapixel value


< So , select an

cameras have lo 1 megapixel


Both
.

->
Compensatory
rondomly E
C Choose another attribute ->
The optical zoom level

choose If compare
a attribute
-> Canon has 4x optimal zoon, Nikon has lox optimal zoom

↳ Nikon is better

2 So choose Nikon b ettribute

↳ Coon is better

Take the lost

to reach decision last time when


you
a

2 Select the attribute that embled you

The Lexicographic Strategy Non-Compensatory

ranked in order of importance just choose the


be
d Used when attributes can Winner

↳ not go to

attribute further one

- Idatify the most importance


that attribute
which is best on

Select the option


-
the highest mexapixels)
comera with
the
↳ Le g
.
.
choose

best the 2nd


performs
on
the which
"tie" on the most important attribute , choose option
C If there is a

on
most imp attribute ,
and so .

if there are few no ties


simple
strategy is cognitively
> The lexicographic
It can work well in enr where

imp then others


- one
attribute is considerably more

scarce
-
inf is
The Semi-Lexicographic Strategy

Like the lexicographic strategy


-

- Except that if options have similar performance on an attribute

3
considered to be tied cheir blue choir
2 They are red us
violation of
transitivity
~
to the next black chair
2
Move on
red chair us

- It may lead to violations of transitivity exiom


blue chair us blue chair

brand
choose the cheaper
otherwise
If Choose higher quality
Example So
:

cent -
dif less than
,

I
Brond Price Quality A B -
- choose A

3 O Low
A choose B
.

B - C ->

B
3 60 .
Heigh
3 40.
A -
C -> choose (
C Medium

video
->
Amos Tversky - Theory
thinking fast & star
-> hoppy
->

(Daniel Kahneman) happy


& -> sed

↳ prospect theory -
1
3
2

(EBA) cutoff point A


Elimination by aspects -

↳ boundary of the acceptable performance

imp
attribute is identified on the attribute
->
The most

is established
-> And a cutoff point

outside this is eliminated


->
Any alternative felling point
on
imp attribute and so

with the and most


continues
-> The process

Choosing a computer system


What is the cutoff
price limit
->
1 74Sm
Aspect
:

time <Lo months


2
:

Delivery a
↓ Aspect Eosy to opply

Involves no comp

smomy eoe
eyor)
time
Delines
- 35

59 Yes

S
Yes

38
Decision Satisficing ->(Simm)
Sequential Making
:

2 alternatives become available


Satisficing is used when sequentially

& Simon (1955) Search


- process stops when an alternative is found that is sufficient

2 The most imp aspect is satisficing is the


espiration
lette the decision maker

E which shows

Ix
100 911 - > I

cost
- .

flat
Suppose ,
you looking for a new

!
Your aspiration level change place

lower then $500


2 The rent is

If I look

this
maybe
I can ->
find a
place

-t Both similar
y
seem

Supplier worked before


Reason Based Choice -B -> Choose A because - you
or it is more popular

and
Tversky (1993)
& Shafir ,
Simonson

↳ When feced with the need to choose ,


decision mokers often seek and construct reasons in order to

↳ Reeson based

I
"
Candidate A selects
-> Candidate B <Which condidate to
satisfactory all characteristic
a Candidate A is - average or on

others
characteristics and very poor an

a Candidate B is i s very good on some

C.
C Which condidate to select/reject
Candidate B
-> Most people choose
This is because
->

Time is limited .

->
make decision
-> Time available to

will involve
given strategy
-> Effect that a

accwrite the effort to make


- Dus try to balance spend
Effort Accuracy Framework
:

a -

decisions
take much
zime Like

Best
-> Decision maker's
knowledge about the env ->
University example

->
Imp of
making an accurate decision

↳According to effort
accuracy framework if high level of accuricy is desired ,
-

a
DMs will effort
spend First try decrease of
,
more to
t -> amount

alternatives
& A combination of also utilized your
strategies are

-
When there is a
long list of alternatives

-
DMs quick and heuristic to
produce a short list
use easy

4 Whether
or not the choice has

->
Decoy Effect (Yem , Tuzok , Fabel
but ,

Rome $500

3
F +
H+ B Then
dominance effect you see ad
asymmetric which
an
offer :

Paris F H+ B
Paris -

F+ A $500
+
$ 500

->
Beginning I am indifferent

-> o -> Ted--> We


not watch the video

Serie
favoriable
o Jerry
-> is more
Tom
Jerry -> Form Tom
-

very similar

->

Dating example
Form B
Com) I Jerry -> Tow is more favorable -

Video :
The Economist Subcription Example Economist . com subcription 167 6
-> .

