Safety Science 101 (2018) 108–120

Contents lists available at ScienceDirect

Safety Science
journal homepage: www.elsevier.com/locate/safety

A Resilience Engineering -related approach applying a taxonomy analysis to MARK

a survey examining the prevention of risks
Toivo Niskanen
Ministry of Social Affairs and Health, Department for Occupational Safety and Health, P.O. Box 33, SF-00023 Government, Helsinki, Finland

A R T I C L E I N F O A B S T R A C T

Keywords: The aim of this study was to devise a way of applying a taxonomy based on the concepts of resilience engineering
Safety and to demonstrate the feasibility of this taxonomy on survey data. An online internet questionnaire was ad-
Health ministered to over 6700 OSH managers; their response rate was 22%. The corresponding values for workers’
Managers OSH representatives were 5300 and 27%. The present study developed a qualitative taxonomy based on a
Workers
Resilience Engineering approach. This study applied a combination of both qualitative and quantitative research
Resilience Engineering
Work system
methodologies. For both managers and workers, ‘Commitment of management’, ‘Cooperation’ and ‘Systematic
Socio-technical model improvement of OSH’ had positive influences on ‘Decreasing accident risks’. Furthermore, ‘Obligations’,
‘Decreasing accident risks’, ‘Improving ergonomics’, and ‘Training’ exerted a positive influence on ‘Development
of the work conditions’. The classification of resilience with the present taxonomy helped to clarify an intangible
quality like resilience and in this context also helped to develop proactive safety measures. The key constructs of
the present taxonomy helped in providing a deeper understanding of the commonalities, differences, and re-
lationships in their sub-categories. The main implication is that the present resilience taxonomy can be applied
by managers and employees for determining the future intervention process in improving OSH. When applying
the present resilience taxonomy with a participatory approach, the organizations may acquire new perspectives
on the working life and learn how they can collectively ensure the successful outcome and fulfilment of needs in
OSH.

1. Introduction be prepared for inevitable surprise while still meeting pressures to

One of the basic principles of Resilience Engineering (RE) was de-
fined by Hollnagel (2011, p. 273) i.e. he defined resilience as the in-
trinsic ability of a system to adjust its functioning prior to, during, or
following changes and disturbances, so that it can continue to perform
the required operations under both expected and unexpected condi-
tions. Woods (2009) defined that ‘resilience’ is a system’s potential for
adaptive action when information varies, conditions change, or new
kinds of events occur in order challenge the viability of previous
adaptations, models, or assumptions. The concept of resilience would
seem also to incorporate two dimensions as follows: First, the capacity
to anticipate and manage risks before they become serious threats to the
operation (McDonald, 2006, p. 172). Second, the ability to cope with
situations in which the operation has become compromised, i.e. the
organization’s survival is dependent on the adequacy of its response to
the threat or challenge (McDonald, 2006, p. 173).
With respect to resilience Woods (2015) argued that “The lines of
inquiry have progressed to tackle the following questions: (1) how
adaptive systems fail in general and across scales; (2) how systems can
improve on efficiency of resource utilization; (3) what mechanisms
allow a system to manage the risk of brittleness at the boundaries of
normal function; and (4) what operational architectures allow systems
to sustain adaptability over long times and multiple cycles of change.”
By adopting a cognitive strategy, the controller should be able to mount
an appropriate response to interruptions and distractions so that the
organization’s core activities are not put at risk (Malakis and
Kontogiannis, 2011, p. 109). The institutional level of the system must
be responsible for collecting suitable information for the training about
potential problems that threaten resilience (Tjeirhom and Aase, 2011,
p.169): (1) Development of guidelines and requirements for addressing
cross-scale interactions; and (2) The training tools should be provided
to participants from different levels and professions.
Woods (2015) stated that “The process of adapting to disruptions,
challenges and surprises over time changes the system in question in
multiple ways; even when adapting to preserve, the process of adapting
transforms both the system and its environment.” Leveson et al. (2006,
p. 116) indicated that once the combined model has been adopted by an
organization and integrated into its procedures, the amount and quality

E-mail address: toivo.niskanen@stm.fi.

http://dx.doi.org/10.1016/j.ssci.2017.08.016
Received 11 March 2016; Received in revised form 17 January 2017; Accepted 16 August 2017
0925-7535/ © 2017 Elsevier Ltd. All rights reserved.
