The transfer of serious games to business research
An example of knowledge transfer between management theory, a serious
game, and practice
1st GaLA Alignment School, 20-24th June, Edinburgh, UK
Christian Schneider, ETH Zürich, cschneider@ethz.ch
The transfer between theory, SG, and practice
Practice
Theory
Tuesday 21 June 2011
?
SG
2
The aim of the analysis is to do better than that…
Tuesday 21 June 2011
3
Content
 Introduction to the logistics game
 Theoretical background
 Results
 Conclusions and practical implications
Tuesday 21 June 2011
4
Basic facts about the logistics game
 Given




Between 11 and 16 players
3 instructors
Approx. 2 hours
A box full of raw material
 Goal
 Produce the right amount of goods in the right quality at the right
time at minimal costs*.
* costs are determined by costs for personnel, stock, late deliveries and bad quality
Tuesday 21 June 2011
5
Impressions from playing the logistics game
WP
3
Shipping
WP
1
Sales
Stock
Rework
Customer
QC
Direct
Internal Transport
External Transport
Tuesday 21 June 2011
WP
2
WP
4
6
Playing the game
taking
measures
Start
Round 1
measuring
performance
Tuesday 21 June 2011
taking
measures
Round 2
measuring
performance
taking
measures
Round 3
measuring
performance
Round 4
Stop
measuring
performance
7
Content
 Introduction to the logistics game
 Theoretical background
 Results
 Conclusions and practical implications
Tuesday 21 June 2011
8
Theoretical background (1/2)
The sand cone model by Ferdows and de Meyer defines a chronological
order in which different manufacturing capabilities should be tackled in
order to achieve lasting improvements in manufacturing.
Question: Can this model be verified in the logistics game?
Tuesday 21 June 2011
9
Theoretical background (2/2)
Instead of specifying the exact parameters of the learning curve, Adler and
Clark model how learning effects emerge by differentiating between firstand second-order learning.
direct labor hours per unit
First-order
learning
Production
activity
Secondorder
learning
Productivity
improvement
cumulative unit number
Question: Does this model help explain the outcome of the logistic games?
Tuesday 21 June 2011
10
Content
 Introduction to the logistics game
 Theoretical background
 Results
 Conclusions and practical implications
Tuesday 21 June 2011
11
120%
120%
35%
35%
80%
60%
100%
80%
60%
Share of bad parts, Round 2
100%
Share of bad parts, Round 2
Share of bad parts, Round 1
Share of bad parts, Round 1
Results: quality performance and costs
30%
25%
20%
15%
30%
25%
20%
15%
Question: Does a sound performance in quality imply that the financial
result is good as well?
40%
20%
40%
20%
0%
0%
1
10
10%
5%
100 10
10000
1000 100 100001000 100000
100000
10 1
1
Costs
perRound
good part,
Round 1
Costs per good
part,
1
0%
1
40%
20%
0%
10
25%
20%
15%
10%
5%
100 10
1000 100000
10000
1000 100 10000
100000
%
%
%
8%
1
4%
2%
100
1000
8%
6%
4%
2%
0%
1
1000
10000
6%
5%
4%
7%
6%
6%
r = 0.06
5%
5%
4%
3%
2%
1%
10 100
1001000
Round 3
4%
2%
3%
2%
1%
1000
4%
3%
2%
1%
1
1 10
10 100
1001000
Costs
perRound
good part,
Round
Costs per good
part,
4
Round 4
Costs per good part
6%
5%
4%
10 100
1001000
Costs
perRound
good part,
Round 3
Costs per good
part,
3
Tuesday 21 June 2011
1000
3%
2%
1%
0%
0%
1 10
1
1
10
10 100
1001000
1000
10000
0%
0%
1 10 10000
100
1000
Finding: While costs correlate with quality in the first two rounds, this
correlation weakens in round three and disappears in round
four.
