Evaluating the Project Coordination Performance
Using Information Centrality
Pelin Ipek TIMURCAN
1
and Sevgi Zeynep DOGAN
2
1
Department of Architecture, Izmir Institute of Technology, 35430, Gulbahce,
Urla, Izmir, Turkey; PH (+90) (232) 750-7054; FAX (+90) (232) 750-6000;
email: pelintimurcan@iyte.edu.tr
2
Department of Architecture, Izmir Institute of Technology, 35430, Gulbahce,
Urla, Izmir, Turkey; PH (+90) (232) 750-7040; FAX (+90) (232) 750-6000;
email: sevgidogan@iyte.edu.tr
ABSTRACT
Evaluation of coordination performance in a project network requires
reliable measures and monitoring methods for effective management. Recent
literature includes studies addressing the relationship between coordinative
activity and the configuration of communication networks. In these works, the
role of network centrality is investigated through the basic standard centrality
measures of degree, betweenness and closeness. Current social network analysis
research emphasizes new formulations of centrality measures for robust structural
analysis of project networks.
This paper presents a novel approach for measuring network centrality
using the concept of information centrality. It is based on the idea that all paths
carry information. The significance of information centrality values for the actors
in a wayfinding signage project at Istanbul Sabiha Gokcen International Airport is
investigated. A centrality index is defined for each firm based on the average of
the four measures of centrality. Findings suggest the existence of a high
correlation between coordination scores and the centrality indices. A centrality
index augmented by information centrality measure has potentials for assessing
the coordination performance in construction management research, and it is
promising for the structural analysis of project communication networks.
INTRODUCTION
Communication networks and the information exchange play a critical role
in the coordinative activity of project participants (Hossain et al. 2006; Hossain
2009). The social network characteristic of centrality has the potential of
identifying project participants in coordinator roles. Traditionally, contractual
links within the project participant network define the coordinator role but recent
findings argue that a central position in the communication network of a project
organization may also be an indication of a party’s coordinator role (Dogan et al.
2012; Dogan et al. 2013; Hossain et al. 2006; Hossain 2009). The correlation
between network centrality and coordination is widely discussed in social studies
(Bavelas 1950; Freeman 1978/79; Leavitt 1951). However, the construction
research domain lacks such discussion apart from the work of Hossain et al.
(2006), Hossain (2009) and Dogan et al. (2012). Recently, Dogan et al. (2013)
proposed an uncomplicated, quantitative method to predict a firm’s coordination
score via a centrality index based on social network analysis. In this recent study,
three standard centrality measures of the firms in a wayfinding signage project at a
major airport construction were calculated and a centrality index was defined for
1119Construction Research Congress 2014 ©ASCE 2014
Construction Research Congress 2014
Downloaded from ascelibrary.org by IZMIR YUKSEK TEKNOLOJI ENSTITUSU on 04/24/17. Copyright ASCE. For personal use only; all rights reserved.
each firm using the average of these centrality measures. Findings showed that the
coordination scores of the firms are highly correlated with the centrality indices.
Centrality measures the distribution of relationships and describes the
influence of a node based on how well connected the node is in a network. Three
basic measures, degree, betweenness and closeness are proposed by sociologists to
determine the centrality of a node in a social network (Freeman 1978/79).
Different measures of centrality capture different aspects of a node according to
its role within the network. As Freeman argued in his seminal paper (1978/79) that
degree of centrality indexes a node’s activity, whereas betweenness centrality
measures a node’s control, and closeness centrality measures its communication
efficiency. Previous empirical studies provide nodes that exhibit high centrality
and thus achieve high levels of performance (Ahuja et al. 2003; Pryke 2004;
Schilling and Phelps 2007; Tallberg 2004; Wasserman and Faust 1994). However,
over the years, researchers have proposed different measures, such as information
centrality (Stephenson and Zelen 1989) focusing on different features of central
nodes.
Our purpose in this paper is to introduce a novel approach for evaluating
network centrality based on the concept of information centrality that has been
overlooked for the analysis of communication networks in construction research.
The measure of information centrality weighs all paths between a pair of nodes
assuming all paths carry information and quantifies the relevance of each of the
nodes in the network. It has a rationale and provides readily interpretable data.
The calculations are relatively straightforward and can be done for large networks.
The use of quantifiable values of information centrality can also illustrate
communication structures, patterns and relationships in an e-mail network
(Stephenson and Zelen 1989).
