ppt - Information Systems - New Jersey Institute of Technology

A Dynamic Delphi System
Connie White, Murray Turoff
New Jersey Institute of Technology
A Dynamic Delphi System, RIP
Delphi – A Brief Introduction
 Decision Driven Rounds
 Methodology
 Pilot Study

Traditional Delphi Defined
“Delphi
may be characterized as a
method for structuring a group
communication process, so that
the process is effective in allowing
a group of individuals, as a whole,
to deal with complex problems.”
(Linstone and Turoff)
Traditional Delphi Defined
 Uses
rounds
 Feedback is selective
 Group Consensus is desired
 Deviates merge towards the
median
 All the time in the world
Dynamic Delphi
Asynchronous interaction
 Participate anytime
 Participate in any part
 All information is presented by all
participants
 Rounds are determined by changes
in decisions

Dynamic Delphi
 Fast
Decision Making
Experts can have a ‘no vote’ or ‘delay
voting until further information comes
forth’
 Visual Feedback
 Aids for non-technical experts for
large heterogeneous group

Thurstone’s Law of Comparative Judgment




Set of unidimensional items
Calculates one group opinion from many
No predetermined scale length - relative
Scale is visual feedback



Semantic differentials (ranking)
Identify equivalence perceptions (clusters)
Indicate extreme differences of group selection
(democrat/republican)
Ordinal Interval Scale
1
2
3
4
5
6
|----------------|----------------|-----------------|-----------------|----------------|
Hospital
Ambulance
Police
Superdome
From low to high of importance during Katrina –
where resources go after the levees break
Thurstone’s Interval Scale
.24 .29
4.5
7.5
---|------|------------------------------------|----------------|---Hospital Ambulance
Police Superdome
From low to high of importance
during Katrina – where resources
go after the levees break
Atomic Decision Making
example Hurricane Katrina
Set of Unidimensional Data
{command and control, hospital, nursing
home, super dome, police station
headquarters, ambulance}
Paired Comparisons




(nursing home, hospital),
(ambulance, police headquarters)
(super dome, hospital)
(ambulance, super dome)
Best reflects expert’s judgment breaking down
complex problems into their atomic units
Changes of Decisions
X1
X1
X2
X3
X4
X2
X3
X4
_______
60%
80%
90%
40%
_______
55%
100%
45%
_______
85%
15%
_______
20%
10%
0%
Changes of Decisions
X1
X1
X2
X3
X4
_______
0
0
0
X2
1
_______
0
0
X3
1
1
_______
0
X4
1
1
1
_______
Changes of Decisions
X1
X1
X2
X3
X4
X2
X3
X4
_______
40%
20%
10%
60%
_______
45%
0%
80% 55%
_______
15%
90% 100% 85%
_______
Changes of Decisions
X1
X1
X2
X3
X4
_______
1
1
1
X2
0
_______
1
1
X3
0
0
_______
1
X4
0
0
0
_______
Thurstone (clustering)
Thurstone (ranking)

D>A>F>G>C>E>B.
-.33 -.28
-.07
.06 .08 .16 .25
- --------------|--------|----------|-------B----|--E----|------------|-C----------|-----G--F-|------A---|----D---|-----------|----------------50
-40
-30
-20
-10
0
10
20
30
40
Future Pilot Study
 Pair-wise
Comparison vs Ranking
 Better reflects expert’s view?
 Faster?
 When n = 5? n = 20?