You are on page 1of 17

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/235281646

System Dynamics Modelling of Construction Safety Culture

Article in Engineering Construction & Architectural Management · May 2011


DOI: 10.1108/09699981111126179

CITATIONS READS
21 916

2 authors, including:

Sherif Mohamed
Griffith University
211 PUBLICATIONS 3,999 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Quantitative assesment of resilient safety culture model using relative importance index View project

Reliability evaluation of resilient safety culture using fault tree analysis View project

All content following this page was uploaded by Sherif Mohamed on 24 May 2015.

The user has requested enhancement of the downloaded file.


The current issue and full text archive of this journal is available at
www.emeraldinsight.com/0969-9988.htm

ECAM
18,3 System dynamics modelling of
construction safety culture
Sherif Mohamed
266 Griffith School of Engineering, Griffith University, Gold Coast, Australia, and
Thanwadee Chinda
Received 24 June 2009 International Institute of Technology, Thammasat University,
Revised 19 March 2010
Accepted 23 August 2010
Pathumthani, Thailand

Abstract
Purpose – This paper forms part of an ongoing research project being undertaken by the authors
into construction safety culture. The paper aims to investigate the interactions among five key
enablers of construction safety culture, as well as the potential impact of each enabler on
organisational safety goals over a period of time.
Design/methodology/approach – Using system dynamics modelling, the paper reports on the
development of a causal model simulating the interactions among safety culture enablers. The model
is developed based on the logical assumption that, by improving the enablers, there will be an
inevitable safety performance improvement. An index is also proposed and used as an indicator for
assessing the maturity level of safety culture.
Findings – The paper presents and reports on simulation results which reveal that an organisation
with ad-hoc safety implementation (starting at a basic level of safety culture maturity) should
primarily focus on enhancing leadership attributes, in the context of safety, to rapidly and successfully
progress through to higher maturity levels in the future.
Practical implications – The use of system dynamic modelling, with the developed index, will help
organisations to plan the most effective safety implementation process to achieve their safety goals
within a planned time frame.
Originality/value – The use of modelling, with the developed index, will help organisations to plan
the most effective safety implementation process to achieve their safety goals within a planned time
frame.
Keywords Safety, Culture, Leadership, Construction industry
Paper type Research paper

Introduction
Throughout the world, the construction industry is disproportionately more dangerous
when compared to other industries (Rosenfeld et al., 2006). In the United States,
although construction jobs account for just 5 per cent of the total workforce, they
account for more than 17 per cent of annual workplace deaths (Goetsch, 2003). In the
United Kingdom, for example, the industry accounts for one third of all work-related
fatalities and, on average, five construction workers are killed every two weeks, while
one member of the public is killed every month by construction activities (Health and
Safety Commission (HSC), 2003). In Australia, the construction fatality rate in
Engineering, Construction and 2001-2002 was five per 100,000 employees, which made it double the all-industry
Architectural Management average (National Occupational Health and Safety Commission (NOSHC), 2005).
Vol. 18 No. 3, 2011
pp. 266-281 Emerging economies and less developed countries are no exception to high fatality
q Emerald Group Publishing Limited
0969-9988
rates. In Thailand, for example, the rate of accidents and fatalities in the construction
DOI 10.1108/09699981111126179 industry is reported as highest of all industries (International Labour Organisation,
2005). Additionally, Thai construction workers are five times more likely to suffer a System
permanent disability than are those in other industries. In India, one of the world’s fast dynamics
growing economies, the construction industry accounts for a major share of
work-related accidents (Damodaran, 2006). A similar trend is experienced in the modelling
international construction “hotspot” – the United Arab Emirates – where construction
accidents dominate work-related accident records (The UAE Ministry of Labour and
Social Affairs, 2001). 267
Construction accidents cause many human tragedies, de-motivate workers, disrupt
site activities, delay project progress, and adversely affect the overall cost,
productivity, and reputation of the construction industry (Mohamed, 1999).
According to Kartam (1997), construction accidents may arise from a variety of
causes, which can generally be classified as:
. physical incidents posing hazardous situations; and
. behavioural incidents caused by unsafe acts.