T. Niskanen
of learning achieved through the investigation and resolution of safety
problems impacts on the effectiveness of system safety efforts and the
quality of subsequent corrective actions, which in turn has a significant
effect on the technical risks present in the system. Resilience cannot be
engineered simply by introducing more procedures, safeguards, and
barriers (Hollnagel and Woods, 2006, p. 348); instead, RE requires a
continuous monitoring of system performance as well as an awareness
of how tasks are actually done. The concept of resilience would seem to
require both the capacity to anticipate and manage risks before they
become serious threats to the operation, and the ability to survive si-
tuations in which the operation has become compromised. One im-
portant aspect of RE is that it strives to clarify the organization's model
of how it creates safety, in order to determine when the model is in
need of revision (Woods, 2006a, p. 22). From some points of view,
quality and safety management are concerned with maintaining stabi-
lity - assuring that a constant standard of work or output (process and
product) is maintained (McDonald, 2006, p. 169).
Little research has been carried out aimed at gathering evidence-
based data in occupational safety and health (OSH) to validate the
scientific theories that underpin the RE approach, especially research
involving a joined quantitative and qualitative analysis. The present
taxonomy in the qualitative study shall specify which fundamental di-
mensions and processes are involved in these relationships, then define
their integrative dynamics, and finally clarify the stages at which these
dimensions take on greater or lesser significance in quantitative terms.
The focus of this work was to develop and apply the OSH taxonomy
to data obtained from a survey concerned with decreasing the risks of
accidents by adopting a systems thinking approach. The present sys-
tematic view utilizes the theories inherent in RE (Hollnagel, 2006) but
applies these concepts in the framework of this analysis. Better under-
standing of systemic structures can then facilitate the design of more
effective safety culture interventions and application of systems
thinking concepts will improve the overall effectiveness of safety
management (Goh et al., 2010).
2. The aim of the study
The main objectives of this study are as follows: (1) Introduce a
procedure for applying the developed taxonomy based on RE using data
from a survey and adopting a socio-technical approach; (2) demonstrate
that a taxonomy based on the RE approach makes it possible to inter-
pret the OSH survey’s data; and (3) show that a RE-based taxonomy is
able to extract OSH domain knowledge from data collected in an in-
ternet questionnaire.
3. Materials
The questionnaire survey was carried out in September-October
2008. The employer respondents were selected from the occupational
safety and health manager (OSH) register (consisting of private business
and municipal sectors) of the Centre of Occupational Safety and from a
similar register compiled by the State Administration consisting of
public servants. The survey was conducted via an online Internet
questionnaire sent to 6710 employers’ OSH representatives (OSH
managers), with 1478 responding, i.e. response rate of 22%. The cor-
responding figures for 5306 workers’ OSH representatives were 1416
(27%). Among employers’ respondents the occupations were according
to the Finnish Standard Industrial Classification as follows:
Manufacturing 26%, Human health and social work activities 17%,
Public administration and defence 17%, Wholesale and retail trade;
repair of motor vehicles 9%, Construction 7%, Education 6%,
Transportation and storage 6%, Electricity, water supply, sewerage and
waste management 4%, Agriculture, forestry and fishing 2%, Mining
and quarrying 2% and others 5%. The occupations of workers’ re-
spondents have similar types of distributions. Although this data was
collected in 2008, the survey was conducted under a similar regulatory
109
Safety Science 101 (2018) 108–120
environment as exists today; thus it is not believed that responses would
differ significantly if collected in 2017. In the present research sample,
both employers’ and workers’ respondents had the similar distributions
according to the different size-classes of the organizations and their
branches.
4. Methods and analysis
4.1. Joined qualitative and quantitative methods and analysis
This study applies a joint qualitative and quantitative research ap-
proach and a taxonomy was developed based on the RE concept.
Onwuegbuzie and Leech (2005, p. 381) postulated that methodological
pluralism should be promoted in order to help as many investigators as
possible to become pragmatic researchers. According to Onwuegbuzie
and Teddlie (2003, p. 352) instead of utilizing terms like quantitative
and qualitative methods, exploratory and confirmatory methods should
be preferred since this would make it possible to unite the quantitative
and qualitative data analysis within the same framework. Creswell and
Plano Clark (2007, p. 9) defined the background used in mixed methods
research in the following way: (1) The voices of participants are not
directly heard in quantitative research; (2) Mixed methods research
helps answer questions that cannot be solved by qualitative or quanti-
tative approaches alone; and (3) Mixed methods research is ‘‘practical’’.
