Key Topics
Decision making stages– Technical support for the stages
DSS GDSS Psychological processes
DSS
Decision support systems typically collections of tools that can help you create optimization and simulation models
They often have database clients, so you can the get the data you need for your models – their parameters– E.g., manufacturing process times, truck
capacity, projected interest rates, …
Components of a DSS
Dialogue Manager Model Base
– Financial Models– Statistical Analysis Models– Graphical Models– Project Management Models
Database
Intelligence
Identify problems and opportunities– Problems sometimes identify themselves
• People complain, profits fall, customers go elsewhere
• But waiting for problems to show themselves makes it harder to solve them
• Sometimes it’s too late
Intelligence
Helps to look for problems that are brewing– Look at key indicators, that is, measures that
can help warn you when there might be a festering problem
– If you were an OU administrator looking at enrollments, what are some key indicators that you would look at to try to predict whether there might be problems on the way?
Key Indicators
External indicators– Predictors of problems outside the firm– What are some key external indicators for
OU administrators? Internal
– Internal indicators of problems– What are some key internal indicators for
OU administrators?
Watching Key Indicators
If you don’t watch, you won’t know
Managers should define key indicators ahead of time, then make sure they watch them
But – there are always new problems
What Can IT Do?
Monitor key indicators automatically, and alert a manager when the indicator is out of bounds– Send an email automatically
Remind the manager to pay attention to certain indicators– Produce regular reports– Email reminders
This Helps, But…
You never can know what all the key indicators are
New problems come up all the time– You can’t be sure you know how to watch
for them There is no substitute for paying
attention
So Far…
Intelligence– Key indicators of problems
• External• Internal
– Opportunities• External• Internal
Opportunities
Are there ways to do new things, or do the same things differently?
Harder to define the key indicators here, but not impossible
If you were an OU administrator, what key indicators would you watch to look for new opportunities?
DIY Many key indicators already exist
– Government data, data from existing systems (accounting, sales, …)
Sometimes you need to create your own key indicators– Gather your own data
What data might an OU administrator gather that is not already available from existing sources?
Design
Identifying and analyzing alternatives Key ideas:
– Group decision making– Surfacing assumptions– Sensitivity analysis
Group Decision Making
Identifying alternatives is one of the most common reasons for group decision meetings
But how can you make sure that the group performs well?
What does it even mean for a group to perform well?– Come up with good alternatives
Brainstorming
Technique for generating alternatives– Gather people with different backgrounds– Ask them to generate ideas, without
evaluating them• Encourage participation• Keep going until people run out of ideas, or for
a fixed time
– Evaluate the ideas, after the idea generation phase
Group Problems
Domination by an individual or clique– Anonymity helps
Group Think– Group develops distorted view of reality– Group members reinforce each other
• E.g., Bay of Pigs
– Devil’s advocate helps• Rotate this unpopular role!
Group Problems Many people are unwilling to abandon a
position once it is taken publicly – Need to defend your position– Takes maturity to change your mind– Takes a boss and a culture that allows you to
change your mind Fear of offering uncertain or unpopular ideas
– Results in a loss of face– Management sets the tone: what is the norm for
reacting to potentially bad ideas?
Modify Brainstorming
Anonymous idea generation Anonymous idea ranking in evaluation
phase Allow people to anonymously change or
retract their ideas Randomly assign the devil’s advocate
role
IT Can Help
Group Decision Support System Software application that consists of
most elements in a DSS, plus software needed to provide effective support in a group decision making
Imagine Excel with features to support brainstorming
GDSS Characteristics Special design Ease of use Flexibility
– Support various group processes
Decision-making support– Spreadsheets, access
corporate DB, access Web, …
Anonymous input
Reduction of negative group behavior
Parallel communication– Several
conversations at once
Automated record keeping
Communication Choices Face-to-face vs. dispersed
– Are group members in the same room? Synchronous vs. asynchronous
– Does everyone meet at the same time? Media
– Text, voice, video Consider cost and effectiveness for different
decisions– E.g., face-to-face offers rich communication, but can
be expensive (travel, hotels, disrupted schedules…)
Delphi
A well-known method Groups of experts generate alternatives,
and rank them over a series of rounds Rankings can be anonymous Discussion between rounds Try to converge (although may not be
able to) Can be done over a network
Other Technology
Less formal techniques are more likely to be subject to error, but are quick and cheap
Chat Forum Email list Teleconferencing
Chat Useful features:
– Record discussion, so you know what was discussed and decided
– Private rooms– Banning (for public chats)– Ability to exchange files– Emoticons
The most important: social standards– Rules for how people treat each other– More important than the technology
Forums Useful features
– Threaded discussion– Searchable – forum used as a knowledge
base as well as a communication mechanism
– Email alerts – forum sends an email to you when there are new postings to a thread you mark as important
– Spell checking, cuss filters, … Social rules!!