The latter has been identified as the main cause of construction accidents (Sawacha
et al., 1999), and is viewed by many as the direct result of having a poor safety culture
(Smith and Roth, 1991). Since poor safety culture can lead to risks to human lives, much
attention has been paid, over the past few years, to organisational safety culture,
especially to its definitions, dimensions, and enablers, as well as to the development of
tools for assessing and monitoring its “health”, in order to identify areas for safety
performance improvement (Choudhry et al. 2007). Wright et al. (1999), for example,
developed a so-called safety culture improvement matrix to be used as a
self-assessment tool in assessing the construction safety culture (CSC). Lardner et al.
(2001) developed a safety culture maturity model to assist organisations in establishing
their current safety culture maturity level and identifying actions needed to improve
their level of maturity. Molenaar et al. (2002) identified a total of 31 characteristics of a
positive safety culture – all of which could collectively be used to provide a
“snap-shot” assessment of CSC. Mohamed (2003) adopted a Balanced Scorecard tool to
benchmark organisational safety culture, and argued that only a performance
measurement tool that has a number of different but complementary perspectives
would enable organisations to pursue incremental safety performance improvements.
The variety of tools, briefly presented above, are an indication of how research is
rapidly progressing towards the development of a reliable and valid instrument to
measure organisational safety culture. A major shortcoming with these tools, though,
is the inability to appropriately capture and present causal links between what the
organisation is doing and what it aims to achieve (in this study called the “Enablers”
and “Goals”, respectively). Another element of weakness lies in a lack of understanding
about the interactions among different safety culture enablers, as well as the extent of
their individual, or combined, effects on the organisation’s ability to achieve safety
performance improvements.
In light of the above introduction, this paper presents a causal model to simulate the
interactions and relationships among CSC enablers, as well as the potential impact of
each enabler on safety goals over a period of time. The causal model is developed based
on the logical assumption that by improving the enablers, there will be an inevitable
performance improvement. This same assumption underlies the EFQM (European
ECAM Foundation for Quality Management) Excellence model (EFQM, 1998), which is
18,3 adopted by this study for use as a conceptual framework.
Given the type of safety improvement method should differ as the organisation
matures (Lardner et al., 2001), an organisation must be able to prioritise areas for safety
improvement to progress sequentially through different levels of cultural maturity.
The developed causal model, therefore, aims to not only assist organisations in
268 appropriately assessing their current CSC maturity level, but also depict the potential
influence of one or more enablers on the ability to achieve pre-determined safety goals.
In developing the causal model, the paper utilises system dynamics (SD) modelling,
which is a computer-aided approach for analysing and solving problems with a focus
on policy analysis and design. The following section briefly presents the main features
of, and rationale for selecting, the EFQM Excellence model as a conceptual framework.
This is followed by a detailed description of the developed causal model. Finally, the
simulation results are reported and discussed.

1. Conceptual framework
Empirical evidence suggests that the application of holistic management models, such
as the EFQM Excellence model, has a positive effect on organisational performance
(Kristensen and Juhl, 1999). In light of this empirical validation, the EFQM Excellence
model, which is a non-prescriptive model, is deemed a suitable conceptual framework
for this research study. As previously mentioned, this framework is based on the
logical assumption that by improving how the organisation operates, there will be an
inevitable improvement in the results.
The EFQM Excellence model consists of nine elements grouped under five
“enablers” criteria namely Leadership (Lds), Policy and Strategy (Pol), People (Ppl),
Partnerships and Resources (Prs) and Processes (Pro), and four “results” criteria
including People, Customer, Society and Key Performance results. The enablers
represent how the organisation operates whereas the results concentrate on achieving
predetermined organisational goals. In this study, however, the focus has been mainly
on the improvements of, and interactions among, the enablers’ criteria to achieve better
results. For this reason the four “results” criteria were combined together into a single
construct (referred to hereinafter as Goals). Accordingly, the proposed CSC model
comprises six constructs (five Enablers and a single set of Goals), as shown graphically
in Figure 1. It is assumed that leadership drives people management, policy and
strategy, as well as resources, and that these three enablers collectively influence the
ability to achieve pre-determined Goals through the implementation and improvement
of suitable processes (see Figure 1).
In accord with the EFQM Excellence model, a total of 1,000 points was evenly split
(500/500) between the Enablers and Goals. The 500 points allocated to the enablers
were distributed as follows:
.
100 points to Leadership (Lds);
.
80 points to Policy and Strategy (Pol);
.
90 points to People (Ppl);
.
90 points to Partnerships and Resources (Prs); and
.
140 points to Processes (Pro) (EFQM, 2000).
System
dynamics
modelling