6%
0%
%
Share of bad parts, Round 4
%
6%
0%
10
100
7%
7%
10%
8%
10%
Round 2
Share of bad parts, Round 4
Share of bad parts, Round 3
%
5%
10%
r = 0.17
Costs
perRound
good part,
Round
2 per good
Costs
perRound
good part,
Round 3
Costs
part,
3
Costs per good
part,
2
14%
12%
10%
101
Round 1
%
15%
1
Costs
perRound
good part,
Round 1
Costs per good
part,
1
%
20%
0%
0%
1
25%
12%
7%
Share of bad parts, Round 4
0%
60%
r = 0.32
12%
Share of bad parts, Round 4
0%
80%
30%
Share of bad parts, Round 3
0%
r = 0.28
100%
30%
10
100
Costs
perRound
good part,
Round
Costs per good
part,
2
14%
14%
Share of bad parts, Round 3
0%
35%
35%
Share of bad parts, Round 2
0%
120%
Share of bad parts, Round 2
Share of bad parts, Round 1
Share of bad parts
0%
5%
0%
0%
1
10%
1000
Costs
perRound
good part,
Round 4
Costs per good
part,
4
12
Results: quality performance and costs
120%
100%
80%
60%
r = 0.24
40%
20%
0%
0
50
100
150
200
250
300
Costs per good part, Round 3
350
Share of bad parts, Round 1
Share of bad parts, Round 1
Question: Do groups that have good quality performance in the beginning
outperform other groups in round four?
120%
100%
80%
60%
r = -0.03
40%
20%
0%
0
50
100
150
200
Costs per good part, Round 4
Finding: Good quality performance in the beginning has no correlation
with the financial result in the last round.
Tuesday 21 June 2011
13
Results: quality performance and measures
taking
measures
Start
Round 1
Round 2
measuring
performance
Round 3
Round 4
Stop
measuring
performance
best performers
evaluate delta
worst performers
Tuesday 21 June 2011
14
Results: quality performance and measures
Question: Do the measures taken by the best performers differ from the
measures taken by the worst performers?
14
# times chosen
12
Performance
change between
round 1 and 2 in
number of bad
parts:
10
8
6
Best performers
4
Worst performers
2
0
Layout
Lot size
Training
Removal int.
transp.
Q-Training
Finding: The measures taken by the two groups are practically the
same.
Tuesday 21 June 2011
15
Results: quality performance and measures
Question: Does the measure “quality training” improve quality more than
other measures?
Average number of bad
parts
Average costs per good
part
With Q-Training
Total
With Q-Training
Total
Round 1
10.86
11.69
4874.00
2443.22
Round 2
3.57
3.21
256.71
109.31
Delta
7.29
8.48
4617.29
2333.91
Finding: Groups choosing “quality training” as a measure improve
quality less than the total average.
Tuesday 21 June 2011
16
Results: quality performance & learning
Players‘ progress in
producing the part in the
right quality.
First-order learning:
based on repetition and
learning-by-doing
Production activity
Productivity
improvement
Second-order learning:
based on managerial or
engineering actions
purposely improving the
manufacturing
capabilities
Measures aimed at
improving the production
capabilities.
Tuesday 21 June 2011
17
Content
 Introduction to the logistics game
 Theoretical background
 Results
 Conclusions and practical implications
Tuesday 21 June 2011
18
Conclusions
 Costs only correlate with quality performance in the first two rounds.
 Good quality performance in the beginning does not assure a superior
financial result in the end.
 Measures taken between round one and two have no influence on
quality.
 Quality training does not improve quality more than other measures.
 The sand cone model can be confirmed inasmuch as quality improves
early on.
 The learning process helps to explain why quality performance
improves independently of measures taken.
Tuesday 21 June 2011
19
Practical implications
 In manufacturing, improving quality helps to bring down costs early on.
 Improving quality is not necessarily a matter of managerial or
engineering actions, some aspects might improve simply as a function
of the cumulated units produced.
Tuesday 21 June 2011
20
What have we seen?
Practice
?
Theory
SG
WP
3
Shipping
WP
1
Sales
Stock
First-order
learning
Production
activity
Tuesday 21 June 2011
Secondorder
learning
Rework
Productivity
improvement
Customer
QC
Direct
Internal Transport
External Transport
WP
2
WP
4
21
THANK YOU!
Tuesday 21 June 2011
22
BACKUP
Tuesday 21 June 2011
23
Results: quality performance and measures
12
# times chosen
10
8
6
4
2
0
Layout /
Training
Tuesday 21 June 2011
Layout /
Removal int.
transp.
Layout / Lot
size
Lot size /
Training
Lot size / QTraining
Training /
Training
Training / QTraining
24
Results: quality performance & learning
 Plateuing in round three and four
 Improvements of costs base on more sources of learning than
improvements in quality.
14
2500
Costs
2000
Quality
12
10
8
1500
6
# Qabs
Costs
3000
1000
4
500
2
0
0
1
2
3
4
Round
Tuesday 21 June 2011
25
Data sources
 Played for the last 16 years
 Results of 81 games
 Played with students (70%) and practitioners (30%)
Tuesday 21 June 2011
26