28
Table 3 – Differences in Mean Outcome for Digital and Board-based Experiments
Descriptive Statistics
Board exercise
(28 experiments)
Digital exercise
(12 experiments)
Discipline
Design Criteria
Design
Target
Mean
St Dev
Mean
St Dev
T-value
Functional Internal
Area (FIA)
≥
100
121.3
43.4
117.8
45.8
0.22
Architect
ure
Maximum Blue Deltas
(% total) (MCD) (§)
≤
60-70
54.73
7.97
52.75
6.89
0.75
Average Internal
Temperature Range
(AITR)
55-65
60.13
9.50
100.2
63.8
-2.17*
Minimum Individual
Delta Temperature
Range (min IDTR)
20-85
32.3
11.4
39.8
21.7
-1.12
Thermal
Engineeri
ng
Maximum Individual
Delta Temperature
Range (max IDTR)
20-85
80.0
14.0
181.9
98.1
-3.58*
Maximum Load at
Anchor Points
(MLAP)
≤
20
15.41
3.58
25.1
26.6
-1.25
Structural
Engineeri
ng
Maximum Internal
Moment (MIM)
≤
40
29.5
12.1
56.1
60.5
-1.51
Total Budget (TB)
≤
!1400
1956
1288
1089
514
3.04*
Project
Manage
ment
Building Time (BT)
Shortest
76.2
41.5
62.8
23.4
1.27
(*) p < 0.05
The comparative analysis of the two data sets suggests few statistically significant differences in
mean performance for the two groups. However, the analysis shows that the digital results tend to
exhibit a higher variability than the board-based results. This suggests that the performance of the
‘virtual’ engineering design teams may be less consistent than that of the co-located teams. We
conjecture that the consistency of virtual team work may be increased if administrators instruct
teams to adopt rigorous communication and design process protocols (Jain and Sobek II 2006).
Indeed, one ‘architect’ observed: “If I were to run the project again, I would implement a formal
introduction process where key stakeholder interests and roles are clarified prior to any
designing. This would go some way to creating a structured design process.”
Specifically, the analysis reveals that, first, there are no statistically significant differences in
the mean performance of architects and structural engineers participating in virtual and co-
located teams. Second, thermal engineers participating in virtual teams tend to perform less well
29
than those in co-located teams. This insight leads us to question whether the digital system, as it
was developed, made it more difficult for the thermal engineers (who were in charge of the more
laborious, time-consuming calculations) to enforce their views throughout the experiments than
to the other participants performing less technical roles. In their reports, for example, some
‘thermal engineers’ noted that they wished they could have had a tool to help them shorten the
time necessary to assess whether a design configuration was satisfactorily meeting their own
design criteria. Third, the results suggest that ‘virtual’ project managers can be more successful
in ensuring that design outcomes meet key criteria (in terms of total budget and building time)
than co-located project managers. It therefore appears that the concerns raised by ‘virtual’ project
managers (in terms of perceived difficulties in asserting their authority) do not necessarily
manifest that they are unable to influence the engineering design process.
The effectiveness of the Repository was unclear. While there is evidence that knowledge
reuse occurs through social knowledge networks (Demian and Fruchter 2006), only a small
number of users invested time and effort to document some of the chat-based dialogues and/or to
add notes into the repository. We found, for example, some evidence of users looking for
benchmarking information in the repository; as stated by one participant: “From the knowledge
database I found information pertinent to architects, for example, the cost of different materials.
This helped to reduce my uncertainties and gave me more confidence in trying to achieve a “right
first time” approach.” However, even users who found the repository useful may have acted
opportunistically in the sense that at the end of their own experiment they did not spend time
adding new insights or documenting valuable parts of their dialogues:
“Before the exercise began, I took time to read through all the knowledge base. This
provided a few useful tips, and shed some early light on the design objectives of the other