4.2. Quantitative methods and analysis
The following indicator values of the Likert scale were used:
1 = strongly disagree, 2 = somewhat disagree, 3 = not disagree/not
agree, 4 = somewhat agree and 5 = strongly agree. The statistical
analysis was done with SAS software (2005). Hypotheses H1–H10 were
tested using regression analysis. The coefficient of determination (R-
squared) is calculated in regression analysis to indicate the percentage
of the dependent variable that can be predicted by the independent
variables; this level of accuracy in prediction of the dependent variable
will change based on which independent variables are included in the
model. For the statistical consistency of the responses, Cronbach’s alpha
was estimated as a coefficient of reliability or consistency (SAS soft-
ware, 2005). It will be noted that the alpha coefficient is included as
one of the statistics reported with differential decisions, and not with
absolute decisions (Cronbach and Shavelson, 2004). A commonly-ac-
cepted rule is that an alpha of over 0.6 is ‘questionable,’ over 0.7 ‘ac-
ceptable’ and over 0.8 ‘good’ and over 0 0.9 ‘excellent’ (Reynaldo and
Santos, 1999).
4.3. Qualitative taxonomy of the RE approach developed for this study
Hollnagel (2011, p. 273) stated that resilience can be defined as the
intrinsic ability of a system to adjust its functioning. One can make a
joint qualitative and quantitative analysis of what happens when an
organizational, human and technical system is stretched to accom-
modate new demands; this offers opportunities for studying aspects of
system resilience. The role of qualitative indicators is crucial when
developing RE, in fact this is an area which has not received as much
attention as other aspects of measurement.
For this reason, the present study has developed qualitative tax-
onomy indicators. The purpose of the present method is to provide
qualitative information about what is happening in intermediate stages
of the processes of the RE. Starting from the level of the whole system,
the present qualitative approach leads to the development of opera-
tional details and specific steps that can be taken on a concrete level.
The developed taxonomy techniques place a strong emphasis on the
analysis of the relationships between strategy, process, procedures and
function within the inherent features of systems.
The present study is being focused on the following characteristics
(Wilson, 2014): (1) Systems and Organization; (2) People and
T. Niskanen
Performance; and (3) Equipment and Interfaces. These subcategories
are based on the systems thinking approach and were developed for the
present study based on a taxonomy originating from RE. The sub-ca-
tegories of these taxonomies are as follows: (1) Systems and organiza-
tion: work systems, rules and leadership; (2) People and Performance:
communication, processes and competencies; and (3) Equipment and
Interfaces: technology, procedure and performance. Furthermore this
taxonomy is classified into sub-categories in the Sections 4.3.1–4.3.3.
The present taxonomy consists of three different sections: “(1)
Systems and organization”, “(2) People and Performance”, and “(3)
Equipment and Interfaces”. Each section aims to reveal important as-
pects and properties of a resilient system. This study started out by
explaining and exemplifying event-based constraints, which are the
contextual factors with which the system must cope. Subsequently, a set
of core resilience functions was identified; these outline the functional
dependencies for adaptive capacities between the functions. This study
also provided examples of resilience-critical properties of each function.
The next section examined the functional dependencies that arise from
coping OSH strategies. A consequence of using the taxonomy, and
thereby understanding the dependencies between constraints, functions
and strategies is that it becomes apparent that focusing and investing in
certain strategies will influence what problems that can be dealt with
and what resilience functions need to be developed (Lundberg and
Johansson, 2015).
4.3.1. Systems and organization
In this study, the developed sub-categories of the novel taxonomy
(see Section 4.3.4) are marked by typeface of italics. The theoretical
proposals are based on the RE -related studies.
4.3.1.1. Work systems. Socio-technical systems: The general system
view has the following implications for business systems (Sundström
and Hollnagel, 2006, p. 242–243): (1) A business system is a dynamic
open system with entities organized on multiple levels, e g. wholes
could be described as being organized in a classical hierarchical
structure.
Strategy: With respect to RE the work systems deal with the fol-
lowing: (1) Each organized whole is defined by its interactions with
other entities as well as its environment (Interactions); (2) Business
systems will define desired behaviours by defining goals, policies,
standards, processes and procedures (Goal-oriented); (3) Organized
wholes can adopt different types of control modes of management be-
havior (Self-regulation); and (4) The property of resilience emerges as a
result of a system's ability to transition from one state to the next
(Proactive).
Adaptation: Resilience, in particular, is concerned with under-
standing how well the system is able to adapt, especially in what range
or to what types of variation (Woods and Cook, 2006, p. 69).
Integrity: The basic stability and integrity of the system are two
important dimensions; another crucial aspect is the system’s capacity to
absorb major disturbances from the operating environment and to re-
cover from failure (McDonald, 2006, p. 179).
Robustness: Systems can be viewed as hierarchical structures where
management and each level of the line-management impose constraints
on the activity of the level below it. In other words, there may or may
not be constraints at a higher level that allow or control lower-level
behavior (Leveson et al., 2006, p. 98).