269

Figure 1.
Proposed CSC model

Importantly, this allocation of points among enablers, reflecting their relative


contribution to the achievement of Goals, is an area of much debate, which is
adequately covered elsewhere (Chinda and Mohamed, 2008). For practical purposes,
however, this study adopted the original enablers’ allocation promoted by the EFQM
Excellence model (see Figure 1). The Goals construct, on the other hand, contained 500
points representing the aggregate scores of people, customer, society, and key
performance results. As stated above, this paper assumes that the CSC builds on the
intricate interactions between both Enablers and Goals, and more importantly among
the Enablers themselves. Therefore, at any level of CSC, the scores for each of the six
constructs could be summed up to form an aggregated score representing the level of
CSC maturity. This aggregated score, called hereinafter as the CSC index, ranges
theoretically from 0 point indicating a very ad-hoc and immature culture to 1,000 points
indicating a very mature culture.
Wright et al. (1999) adopted a similar approach in developing a safety culture
improvement matrix where a weighted score for each enabler is obtained, then
combined, together to achieve an overall score. However, Wright et al.’s study failed to
consider the interactions among different enablers. Without examining safety culture
holistically, it is impossible to determine the influence of one or more enablers on the
realisation of safety goals. Accordingly, these interactions and causal links were first
examined using the statistical technique of structural equation modelling. This was
carried out to gain an insight into the associations (represented by correlation
ECAM coefficients’ values) among different constructs of the proposed model. Full details of
18,3 how these values were obtained are published elsewhere (Chinda and Mohamed, 2008).
As shown in Figure 1, a correlation coefficient of 0.90 was obtained indicating a
very strong correlation between Pro and Goals. The Pro enabler, in turn, appears to be
directly related to both the Ppl and Pol enablers, and indirectly related to the Prs
enabler. Both Ppl and Pol, on the other hand, appear to be strongly influenced by Lds,
270 whereas Prs is indirectly influenced by Lds through Ppl. The correlation coefficients,
shown in Figure 1, also demonstrate that Lds has an indirect relationship with Pro
through the mediating effects on Ppl and Pol. All these statistically significant
relationships were used in developing the causal SD model as elaborated on in the
succeeding sections.
To be able to assess the level of CSC maturity, each maturity level needs a
score-range (zero to 1,000 points). According to the EFQM (1998), the total score of
1,000 points could be divided into five levels, as follows: Uncommitted (0-249); Drifters
(250-499); Improvers (500-749); Award winners (750-999); and World-class (an ultimate
score of 1,000 points). Many researchers (e.g. Dale and Smith, 1997; Ahmed et al., 2003),
however, report the use of the EFQM Excellence model with a number of different
levels and respective score ranges. In this study, the authors adopt five levels of safety
culture maturity as represented in Figure 2, with each level having a score-range of 200
points. These score ranges are used, together with the calculated CSC index, to indicate
the CSC maturity level.

2. System dynamics modelling


2.1 SD applications in construction
In the construction domain, many researchers have reported SD modelling
applications. Love et al. (2002), for example, developed a SD model to capture the
interrelationships among factors that contribute to design errors and reworks in
construction projects. Chritamara and Ogunlana (2002) developed a SD model to better

Figure 2.
CSC maturity levels
understand the interacting nature of the problems inherent in the design-and-build System
procurement of construction projects in Thailand. Tang and Ogunlana (2003) used SD dynamics
modelling to gain insights into the interactions between a country’s construction
market and the organisation’s financial, technical, and managerial capabilities. modelling
In sum, the SD modelling approach has the capability to capture the interactions
among a range of system variables and to predict the implication of each over a period
of time (Khanna et al., 2003). This key feature coupled with its ability to: 271
.
deal effectively with dynamic changes over time;
.
capture feedback processes;
.
permit the use of soft data; and
.
facilitate testing alternative strategies without actually having to implement
them, made SD an ideal approach to model the individual and/or combined
influences of different CSC enablers on safety goals.

2.2 CSC causal loop diagram


To conceptualise a real world system under investigation, the SD focuses on the
structure and behaviour (over time) of the system using multiple feedback loops (closed
chains of cause-and-effect links, in which information about the result of actions is fed
back to generate further action). These feedback loops are presented graphically using
a causal loop diagram (see Figure 3) where the arrows indicate the direction of
influence and þ /2 signs indicate the type of the influence (Khanna et al., 2003). If a
change in one variable generates a change in the same direction in the second variable,
relative to its prior value, the relationship between the two variables is referred to as
positive (represented by a plus sign). On the other hand, if the change in the second
variable takes place in the opposite direction, the relationship is negative (represented
by a minus sign) (Forrester, 1985).
The proposed causal loop diagram of CSC, shown in Figure 3, consists of seven
elements to explain the relationships among the Enablers, Goals, and the CSC index.
These seven elements are:
(1) Enablers. Score at point (t) in time (maximum 500 points): This score is equal to
the sum of the Lds score (maximum 100 points), the Ppl score (maximum 90