Teleconferencing
Ability to see and hear other people in real time
Sometimes ability to work on applications together
Much cheaper than face-to-face meetings, but gives you relatively rich communication
Design
Identifying and analyzing alternatives Key ideas:
– Group decision making– Surfacing assumptions– Sensitivity analysis
Surfacing Assumptions
Sometimes it’s hard to see what you are assuming about a situation that may not be true
In IT, it’s easy to assume that users know things, have resources, have time to learn, …
A project can be in trouble if these assumptions turn out to be wrong
Example: IT for Voluntary NPOs
Suppose you’re creating an IS for small voluntary organizations
You cannot assume that:– They have office space to house PCs
• Much work is done on volunteers’ home PCs
– The home PCs are maintained• Virus protection, backups, …
– Management can force workers to use IT– Workers will see the value of IT
• “Bite me, nerd child”
ACK!
Any of these things can sink a project– How do you deal with the mix of hardware,
software, connectivity, expertise, …?– How do you train people who don’t want to
be trained? It’s enough to drive you nuts!
What To Do?
1. Involve people who know the operational constraints in system development
2. Try the system out in operational environments early in the project
3. “Laboratory” testing Usability testing – give people tasks to do
with a system prototype, and video tape them talking aloud as they do it
Design
Identifying and analyzing alternatives Key ideas:
– Group decision making– Surfacing assumptions– Sensitivity analysis
Sensitivity Analysis
What happens if you’re wrong?– If exchange rates fluctuate– If competition is stronger than predicted– If enrollments are less than you thought– If costs are higher than you forecast?
What’s the effect on your decision?
Sensitivity Analysis
Identify the assumptions that have the most effect on outcomes– Hopefully surfacing assumptions will help
See how much violations of your assumptions change the value of various alternatives
Choice
Ranking the alternatives Issues:
– Identifying outcomes of interest• Account for all important effects of each
alternative
– Valuation of outcomes• Objective functions
– Types of models• Optimization, simulation
Eliminating Alternatives
Constraints– Criteria that an alternative absolutely must
meet– E.g., marketing plan is legal and ethical
Dominated solutions– If alternative A is better than alternative B
in all respects, then no need to consider B
Objective Function
Specifies the relative importance of the various outcomes of an alternative
Ideally, summarizes the value of an alternative in one number
May not work that way in practice – Important qualitative variables that are
difficult to measure– Uncertainty
Objective Function Example
Suppose you make hats and booties for llamas
The more hats you make, the fewer booties you can make, and vice versa
Suppose you make $12 for every hat and $8 for every pair of booties.
Objective Function Example
The objective function is:
You want to maximize Z for the various combinations of H and B that you can make.
Z = 12H + 8B
Objective Function Example Suppose you have the following three
production possibilities:
Which one do you pick?
1 2 3
H 33 42 25
B 48 21 62
Objective Function
In general:
Where:Z = value of alternative i
k = number of attributes
b = importance of attribute j
x = rating of alternative i on attribute j Linear functions are the most common in
business
Zi = b1xi1 + b2xi2 + …+ bkxik
Objective Function
So you need to know:– What the alternatives are– What attributes you will judge each
alternative on– How to weight the attributes
Choice
Ranking the alternatives Issues:
– Identifying outcomes of interest• Account for all important effects of each
alternative
– Valuation of outcomes• Objective functions
– Types of models• Optimization, simulation
Optimization Models
Guaranteed to give you the best choice– Linear programming– Dynamic programming– Queuing models– Inventory models– Data envelope analysis– …
Simple Example You have a supply depot on a river You want to install a security fence You have 1,000 feet of left-over fencing What’s the most space you can
enclose?Jetty
Buildings
A = W ( 1000 – 2 W )
A = 2W2 – 1000 W
dA / dW = 4W – 1000
dA / dW = 0 when W = 250
Jetty
BuildingsW
1,000 – 2W
Another Way
W B A
247 506 124,982
248 504 124,992
249 502 124,998
250 500 125,000
251 498 124,998
252 496 124,992
253 494 124,982
Build a fence with this width
Get this area
In Other Words…
Create a mathematical model that acts like the one you are interested in– Outputs from an input, e. g., area resulting
from building a fence of a given width Try different inputs, and watch the
output This is simulation
Differences?