4.3.1.2. Rules. Tjeirhom and Aase (2011, p.169) proposed that the
following actions should be implemented in order to strengthen
downward resilience: (1) Development of clear safety goal rules at
the governmental level (Legislation); (2) Downward resilience will be
threatened by an unwillingness to define clear goal rules at the strategic
level (Command-and-control).
Guidelines: With respect to safety tools implemented at this stage,
the priority is to standardise people (competence), work (procedure)
110
Safety Science 101 (2018) 108–120
and technology (ergonomics) with recommended procedures in design
(ergonomics) and operations (guidelines, protocols) as the main generic
tools (Amalberti, 2006, p. 260).
Rule hierarchy: The tools for standardisation will proceed to more
official prescriptions at the next level, and ultimately will form the basis
for new federal or national laws in the last stage (Amalberti, 2006, p.
260). In the context of organizational systems, resilience would seem to
depend on increasing standardisation. The following are examples of
such tendencies (McDonald, 2006, p. 158–159): (1) Stronger co-ordi-
nation of processes by routinisation of procedures in operations and
organizational systems (Collaboration); (2) Increased reliability
through the removal of variance due to individual skill, and ensuring
the substitutability of different people, through standardised selection
and training (Mobilization); (3) Ensuring, through supervision, in-
spection, auditing, etc. that the standardisation of the work-process is
actually controlling the routine flow of work (Controlling); (4) Better
standardisation of the outputs of the process is made possible through
better monitoring as well as recording of those outputs (Monitor-
ing);and (5) Automation of routine or complex functions (Anticipating).
4.3.1.3. Leadership. Policy: The ability to interpret the environment
appropriately is an important aspect as is the capacity to anticipate,
plan and implement appropriate adjustments to address perceived
future requirements (McDonald, 2006, p. 156).
Operations: The design must not only enforce constraints on beha-
vior to ensure safe processes, but it must continue to operate safely as
changes and adaptations will inevitably occur over time (Leveson et al.,
2006, p. 97).
Change management: Resilience appears to convey the property of
being able to adapt to the requirements of the environment, or other-
wise being able to manage the variability or challenging circumstances
emerging from a changed environment (McDonald, 2006, p. 156).
Pro-active: The training aimed at the development can force people
into proactive thinking and to express their expectations of what might
happen in such an event (Bergström et al., 2011, p. 49). When con-
ducting an organizational analysis and diagnosis, it is often rather easy
to identify the apparent imperfections, deficiencies and inconsistencies
to which most organizations are subjects (McDonald, 2006, p. 164).
Flexibility: The enhancement of adaptive and flexible competencies
of the leaders can support problem-solving in a group by enabling
group members to step back from prescribed role behaviours and rou-
tines (Bergström et al., 2011, p. 46).
Adaptation: If one wishes to adopt resilient control and manage-
ment, then the system must have the ability to reflect on how well it is
adapted, what it is adapted to and what is changing in its environment
(Woods and Branlat, 2011, p. 128).
Re-balance: From the point of view of managing resilience, a safety
organization should monitor and dynamically re-balance the equili-
brium between production pressure and risk (Woods, 2006b, p. 318).
Adjustments: There are two sides to organizational safety; first, the
system or organization must be resilient so that it can avoid failures and
losses, second, it must be able to respond appropriately after the fact
(Leveson et al., 2006, p. 95).
Mobilization: By adopting the concept of an organizational system,
this emphasizes the importance of understanding the functional char-
acteristics of systems (McDonald, 2006, p. 157). Resilience is the ability
of systems to prevent or adapt to changing conditions in order to
maintain (i.e. exert control over) a system property (Leveson et al.,
2006, p. 95).
4.3.2. People and performance
In this study, the developed sub-categories of the novel taxonomy
(see Section 4.3.4) are marked by typeface of italics. The theoretical
proposals are based on the RE -related studies.
4.3.2.1. Communication. Interactions: Communication includes the
T. Niskanen
ability to assess potential actions, the consequences of which are both
important but difficult to foresee and that might significantly influence
people's ability to control the problem in the near future (Hollnagel and
Woods, 2006, p. 350).
Information exchange: Communication depends mainly on in-
formation exchange and in order to be accomplished, it requires both
sufficient time and competent cognitive resources (Malakis and
Kontogiannis, 2011, p. 110).
Management of meaning: As the organization goes through this
ongoing process of problem identification, investigation, and resolu-
tion, some lessons will be learned. The problem is that these lessons
may be of variable quality, depending on the investigation process and
its thoroughness (Leveson et al., 2006, p. 117).