Figure 3.
Causal loop diagram of the
CSC index
ECAM points), the Prs score (maximum 90 points), the Pol score (maximum P 80 points),
18,3 and the Pro score (maximum 140 points), i.e. Enablers score ¼ (Lds score þ
Ppl score þ Prs score þ Pol score þ Pro score).
(2) Goals. Score at point (t) in time (maximum 500 points).
(3) CSC Index. Score at point (t) in time (maximum 1,000 points): This score is equal P
to the sum of the Enablers score and the Goals score, i.e. CSC index score ¼
272 (Enablers score þ Goals score) at point (t) in time.
(4) Desired Goals Score. This is the ultimate score that each and every organisation
aspires to achieve. So, it must match the maximum value of the Goals Score. In
other words, it is set as 500 points.
(5) Gap of Goals at point (t) in time. It is equal to the difference between the Desired
Goals Score and Goals Score at point (t) in time, i.e. Gap of Goals ¼ Desired
goals score – Goals score.
(6) Desired CSC Index Score. This score contains five values: 200, 400, 600, 800, and
1,000 points, to match the five CSC maturity levels, respectively (see Figure 2).
Deciding which value to be used depends on the CSC index score at that point of
time. For instance, if the CSC index score at point (t) in time equals 100, meaning
that the organisation is at the first maturity level, and thus, the Desired CSC
Index at this point of time is set as 200 points (representing the threshold for its
current maturity level).
(7) Gap of CSC Index at point (t) in time. This index score is equal to the difference
between the Desired CSC Index Score and CSC Index at point (t) in time, i.e. Gap
of CSC index ¼ Desired CSC index score – CSC index score.

The relationships between the above seven elements, shown in Figure 3, could be
further explained as follows. At any point of time, the CSC Index Score (representing
the sum of the Enablers Goals scores) is compared with the Desired CSC Index Score,
resulting in a Gap of CSC Index reflecting the difference between these two values. As
the CSC Index Score increases (as a result of an improvement in the Enablers and Goals
scores), the Gap of CSC Index decreases, forming a negative (2 ) relationship. An
example of a more detailed causal loop diagram, showing the relationships among
Enablers and Goals, is illustrated below.
As shown in Figure 4, increased management commitment towards safety (Lds) will
tend to increase staff participation in safety activities (Ppl) (positive “ þ ” link) (Teo
et al., 2005), which, in turn, will increase resource requirements (Prs) (positive “ þ ”
link) (Pipitsupaphol and Watanabe, 2000). This increase in resource requirements (Prs)
is likely to enhance safety policy formulation (Pol) (positive “ þ ” link), which
sequentially improves safety implementation (Pro) (positive “ þ ” link). This
assumption reflects the recommendations made by Wright et al. (1999), which
implied that resource requirements are a fundamental element in formulating effective
policies to improve safety process implementation. Further, improved safety
implementation (Pro) will tend to reduce the cost of accidents (Goals) (negative “ –”
link) (Pannirselvam and Ferguson, 2001). Undoubtedly, this reduced cost of accidents
(Goals) leads to more management commitment towards safety (Lds) (negative “ –”
link) (Turner, 1991). Thus, the feedback loop between Lds, Ppl, Prs, Pol, Pro, and Goals
is a positive one.
System
dynamics
modelling

273

Figure 4.
Causal loop diagram of
Enablers and Goals

Similar causal loop diagrams of the CSC model, shown in Figure 1, were then converted
into the so-called “stock-flow” diagrams, using a SD based software package
“STELLA” (Ithink, 2003), to enable the simulation. Model details are described next.

2.3 Construction safety culture dynamic model


The STELLA program is a dynamic tool for building stock-flow diagrams. It provides
the modeller with a menu of symbols for creating a system diagram on a computer
screen. Symbols are selected and moved onto the screen and then connected. Modellers
are constrained by the SD connection rules to produce diagrams, which connect
symbols in a set sequence. In addition to symbols, STELLA provides guidelines for
equation formulation. These guidelines can be thought of as rules for converting
symbols, text and words into algebra (Morecroft, 1988).
The basic CSC dynamic model is illustrated in Figure 5 where interactions among
the enablers were omitted to improve diagram clarity. This model reflects the
assumption that the CSC index can be “healthier”, provided that the organisation
focuses on improving the five Enablers and, accordingly, achieves higher safety Goals.
Within this model, each enabler has its own detailed stock-flow diagram. Those
diagrams are connected to each other in a pattern reflecting the direct influences shown
in Figure 1. For illustration purposes only, the Leadership (Lds) stock-flow diagram is
explained herein in detail.

2.4 Leadership stock-flow diagram


Figure 6 provides a simple representation of the leadership stock and its flow. Each
simulation run is for a time interval (dt) so at any point in time (t), the value of the
leadership enabler (vlds) is represented by Equation 1, where rlds ¼ leadership enabler
rate. This rate is calculated using Equation 2, where rlds depends mainly on:
ECAM
18,3

274

Figure 5.
Basic CSC dynamic model

Figure 6.
Leadership stock-flow
diagram (acronyms are
listed in the Appendix)