The paper is organized as follows: Descriptions and formulas for
information centrality are presented in detail followed by the description of the
construction project case study. The methodology part includes the calculations
for information centrality measures for each of the firms in the construction
project and a centrality index based on the average of all four centrality measures
is defined. Finally, the correlation between coordination performance and network
centrality are evaluated, findings are discussed and concluding remarks are made.
INFORMATION CENTRALITY
The concept of information centrality is essential for the proposed
approach in this paper. As an earlier reference for centrality measures, Freeman’s
(Freeman 1978/79) betweenness centrality encompasses the betweenness counts
focusing only on geodesics of the given network paths. This leads to the neglect of
the paths with distances greater than the minimum path length attained by the
geodesics. In communication relations, actors might choose message paths that are
longer than the geodesics under the influence of strong reliability reasons or
intrinsic system characteristics. Information might be deliberately channeled
through various communication paths created via many intermediaries
(Stephenson and Zelen 1989). In these form of networked communications,
information robustness in a network with its all-possible paths is vital where
information might take an indirect route.
The index of centrality concept developed by Stephenson and Zelen (1989)
comprises all the paths between network actors by assigning a weight coefficient
to the each path depending on its length. In this framework, a weighted function of
1120Construction Research Congress 2014 ©ASCE 2014
Construction Research Congress 2014
Downloaded from ascelibrary.org by IZMIR YUKSEK TEKNOLOJI ENSTITUSU on 04/24/17. Copyright ASCE. For personal use only; all rights reserved.
the paths is calculated using the inverses of the lengths of the paths as weights.
Geodesics are given as weights of unity, while longer paths than the geodesic
length receive smaller weights based on the contained information (Wasserman
and Faust 1994). The information of a path is defined quite simply as the inverse
of its length.
The concept of information is used extensively both in the communication
and the statistical estimation theory. Information is statistically defined as the
inverse of the variance of an estimator. If an estimator has a small variance, it has
large information which is considered positive. The opposite case is also a valid
argument where poor estimators with large variances have little information. This
approach is applied to the network centrality framework by extending
betweenness on geodesics to all possible paths and weighting according to the
information contained by these particular paths (Stephenson and Zelen 1989).
Stephenson and Zelen (1989) argued that the information centrality of an
actor is a function of all the information for the paths flowing out from that
specific actor. The chosen function is the harmonic average. The procedural
arguments given below presents Stephenson and Zelen’s (1989) approach in
mathematical details:
If
refers to a pair of nodes in the graph, the first phase is the
identification of all possible paths connecting
and . Suppose that there
are
paths connecting and ; accordingly the existing paths for are
.
Then topological distance
is defined for all the denoted paths, which
is the sum of the number of existing links on the specified path. Distance
calculations, i.e.
are done for the each path
.
The information measure between two nodes i and j
is defined as the
reciprocal of the topological distance
between them: . The
information of node
is the sum of all information content with regard to
all other nodes
in the network: ; n indicating the
total number of nodes in the network.
The centrality information measure
of node is defined by the
harmonic average of the information flowing from
to the all other
nodes in the network:
(1)
CASE STUDY
This study re-evaluates the correspondence of coordination scores and
network centrality measures of the project participants of the Istanbul Sabiha
Gökçen International Airport (ISGIA) wayfinding signage project by using
information centrality concept. Prior to this study, the interplay between
coordinative activity and network centrality in the project was initially studied by
using degree, betweenness and closeness measures and the findings were reported
(Dogan et al. 2013).
1121Construction Research Congress 2014 ©ASCE 2014
Construction Research Congress 2014
Downloaded from ascelibrary.org by IZMIR YUKSEK TEKNOLOJI ENSTITUSU on 04/24/17. Copyright ASCE. For personal use only; all rights reserved.
The wayfinding signage design project of the ISGIA’s new terminal
building’s interiors and exteriors is used as a case study in this paper. Being one of
the largest capital investment project in Turkey, ISGIA is a major airport located
on the Asian part of Istanbul and ISGIA project was undertaken by using build-
operate-transfer (BOT). The owner of the project was Airport Management and
Aviation, Inc. and the other party in the BOT arrangement was the Limak-GMR-
MAHB consortium. The consortium awarded the architectural design project
through a design competition to the Tekeli-Sisa Architectural Partnership, and the
construction contract to the joint venture Limak-GMR. The joint venture Limak-
GMR subcontracted the wayfinding signage design/build (D/B) project to
YONSIS, who used Woodhead as a consultant to perform the schematic design,
developing signs, specifying materials, fabrication details, and font styles and
colors. YONSIS also engaged RGB as a management consultant. RGB
coordinated the activities of the participants of the wayfinding project by
providing consulting services such as scheduling work, organizing meetings,
monitoring RFIs, submittals and change orders, and overseeing the project
closeout. Table 1 shows the roles of the project participants (Dogan et al. 2013).