Optimization models have strict mathematical assumptions– E. g., linear programming assumes that all
constraints can be represented by linear equations
Simulation models are more flexible– But they won’t give you the “right answer”– If the model doesn’t reflect the real
situation, they are of limited value
Everything Depends on the People
The tech is useless without savvy people
To use DSS well, they need to know the business, the tech, and themselves
People exhibit systematic psychological biases when making decisions
If you know about them, you can do allow for them
Confirmation
Tendency to look for confirming information rather than disconfirming information.– Given a frequency question, search our
memories for matches– Think that the real frequency matches what
we recall– Usually makes sense
Confirmation Problems
But:– We hear about some things more than
others. E.g., the media is more likely to cover murder than suicide
– We remember more salient than less salient things, i.e, some things draw attention more than others.
– A murder in your neighborhood might get more attention than a suicide. Why?
Confirmation
If an issue is important, try to get some data
Don’t assume that what comes to mind first is right
The contingency table is a useful tool
Contingency TableDoes not wearing hats cause accidents?
Wearing
Accident
NoAccident
10
NotWearing
50
120 2,200
60
2,320
130 2,250 2,380
Emotions Emotions set goals
– Fear of getting fired – don’t take risk– Don’t like Bill – do the reverse of what he
says– Dominate – say an idea is stupid to put
someone in their place Managers who want their employees to
make good decisions need to intentionally create that environment
Emotions
Emotions can set positive goals– Service– Mastery
Applies to all of your life– People who have goals are more satisfied
with the lives– What are some personal goals people
have?
Emotions
Emotions change the way you process information– Anger can reduce your ability to think
• Emotional cascade
You cannot fully control your emotions– Amygdala
Emotional Control
But there are things you can do Most important: be mindful
– Pay attention to yourself, so you can detect when you are getting out of control
– When others are out of control, ask yourself why
Cool off - 30 minutes Talk about assumptions
Schemas
Schemas are mental templates
How do you know what they are talking about?
A: Did you order it? B: Yeah, it will be here in about 45 minutes. A: Oh... Well, I've got to leave before then. But save me a couple of slices, okay?
Expertise
Experts have more and better schemas than novices
Experts know which schemas to apply when– Diagnosticity
Experts can ask the right questions– The questions that will tell when what the
problem is (i.e., choose the right schema)
Schemas and Stereotypes
Schemas can have negative consequences
Example – women are bad at math Affects not only how people with this
stereotype view women, but how women view themselves, and how they perform
Sample Study
Women perform as much as 12 percent better on math problems when tested in a setting without men…
…women tested in single-sex groups scored a 70-percent accuracy rate on math exams; women tested in groups in which they were outnumbered by men scored a 58-percent accuracy rate
From ScienceDaily, 9/13/2000
Sample Study
Stanford psychology professor Claude Steele has conducted experiments in which he brings in black students and white students to take a standardized test.
The first time, he tells the students that they will be taking a test to measure their verbal and reasoning ability. The second time, he tells them the test is an unimportant research tool.
Steele has found that the black students do less well when they are told that the test measures their abilities.
Frontline, 1999
Schemas and Stereotypes
Schema use is unavoidable– Built in to the way we think
When you see a Chinese man or a black woman, your stereotypes will be activated
But those who show less prejudice add a second step– They question their own thinking
Confidence
Evolution has equipped us with the ability to make fast, confident decisions– The leopard and the human
Social pressure to be sure– “If you’re not sure, you can’t be competent”
Result: People are more sure of some things than they should be
Confidence
People who are sure can lead you to disaster– Many people want certainty from their
leaders– But the world is an uncertain place– Ask why you or someone else is so certain,
and what the consequences of error are
Cognitive Limits
The magic number seven plus or minus two
Keep things as simple as possible, but no simpler
Allow people time to make good decisions
Confidence
People who are sure can lead you to disaster– Many people want certainty from their
leaders– But the world is an uncertain place– Ask why you or someone else is so certain,
and what the consequences of error are