.
vlds at time (t-dt);
.
the corresponding gap of goals (ggoals); and
.
the leadership rate fraction (rflds).
It also depends on the product of two more parameters namely; the gap of the
leadership enabler score (glds), and the percentage of that extra effort to be
provided (rather than what is normally being provided as a result of the existing
“ggoals”) to improve the Leadership score in a shorter period of time. To simplify
the simulation analysis, the leadership rate fraction (rflds) was assumed to have a
constant value represented by the average annual increase of safety budget
compared with the corresponding increase in total budget. This assumption was
made on the premise that an increased safety budget (albeit how little that could be)
from one year to the next is a common mechanism used by many, especially top
management, to signal organisations’ continued commitment towards enhancing
safety at the workplace:

vldsðtÞ ¼ vldsðt 2 dtÞ þ ðrldsÞ* dt ð1Þ


rlds ¼ ððvlds þ ggoalsÞ* rfldsÞ þ ðglds* pldsÞ ð2Þ System
Figure 6 may be explained as follows: At any pint in time, when the gap of goals
dynamics
“ggoals” is large (in other words the current Goals score is low compared to the 500 modelling
targeted score), organisations are expected to seriously attempt reducing this gap by
adopting different means. Such means (in the context of leadership) include leaders:
committing more time and effort to safety activities, leading by example, 275
encouraging more effective two-way communication, assigning safety accountability
to staff, and aligning productivity and safety targets. As a result, the “rlds”
increases. Naturally, the increased “rlds” increases the “leadership” stock, which, in
turn, increases the “vlds” value. It is worth noting that the maximum score of “vlds”
is controlled, however, by the maximum “desired value of leadership” (dlds), which
is equal to 100 points (see Figure 1).
Given that Lds is assumed to drive Ppl, Prs, and Pol, the newly obtained “vlds”
value is transferred to the corresponding stock-flow diagrams for these three connected
enablers. This transfer is, however, influenced by the strength of the correlation
between Lds and the three enablers as shown in Figure 1. For example, the “vlds” value
that is being transferred to the Ppl stock-flow diagram will be multiplied by the
correlation coefficient between Lds and Ppl, which is equal to 0.64. Consequently, the
transferred “vlds” value affects the People enabler rate (rppl) which in turn, influences
the People enabler “vppl” value, and so on. The simulation continues to iterate as
cycles, from the Lds to the Goals passing through all the enablers’ stock-flow diagrams.
In each cycle, the Enablers score and the CSC index are calculated, as illustrated in
Equations 3 and 4 below. The cycles continue until the CSC index reaches a score of
800 or higher, indicating that the organisation has now achieved the “continually
improving” (fifth) maturity level (see Figure 2). The next section describes the
simulation results of a base case of the model:
Enablers Score ðtÞ ¼ vldsðtÞ þ vpplðtÞ þ vprsðtÞ þ vpolðtÞ þ vproðtÞ ð3Þ

CSC Index ðtÞ ¼ Enablers Score ðtÞ þ Goals Score ðtÞ ð4Þ

3. Simulation results
The CSC dynamic model is simulated using data collected from a questionnaire survey
targeting the construction industry sector in Thailand. Medium and large local
construction contracting organisations, with staff of 100 or more, were selected for
sampling purposes. Targeted respondents were selected on the assumption that they
held senior appointments, such as executive directors, managing directors, and senior
project managers, within their respective organisations. Full details of the survey and
collected data are published elsewhere (Chinda and Mohamed, 2008). This section,
however, demonstrates the capabilities of the developed model.

3.1 Base case scenario


In the “base case” simulation, the initial values of the five Enablers were set as zero to
explore the hypothetical situation of organisations with no prior safety implementation
whatsoever (i.e. at the bottom of the first level of maturity). The corresponding initial
value of Goals, however, was not equal to zero. The linear regression performed, using
ECAM the survey data gathered from more than 100 organisations, between the Enablers
18,3 score and the Goals score reveals that when the former equals the hypothetical base
value of zero, the latter is equal to 68 points (out of 500) giving rise to Equation 5. This
may be explained as follows: the Goals construct does not totally and exclusively
depend on the implementation of the Enablers. For example, a medium-size
organisation with no commitment to safety may experience very low number of
276 accidents due to light workload, sheer luck, or both:

Goals ¼ 68 þ ð0:79 * EnablersÞ ð5Þ

Initial values of the Enabler score (set at zero) and the Goals (68 points) were
substituted in the simulation model, and the results are displayed graphically in
Figures 7 and 8. The time units used in the simulation can be varied as the SD model
allows the user to define their own units. For simplicity, however, the time unit used in