Sense-making procedures: Paries (2011, p. 26) has stated that resi-
lience can be interpreted in the following ways (1) generic anticipation
schemes, providing (common) sense-making frameworks of what hap-
pens, at a level of abstraction which is high enough to wrap around all
the countless and unpredictable variations of real scenarios (Flex-
ibility), and (2) fast and efficient implementation protocols and skills,
capable of forcing the available generic schemes to fit the parameters of
the day, under critical time constraints (Anticipation).
Adaptation: The complexity of controlling a process requires that
there be elaborate strategies of adaptation of the human-system en-
semble. Three types of patterns characterize the degree of adaptation at
work (Malakis and Kontogiannis, 2011, p. 113): (1) Coordination (be-
tween practitioners): (2) Resilience (between practitioners and de-
mands) and (3) Affordances (between practitioners and organizational
artefacts).
4.3.2.2. Processes. Procedures: The descriptions of resilience processes
are useful when they specify the harms posed to these processes relative
to challenges originating from external events (Woods and Cook, 2006,
p. 71). Communication can be useful to make descriptions of resilience
processes when they specify the harms that can happen to these
processes relative to challenges posed by external events (Woods and
Cook, 2006, p. 71).
Operations: Downward resilience is important because the context
and structures of a system either foster resilience or induce pressure
towards resilient operations (Tjeirhom and Aase, 2011, p.160). In-
dicators: Sundström and Hollnagel (2006, p. 239) revealed that so-
called state variables should be used to track progress towards these
goals (i.e. leading indicators) and to establish whether the goals have
been achieved (i.e. lagging indicators).
Flexibility: The resiliency allows one to detect undesirable declines
in adaptive capacity and to intervene to increase flexibility in the
adaptive capacity (Woods and Cook, 2006, p. 69).
Relationships: The key to understanding the possibilities of change
concerns the causal relationship between system and action - how the
processes of the organization's systems are understood as constraining
or allowing certain courses of action in order to meet the contingencies
of a particular situation (McDonald, 2006, p. 173).
Micro- and macro-levels: Upward resilience as performed by the
micro-level actors might be reflected in decisions made at the macro-
level such as setting new strategic goals, development of new proce-
dures or implementation of new technology (Tjeirhom and Aase, 2011,
p.161).
Control and interactions: In the strategic control mode, the time
horizon is wider and sets its focus on higher-level goals (Sundström and
Hollnagel, 2006, p. 242). At this level, the dependencies between tasks
and the interaction between multiple goals are also taken into account;
adoption of a strategic control mode is facilitated by introducing system
thinking throughout the organization.
4.3.2.3. Competencies. Evaluation: The role of information exchange is
crucial if an organization is to be able to take evaluation of coordinated
action and perform effectively in critical situations (Malakis and
111
Safety Science 101 (2018) 108–120
Kontogiannis, 2011, p. 110).
Knowledge: Competence and resources are both important for the
system's ability to respond rationally to a threat (Hollnagel and Woods,
2006, p. 349).
Comprehension: In order for effective learning to take place, there
must be a sufficient opportunity to learn and it must be possible to
determine whether the learning (the change in behavior) has the de-
sired effect (Hollnagel, 2011, p. 287).
Application: The competence refers to knowing what to do and
knowing how to do it, whereas the application resources refer to pos-
sessing the ability to do it (Hollnagel and Woods, 2006, p. 349).
Exchange: The organization’s members exchange information in
order to clarify the planning, the actions and the responsibilities
(Malakis and Kontogiannis, 2011, p. 110).
Analysis: The training aimed at the development /of generic and
non-domain specific competencies can increase an analysis competence
in unexpected and rapidly unravelling situations (Bergström et al.,
2011, p. 46).
Synthesis: The safety organization should be able to spot 'holes' in
the organization's decision processes, reframe assessments about how
risky the organization has been acting, question the organization's as-
sumptions about how it achieves safety (Woods, 2006b, p. 318).
4.3.3. Equipment and Interfaces
In this study, the developed sub-categories of the novel taxonomy
(see Section 4.3.4) are marked by typeface of italics. The theoretical
proposals are based on the RE -related studies.
4.3.3.1. Technology. Socio-technical system: When resilience is a
system property, then it probably needs to be considered as an aspect
of the relationship between a particular socio-technical system and the
environment in which that system operates (McDonald, 2006, p. 156).
In order to manage a business system, it is necessary to establish a view
of the organizational regularities (e.g. OSH legislation) at each system
level.
Application: General management, policies and control principles
need to be based on principles applicable to all types of organized en-
tities (Sundström and Hollnagel, 2006, p. 237).
Expertise: The principal measures related to the human resources
used by business systems are of particular importance, i.e., financial,
human and (technology) systems resources (Sundström and Hollnagel,
2006, p. 239).