Table 1. Roles of the project participants
Wayfinding Signage D/B
Project Participant Firms
Roles
YONSIS Subcontractor
RGB Management Consultant
Woodhead Design Consultant
LIMAK-GMR General Contractor
Tekeli-Sisa Designer
The main communication channel among the participating firms was the e-
mail exchange due to the geographic separation of the project parties. The general
contractor (Limak-GMR-MAHB) was an international joint venture (Limak-
Turkey, GMR-India, MAHB-Malaysia), the designer (Tekeli-Sisa) and the
consulting firm (RGB) were local project participants. The subcontractor in
charge of the wayfinding signage design project (YONSIS) was located in Izmir,
Turkey and the sub-subcontractor (Woodhead) was located in Adelaide, Australia.
E-mail traffic took place over a period of six months. The complete e-mail
communication data of 216 sent and received e-mails were reviewed and analyzed
by Erbasaranoglu (2011).
The e-mail communications constituted a good source of data for
measuring the network relationships and the coordination patterns during the
project process. In order to analyze this kind of data more effectively, Dogan et al.
(2013) proposed a simple procedure for monitoring the coordinative performance
of project participants. The degree, betweenness and closeness centrality measures
of the project participants are calculated using social network analysis on the e-
mail communication network between the participants. For defining the level of
coordinative role within the project network, a centrality index is defined for each
firm based on the average of the three standard centrality measures. The firm’s
1122Construction Research Congress 2014 ©ASCE 2014
Construction Research Congress 2014
Downloaded from ascelibrary.org by IZMIR YUKSEK TEKNOLOJI ENSTITUSU on 04/24/17. Copyright ASCE. For personal use only; all rights reserved.
coordination score is also calculated based on the content analysis of the sent and
received e-mails between the project participants. The data analyses returned
highly correlated values between the coordination scores and centrality indices.
Findings suggested that the coordinative role of a firm can be identified by its
particular centrality index. Referencing the proposed method in this paper, the
centrality index can be measured relatively easy by using simple software and
capturing only the number and direction of emails exchanged between the
participants, rather than performing time-consuming content analysis of e-mail
exchanges to calculate coordination scores.
METHODOLOGY OF THE STUDY
The methodology of the study involves two parts: (1) Conducting social
network analysis for the email dataset obtained from the wayfinding signage
design project; calculating information centrality measures; and defining centrality
indices for project participants; and (2) Evaluating correlation between
coordination performance and network centrality of project participants.
Measuring Information Centrality
Network centrality identification of social network analysis is used to
examine the electronic mailing of firms participating in wayfinding signage design
project. Information centrality measurements are calculated by UCINET,
software developed by Borgatti et al. (2002). UCINET analyzed the information
exchange data matrix for email communications between the participants of the
project. The sent and received e-mails by each firm were added up provided that
the recipient is the direct one other than the recipients of carbon copy (“CC” or
“BCC” types); for CC and BCC e-mails are usually intended as passive
information propagation, rather than establishing a two-way relationship (Klimt
and Yang 2004).
Based on Freeman’s (1978/79) definitions, degree centrality denotes the
number of nodes connected to one node in particular; betweenness centrality is the
extent to which a firm lies between other pairs of firms (it is the proportion of all
the shortest paths (i.e., geodesic distances) between pairs of other firms that pass
through the firm); closeness centrality is based on the sum of the geodesic
distances from each node to all other nodes. Then, following Stephenson and
Zelen (1989)’s definition information centrality is the harmonic average of all the
information included in all the paths flowing from that firm.
Stephenson and Zelen (1989) stated that the information centrality
calculations could be practically conducted by simply inverting an incidence
matrix rather than the principal information centrality formula (Eq. 1) as
previously introduced. A crucial component of this later formula is the sum of the
strengths or values for the lines incident with a node (Wasserman and Faust 1994).
The number of sent and received emails present in the weighted information
exchange network of the wayfinding signage project implies compatibility for
such computation. For a network with n nodes,
matrix is defined:
(2)
and off-diagonal elements
1123Construction Research Congress 2014 ©ASCE 2014
Construction Research Congress 2014
Downloaded from ascelibrary.org by IZMIR YUKSEK TEKNOLOJI ENSTITUSU on 04/24/17. Copyright ASCE. For personal use only; all rights reserved.