Figure 7.
Graphical results of
Enablers, Goals and CSC
index values over time

Figure 8.
Graphical results of
Enablers over time
this study will be referred to as “years”, henceforth. Nevertheless, it is worth noting System
that this “years” unit does not represent the calendar year. dynamics
As shown in Figure 7, initially, the gap of the Goals (ggoals) value was relatively large
(500 – 68 ¼ 432 points). This, then, “kicked-start” the value of the Lds enabler (vlds) modelling
which in turn increased the values of the remaining enablers. As the five Enablers’ score
values increased (identifying an improvement in safety culture’s implementation), the
Goals score value, and the CSC index consequently increased. The simulation continued 277
until the CSC index reached the maximum score of 1,000 points. The results showed the
CSC index exceeding the 800 points threshold at the end of year 11 indicating that it took
11 years for this hypothetical organisation, with a non-existent safety policy and safety
implementation process, to progress from the first to the fifth levels of CSC maturity.
The graphs have similar S-shaped patterns, with a slow increase at the beginning of
the simulation. It took six years to progress from the first to the second levels of culture
maturity. This result demonstrated that for an organisation with a non-existent safety
culture policy and implementation process, it is hard to improve the CSC in the early
stage of the safety implementation. This is shown by a slow increase in the rate for
Enablers, and the even slower increased rate for Goals. After the organisation reached
the second maturity level, however, the Enablers and Goals score values increased
rapidly, as depicted by the sharp rises in the curves shown in Figure 7. This, in turn,
enhanced the CSC index, which could be seen as a steep incline of the graph. The
organisation progressed from the second to the fifth maturity levels over five years (at
the end of year 11). After year 11, it becomes too difficult for the organisation to keep on
increasing the Enablers score value at this high rate, as it continues to close the Goals
gap (ggoals) getting closer to the target score of 500 reflecting a mature safety culture.
This, in turn, slows the increase rates of the Goals score value and the CSC index.

3.2 Sensitivity and policy analyses


Close observation of the increasing rate of the five Enablers’ values at the early stage of
the simulation (see Figure 8), indicates that Lds was surprisingly the weakest enabler in
boosting the CSC index, as it produced the least score compared with the other four
enablers. To explain, at the end of year six (when the organisation reached the second
maturity level), the Lds score was only 23.2 out of 100 points. On the other hand, Ppl, Pol,
Prs, and Pro produced 28.2, 39.9, 39.3, and 42.9 per cent of their maximum scores,
respectively. In reaction to a large gap in the Lds score, the organisation must allocate
more effort to improving the leadership’s attributes. Therefore, to expedite the Enablers
value, and to achieve higher CSC index in the early stages, the organisation should focus
more on enhancing the implementation of key attributes of the leadership enabler.
To confirm whether the organisation should concentrate on improving the
implementation of Lds, a number of model runs were needed. First, the organisation
was said to allocate more effort focusing on Lds improvement, i.e. apart from normally
implementing this enabler, the organisation puts extra effort (say 10 per cent) into
further enhancing this particular enabler’s implementation. This means that the
organisation maintained its improvement of the five enablers, but more attention was
given to Lds. Consequently, the “plds” value was set to 0.10, while its peers for the
other enablers were still set at zero. The model was then simulated, and the results
were recorded. Next, the “plds” was set back to zero, then the “pppl” was set as 0.1
(meaning that the organisation now changed its focus, from improving the Lds, to the
ECAM Ppl enabler). The model was re-simulated, and the results were recorded, then the
18,3 “pppl” was set back to zero. This process continued (putting 10 per cent extra effort for
each enabler at a time), and simulations were performed for all five enablers.
The results shown in Table I demonstrate that, by focusing more on Lds, the
organisation reached the second maturity level in a much shorter time (three years), and
achieved the fifth level of maturity four years earlier (it took seven years, instead of 11
278 years, to achieve the fifth maturity level). Therefore, for the organisation starting at level
one of maturity, attention should be paid, in the main, to improving the key attributes of
leadership to successfully progress much-faster through to higher maturity levels. This
finding attests to the critical role played by top management through:
.
being a role model (Teo et al., 2005);
.
assigning and communicating safety responsibilities clearly to all staff (Lardner
et al., 2001); and
.
ensuring that the workload was reasonably balanced among workers to avoid
unsafe behaviours (Mohamed, 2002).

As stated previously, the five Enablers include those necessary elements that enable
the organisation to improve its safety performance. Each enabler comprises a number
of attributes. As a result, any positive interventions in these attributes are expected to
enable the organisation to meet its safety goals. These interventions can be in the form
of improving safety training, providing more safety resources, integrating safety in
business goals, etc. All such interventions could be grouped under enabler-specific
improvement efforts, and referred to these as a set of policy scenarios. The developed
model can assist in testing those policy scenarios to improve organisational safety
culture, without having to implement them. Consequently, organisations may need to
realistically establish their current maturity level (e.g. base-line), then experiment with
different safety policy scenarios to enhance their CSC indices, and select the best policy
that matches their situation.
As a final word of caution, the CSC index, in practice, may never reach its maximum
score. One key reason is top management’s view of the fifth maturity level as a target,