Flexibility; At the operational level, resilience may be a function of
the way in which organizations approach and manage apparently
contradictory requirements; on the one hand, of good procedures and
good plans, and on the other hand, appropriate flexibility to meet the
real demands of the operation (McDonald, 2006, p. 167–168).
Implementing: A resilient system must have the ability to anticipate,
perceive, and respond (Hollnagel and Woods, 2006, p. 350); RE must
therefore address the principles and methods by which these qualities
can be implemented.
Controlling: The control system of the modelling technological
barriers consists of a combination of hardware, software and beha-
vioural/procedural elements (Hale et al., 2006, p. 292).
Self-regulation: It is self-evident that the hardware/software must
be designed, installed, adjusted, used, maintained and monitored, but it
is also important that behavioural/procedural aspects are defined,
framed, tested, used and monitored (Hale et al., 2006, p. 292).
4.3.3.2. Procedure. Best practices: There tends to be a growing gap
between the real system and the system as it is imagined to be, in part
through local feedback loops. Resilience, as a form of adaptive capacity,
is a system's potential for adaptive action when information varies,
conditions change, or when new kinds of events occur, i.e. challenges to
the viability of previous adaptations, models (Woods and Branlat, 2011,
p. 129).
T. Niskanen
Adaptation: Resilience focuses on assessing the organization's
adaptive capacity relative to the challenges confronting that capacity
(Woods, 2006a, p. 22).
Relationships: The resilience information consists of knowledge of
the relationships between the state of the process being controlled and
the kinds of responses that have been made to recent challenges (Woods
and Cook, 2006, p. 74).
Flexibility and balance: Resilience reflects the tension between
stability and change in organizational and operational systems, medi-
ated by the notion of appropriate adaptation (McDonald, 2006, p. 156).
Calibration: One marker of resilience that has emerged from con-
verging lines of evidence is the distance between operations as man-
agement imagines they should be conducted and how they actually are
done (Dekker, 2006, p. 88); this distance tells something about the
models of current risk, and how well calibrated are organizational de-
cision makers. This, however, requires comparisons and perhaps a type
of quantification (even if only conceptual) that may be challenging to
attain.
Up-to-dateness: It requires that management should be kept up-to-
date about the real nature of work (and their deficient perception of
what is being done) since it is necessary that messengers go beyond
simply sending 'reassuring' signals to management (Dekker, 2006, p.
78).
Preparedness: The definition of risk management needs to be
broader than considerations that restrict it to particular activities or
tools (Leveson et al., 2006, p. 98).
4.3.3.3. Performance. Hierarchical activities: According to Cook and
Nemeth (2006, p. 214), resilient performances are remarkable because
the sudden demands that provoke them generate the need for
practitioners to engage in quite high levels of so-called goal-and-
means hierarchical activities.
Control: The outcome can be a more or less orderly series of ac-
tivities, where the orderliness or regularity of performance reflects the
degree of control (Sundström and Hollnagel, 2006, p. 241).
Flexibility: If management adopts a RE approach then it means that
the organization will have the capacity to deal with informal and
flexible ways of working (McDonald, 2006, p. 168); one aspect of this
concerns the capacity to change rules and procedures to deal with
specific circumstances.
Cognitive performance: A resilient performance is a cognitive per-
formance that is able to traverse or bypass the hierarchy in order to
resolve a conflict (Cook and Nemeth, 2006, p. 214).
Procedures: Training efforts are completely oriented towards the
objective: learning how to recognize a set of anticipated situations, and
how to respond properly to them with the relevant pre-established
procedure (Paries, 2011, p. 25).
Experience: RE is influenced by the context, by knowledge or ex-
perience of dependencies between actions, and by expectations about
how the situation is going to develop - in particular, about which re-
sources are and will be available (Sundström and Hollnagel, 2006, p.
241).
Anticipation: The control structure itself, therefore, must be care-
fully designed and evaluated to ensure that the controls are adequate to
maintain the constraints on behavior will be able to control risk
(Leveson et al., 2006, p. 97-98). One important concept is to view the
management of resilience as a matter of finding the correct balance
between individual resilience (individual responses to operational
challenges) and system resilience (Dekker, 2006, p. 88).
4.3.4. The summary of the developed taxonomy
For “Systems and Organizations” the sub-categories of the taxonomy
are the as follows (Fig. 1):
(1) Work systems: Socio-technical systems; Strategy; Interactions; Goal-
oriented; Self-regulation; Proactive; Adaptation; Integrity;
112
Safety Science 101 (2018) 108–120
Robustness.
(2) Rules: Legislation; Command-and-control; Guidelines; Rule hier-
archy; Collaboration; Mobilization; Controlling; Monitoring;
Anticipation.