Printing subcription
& ->
eleminate
Printing & Web subcription 84 % 3
heavy
Phantom Decoy Effect -
Computer -
or

W net
available
↳ ↓ not

Very similar Decoy effect except that asymetrically dominate one of o


* o -> phonton

a
Popular explanations
-> previously
Discounted create reference point
·
-

price a

When choices
selecting among

2 The attributes

choose concentrate unique alternatives


->

-
Choice
->
by unique attributes

Destination B
Destination A Destinatio C

tree 2 focus -> possible unique


First
e alternatives

good (t)

three
Last
-
(-)
bod
Emotion and Choice Sod moods

->
Emotion can inf how inf is
proceed
-> Happy moods

a
Cloudy

Sunny
Days] Student -> study
Doys
small aword
2 Chocolate Cookie -
example
awerd
->
just saying
of the risks
2 Affect heuristic -
feeling generally positive or
negative about a decision option influence one's perceptions or

benefits its might bring .

↳ Nuclear Power Plant example

Justifying Choices
Already Made

2 Then asked to choose a


job

C
Rerete >
justify
> t not try to
justify
&

Bidding kind of things - good example

Eboy ->
webid - limit
you set a

C bid
another
guy giving year max

what will do ?
you

cost of a television

"]
A $188 + $30 shipping attractive
:

Option to be more
A appeared

B
:
$228 including

-> Check the video


26 Ekim

SMART

a
deterministic and involves certainity
related to each other
the important aspects
and how they are
& SMART -> illuminates
i "Min" or "max"
direction of movement . e.

preffered
,

indication of
is
& objective objective
in relation to
an

is used to measure performance


& attribute

Example
TV ad alternatives -> which is best
Max the impact of a I have
Objective =

recall seeing that


ed.
who
Attribute
=
# of people

a measure a score

velve -
SMART

[ utility -> uncertanity

2 Axioms of SMART
DM prefer one of them
option be chosen over c
then A should
+
two
Decidability chosen over B
and B is over C ,

If strength of
-

A is the
A-c
....

than
&A-B DB-C of A a should be
greater
- Transitivity then the strength
of preference over

and B is over C ,

Summation If A is chosen over B


-

↳ ALB B>
applied
-

method to be
bisection
-solvability - Allow

lower bounds for value


-
Finite upper and

based on another attribute


the choice
should not influence
bosed on an attribute
The choice
indifference
↳ Mutual preference
min
2
↳ important
Att 1 At+
Assume both vs in
2

I divide - smaller piece/) A 10 1000

1000
B 20

Location of office Annual Renz

A 30000

15000
B
5000

B
C
12010

6 D

I 300
F
10000
6

We need to have a set of ettributes to measure performance on a numeric stage

-
1-Identify DN
The initial attributes obtained from a Du may be to
vegue Value Tree
Il alternative courses of action be useful for this
2 tree
- might
.
-

3 -
"relevent attributes ↳ Start with general
3- different attributes
Asses the performance of alternatives on

Costs
S-Determine
Benefits
a
weight for each attribute

I
I
6- For each

7 Make
alternative . ~

Turnover
working
conditions
provisional Decision
-

8 Perform
sensitivity analysis
-

·
S
-

to the Di should be included


that
.

All the attributes


are of concern
1- Completeness ->

2 -
Operationality

3 -
Decomposibility
4- Absence of
redundancy
S-Minumum size -> Large tree -> any meaningful enelysis may be impossible
Just ancadirate office image

razing
4- Asses the performance of alternatives or different attributes

Direct

cost Use of Value function


associate ->
uncertainty
za -> It is the interval (improvement) btw the points in the scale that we can
compare

Framin S
ene
value
10 -

90 -

88

70

d
60

Se

40

I -

10

&
L
-

You might also like