Lds Pol Ppl Prs Pro


Year Index * Level þ Index Level Index Level Index Level Index Level

Initial 68.0 1st 68.0 1st 68.0 1st 68.0 1st 68.0 1st
1 87.0 1st 82.3 1st 85.5 1st 81.7 1st 86.9 1st
2 131.7 1st 109.0 1st 121.5 1st 104.5 1st 113.3 1st
3 210.9 2nd 150.0 1st 177.8 1st 141.1 1st 147.9 1st
4 357.0 2nd 203.1 2nd 272.4 2nd 193.5 1st 189.9 1st
5 540.4 3rd 315.4 2nd 394.6 2nd 290.7 2nd 282.8 2nd
6 735.8 4th 414.3 3rd 549.4 3rd 401.7 3rd 377.4 2nd
7 897.1 5th 561.6 3rd 700.7 4th 558.6 3rd 498.5 3rd
8 970.5 5th 693.8 4th 794.9 4th 699.8 4th 602.1 4th
Table I. 9 991.1 5th 781.9 4th 903.3 5th 792.6 4th 740.2 4th
Experimentation with 10 997.0 5th 876.1 5th 949.4 5th 902.9 5th 797.0 4th
extra efforts given to 11 998.9 5th 937.2 5th 967.1 5th 951.8 5th 896.9 5th
improve each individual
Enabler Notes: ( *) Index ¼ CSC index; (þ) Level ¼ CSC maturity level
not as means of continual improvement. Once the fifth maturity level is reached, top System
management tends to slow the momentum behind all safety activities. This
phenomenon is known as “attention withdrawal” where top management gradually
dynamics
and slowly withdraws its attention to safety when safety performance reflects the modelling
highest level of maturity.

4. Conclusion 279
The main objective of this study was to investigate the interactions and causal
relationships among the key factors (the five Enablers and Goals) of the construction
safety culture. A model was developed utilizing the System Dynamics software to
examine those interactions and causal relationships over a period of time. The CSC
index, developed through modelling, represented the sum of the five Enablers and
Goals’ values at a point in time, and was used together with the five levels of CSC
maturity to indicate the current CSC maturity level.
Simulating the model revealed that an organisation with ad-hoc safety
implementation (organisation starting at level one of CSC maturity) should primarily
focus on enhancing the Lds enabler to rapidly and successfully progress through to
higher CSC maturity levels in the future. The developed CSC index has the potential to
assist organisations in assessing their current safety culture maturity levels, and
identifying areas for safety improvement to enable progress through to higher maturity
levels. Organisations with different maturity levels will need different safety policy and
safety implementation processes, which cannot be imitated. The use of SD modelling,
with the developed CSC index, will help organisations to plan the most effective safety
implementation process to achieve their safety goals within a planned time frame.