(3) Leadership: Policy; Operations; Change management; Pro-active;
Flexibility; Adaptation; Re-balance; Adjustments; Mobilization.
For “People and Performance” the sub-categories of the taxonomy
are the as follows (Fig. 2):
(1) Communication: Interactions: Information exchange; Management
of meaning; Sense-making procedures; Flexibility; Anticipation;
Adaptation.
(2) Processes: Procedures; Operations; Indicators; Flexibility;
Relationships; Micro- and macro-levels; Control and interactions.
(3) Competencies: Evaluation; Knowledge; Comprehension;
Application; Exchange; Analysis; Synthesis.
For “Equipment and Interfaces” the sub-categories of the taxonomy
are the as follows (Fig. 3):
(1) Technology: Socio-technical system; Application; Expertise;
Flexibility; Implementation; Control; Self-regulation.
(2) Procedure: Best practices; Adaptation; Relationships; Flexibility and
balance; Calibration; Up-to-dateness; Preparedness.
(3) Performance: Hierarchical activities; Control; Flexibility; Cognitive
performance; Procedures; Experience; Anticipation.
4.4. The Hypotheses of the study
4.4.1. Review about the research results relating to how RE can be
integrated into Hypotheses H1–H5
The dependent variable “Decreasing accident risks” (as it refers to
Hypotheses: H1-H5)
In the resilience model, a combination of reactive problem in-
vestigation and proactive hazard mitigation efforts leads to effective
safety-related decision making that reduces the technical risk associated
with the operation (Leveson et al., 2006, p. 115).
4.4.1.2. The independent variable: “Identification of OSH dangers” (issues
which will be evaluated in Hypothesis H1). Resilience relates to a very
sensitive awareness of where the organization is in relation to that
danger area and whether it can mount a very rapid and effective
response when signals of approaching or actual danger are detected,
even unexpected or unrecognized signals (Hale and Heijer, 2006, p.
36). Resilience is the ability to steer the activities of the organization so
that it will always remain outside of that dangerous area (Hale and
Heijer, 2006, p. 36). One problem here is that operations 'as imagined'
are not always as amenable to capture or analysis as the idealized
control model. Thus, the starting position would be understanding the
gap between the system as imagined and the system as actually
operated requires (Dekker, 2006, p. 89). These data and the
arguments discussed above lead to the following hypothesis:
Hypothesis 1: For managers and workers “Identification of OSH
dangers” would be positively related with “Decreasing accident
risks” in the joint framework of RE.
4.4.1.3. The independent variable: “Carrying out risk assessments” (to be
considered in Hypothesis 2). Risk assessment focuses on future threats
and is suitable for systems where the principles of functioning are
known, where descriptions do not contain too many details and can be
made relatively quickly and where the systems’ environments are
sufficiently stable (Hollnagel, 2011, p. 286). The key to
understanding the possibilities of risk assessments concerns the
relationship between system and action - how the processes of the
T. Niskanen
organization's systems are understood as either constraining or allowing
certain courses of action in order to meet the contingencies of a
particular situation (McDonald, 2006, p. 173). The above discussion
leads to the following hypothesis:
Hypothesis 2: For managers and workers “Carrying out risk as-
sessments” would be positively related with “Decreasing accident
risks” in the joint framework of RE.
4.4.1.4. The independent variable: “The commitment of upper
management” (will be considered in Hypothesis 3). The upper echelons
of managers must recognize the concerns of human performance, shall
strive to address them and commit themselves to OSH. This can be
achieved by infusing the organization with a sense of how much it
values this aspect of the organization’s work by ensuring that it
provides continuous and extensive follow-through to actions or
activities which damage human health or performance (Wreathall,
2006, p. 279). Control functions, i.e., management actions at different
dynamic business system levels, need to be geared towards promoting
behaviours that lead to a healthy state of the individual in the system,
collection of wholes, and/or the overall business system (Sundström
and Hollnagel, 2006, p. 239). In RE, the task of the managers is to
monitor the boundary conditions on the current system's adaptive
113
Safety Science 101 (2018) 108–120
Fig. 1. “Systems and Organizations”: the sub-ca-
tegories of the taxonomy.
capacity and to target specific points where investments would be
valuable to preserve or expand that adaptive capacity, given the factors
of change being experienced by that field of practice (Woods and Cook,
2006, p. 71). The following hypothesis is put forward:
Hypothesis 3: For managers and workers “The commitment of
upper management” would be positively related with “Decreasing
accident risks” in the joint framework of RE.