References
Ahmed, A.M., Yang, J.B. and Dale, B.G. (2003), “Self-assessment methodology: the route to
business excellence”, The Quality Management Journal, Vol. 10 No. 1, pp. 43-57.
Chinda, T. and Mohamed, S. (2008), “Structural equation model of construction safety culture”,
Engineering, Construction and Architectural Management, Vol. 15 No. 2, pp. 114-31.
Choudhry, R.M., Fang, D. and Mohamed, S. (2007), “The nature of safety culture: a survey of the
state-of-the-art”, Journal of Safety Science, Vol. 45 No. 10, pp. 993-1012.
Chritamara, S. and Ogunlana, S.O. (2002), “System dynamics modelling of design and build
construction projects”, Construction Innovation, Vol. 2 No. 4, pp. 269-95.
Dale, B.G. and Smith, M. (1997), “Spectrum of quality management implementation grid:
development and use”, Managing Service Quality, Vol. 7 No. 6, pp. 307-11.
Damodaran, R. (2006), “Safety in construction organisations”, Property Bytes: The Indian Real
Estate Blog, available at: http://propertybytes.com/?p¼193 (accessed February 20, 2007).
EFQM (1998), Self-Assessment: Guidelines for Companies, EFQM, Brussels.
EFQM (2000), Introducing Excellence, EFQM, Brussels.
Forrester, J.W. (1985), Industrial Dynamics, MIT Press, Cambridge, MA.
Goetsch, D.L. (2003), Construction Safety and Health, Pearson Education, Upper Saddle River, NJ.
Health and Safety Commission (HSC) (2003), Health and Safety Statistics Highlights 2002/03,
HSE Books, Sudbury.
International Labour Organisation (2005), “Thailand – occupational safety and health in the
construction industry”, available at: www.ilo.org/public/english/region/asro/bangkok/
download/background/osh/conth05.pdf (accessed August 13, 2007).
ECAM Ithink (2003), Software Reference Guide: Ithink 8 Technical Documentation, Isee System Inc.,
Lebanon, NH.
18,3 Kartam, N. (1997), “Integrating safety and health performance into construction CPM”, Journal of
Construction Engineering and Management, Vol. 123 No. 2, pp. 121-6.
Khanna, V.K., Vat, P., Shankar, R., Sahay, B.S. and Gautam, A. (2003), “TQM modelling of the
automobile manufacturing sector: a system dynamics approach”, Work Study, Vol. 52
280 No. 2, pp. 94-101.
Kristensen, K. and Juhl, H.J. (1999), “Beyond the bottom line – measuring stakeholder value”,
in Edvardsson, B. and Juhl, H.J. (Eds), The Nordic School of Quality Management,
Studentlitteratur, Lund.
Lardner, R., Fleming, M. and Joyner, P. (2001), “Towards a mature safety culture”, IChemE
Symposium Series, Vol. 148, pp. 635-42.
Love, P.E.D., Holt, G.D., Shen, L.Y., Li, H. and Irani, Z. (2002), “Using systems dynamics to better
understand change and rework in construction project management systems”,
International Journal of Project Management, Vol. 20, pp. 425-36.
Mohamed, S. (1999), “Empirical investigation of construction safety management activities and
performance in Australia”, Safety Science, Vol. 33, pp. 129-42.
Mohamed, S. (2002), “Safety climate in construction site environments”, Journal of Construction
Engineering and Management, Vol. 128 No. 5, pp. 375-84.
Mohamed, S. (2003), “Scorecard approach to benchmarking organisational safety culture in
construction”, Journal of Construction Engineering and Management, Vol. 129 No. 1,
pp. 80-8.
Molenaar, K., Brown, H., Caile, S. and Smith, R. (2002), “Corporate culture”, Professional Safety,
July, pp. 18-27.
Morecroft, J.D.W. (1988), “System dynamics and microworlds for policymakers”, European
Journal of Operational Research, Vol. 35 No. 3, pp. 301-20.
National Occupational Health and Safety Commission (NOSHC) (2005), Regulation Impact
Statement, NOHSC, Sydney.
Pannirselvam, G.P. and Ferguson, L.A. (2001), “A study of the relationships between the Baldrige
categories”, International Journal of Quality & Reliability Management, Vol. 18 No. 1,
pp. 14-34.
Pipitsupaphol, T. and Watanabe, T. (2000), “Identification of root causes of labour accidents in
the construction industry”, Proceedings of the 4th Asia Pacific Structural Engineering and
Construction Conference, Malaysia, pp. 93-8.
Rosenfeld, Y., Rozenfeld, O., Sacks, R. and Baum, H. (2006), “Efficient and timely use of safety
resources in construction”, Proceedings of the CIB W99 2006 International Conference on
Global Unity for Safety and Health in Construction, 28-30 June 2006, Beijing, China,
pp. 290-7.
Sawacha, E., Naoum, S. and Fong, D. (1999), “Factors affecting safety performance on
construction organisations”, International Journal of Project Management, Vol. 17 No. 5,
pp. 309-15.
Smith, G.R. and Roth, R.D. (1991), “Safety programs and the construction manager”, Journal of
Construction Engineering and Management, Vol. 117 No. 2, pp. 360-71.
Tang, Y.H. and Ogunlana, S.O. (2003), “Selecting superior performance improvement policies”,
Construction Management and Economics, Vol. 21 No. 3, pp. 247-56.
Teo, E.A.L., Ling, F.Y.Y. and Chong, A.F.W. (2005), “Framework for project managers to manage
construction safety”, International Journal of Project Management, Vol. 23 No. 4,
pp. 329-41.
Turner, B.A. (1991), “The development of a safety culture”, Chemistry and Industry, Vol. 1, System
pp. 241-4.
(The) UAE Ministry of Labour and Social Affairs (2001), The Reality of Occupational Health and
dynamics
Safety in UAE 2000-2001, The Department of Research and Studies, Ministry of Labour modelling
and Social Affairs, Dubai.
Wright, M.S., Brabazon, P., Tipping, A. and Talwalkar, M. (1999), Development of a Business
Excellence Model of Safety Culture: Safety Culture Improvement Matrix, Entec UK Ltd,
London.
281

Appendix
dlds Desired leadership value.
ggoals Gap of Goals.
glds Gap of Leadership.
Lds Leadership.
plds Percentage of extra effort provided to improve leadership.
Pol Policy and Strategy.
Ppl People.
pppl Percentage of extra effort provided to Improve the People enabler.
Pro Processes.
Prs Partnerships and Resources.
rflds Leadership Rate Fraction.
rlds Leadership Enabler Rate.
rpol Policy and Strategy Rate.
rppl People Enabler Rate.
vlds Leadership Enabler Value.
vpol Policy and Strategy Enabler Value.
vppl People Enabler Value.
vpro Processes Enabler Value.
vprs Partnerships and Resources Enabler Value.

Corresponding author
Sherif Mohamed can be contacted at: s.mohamed@griffith.edu.au

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com


Or visit our web site for further details: www.emeraldinsight.com/reprints

View publication stats

You might also like