4.4.1.5. The independent variable: “Cooperation between employers and
their workers” (will be evaluated in Hypothesis 4). Effective collaboration
channels are needed between the hierarchical levels of each control
structure. There should be both a downward reference channel
providing the information necessary to impose constraints on the
level below and a measuring channel to provide feedback about how
effectively the constraints have been enforced (Leveson et al., 2006, p.
100). It is important to show how the manner in which an organization
deals with the physical, social or economic material it encounters in its
operating environment, leads to outcomes that maintain a stable (or
otherwise positive) relationship with that environment (McDonald,
2006, p. 157). Thus, the 'resilience' of the operational level needs to be
seen in the context of some form of monitoring, together with the
possibility of modification, change or development at a wider
T. Niskanen
collaborational level (McDonald, 2006, p. 168). Therefore, this study
generated the following hypothesis:
Hypothesis 4: For managers and workers “Cooperation between
employers and their workers” would be positively related with
“Decreasing accident risks” in the joint framework of RE.
4.4.1.6. The independent variable: “Systematic improvement of OSH” (will
be considered in Hypothesis 5). One aspect of RE is that it should monitor
organizational decision-making to identify the risk that the
organization may be operating nearer to systems safety boundaries
than realized (Woods, 2006a, p. 22). A socio-technical system cannot be
treated as a static entity, but rather as a dynamic process that is
continually adapting to achieve its ends and to react to changes in
itself and its environment (Leveson et al., 2006, pp. 96–97). Monitoring
resilience should lead to interventions to manage and adjust the
adaptive capacity as the system faces new forms of variation and
challenges of systems (Woods, 2006a, p. 22). The concept of resilience
has to work on at least three levels of the systems thinking(McDonald,
2006, p. 179): 1) the operation; 2) the organization; and 3) the
industrial system which designs and produces the technologies that
make the operation possible. Safety-related constraints specify those
relationships among system variables that constitute the non-hazardous
or safe system states (Leveson et al., 2006, p. 98). These data and the
arguments discussed above lead to the following hypothesis:
Hypothesis 5: For managers and workers “Systematic improvement
of OSH decreasing accident risks” would be positively related with
“Decreasing accident risks” in the joint framework of RE.
4.4.2. Review about the research results relating to how RE can be
integrated into Hypotheses H5–H10
The dependent variable “Development of the work conditions” (as it
114
Safety Science 101 (2018) 108–120
Fig. 2. “People and Performance”: the sub-cate-
gories of the taxonomy.
refers to Hypotheses: H5–H10)
RE emphasizes the need to be proactive in the management of core
processes, including safety, and to anticipate any major changes in
safety and other critical performance domains (Wreathall, 2011, p. 63).
Proactive adjustments mean that the system can change from a state of
normal operation to a state of heightened readiness, and possibly also
act, before something untoward happens (Hollnagel, 2011, p. 278).
4.4.2.1. The independent variable: “Obligations of the employers in risk
assessments” (will be considered in Hypothesis 6). The management
reviews the events through using its 'model' about obligations of the
processes, makes some changes and controls that this will suffice to
allow the business to continue (Wreathall, 2011, p. 63). In this context,
the models and analysis procedures can be used to investigate accidents
and incidents to determine the obligations of the different levels of
organizations. Another aspect is to learn with respect to obligations
how to prevent related accidents in the future, to proactively perform
hazard analysis and design so that it can reduce risk throughout the life
of the system. In this way, it should be possible to achieve a continuous
risk management program where risk is monitored and controlled
(Leveson et al., 2006, p. 106). The above discussion leads to the
following hypothesis:
Hypothesis 6: For managers and workers “Obligations of the em-
ployers in risk assessment” would be positively related with
“Development of the work conditions” in the joint framework of RE.
4.4.2.2. The independent variable: “Decreasing accident risks” (will be
evaluated in Hypothesis 7). These data and the arguments discussed in
this section lead to the following hypothesis:
Hypothesis 7. For managers and workers “Decreasing accident
risks” would be positively related with “Development of the work
T. Niskanen
conditions” in the joint framework of RE.
4.4.2.3. The independent variable: “Improving the ergonomics of work
methods and/or work equipment” (will be considered in Hypothesis 8). The
ability of adjusting performance is the key aspect of resilience, despite
the focus on individuals, teams, organizations or communities (Righi
et al., 2015). Furthermore, the safety of the organization cannot be
derived from a linear combination of the parts, but rather depends on
the ways in which they are coupled and how coordination across these
parts is fragmented or synchronized (Hollnagel and Woods, 2006, p.
351) e.g. in a system approach of ergonomics. Safety must be designed
into the human factors system, and safety when the system is operating
depends partly on the original system design and partly on ensuring
that there is effective control over operations (Leveson et al., 2006, p.
99). RE recognizes that complex system performance in human factors
is always variable, not only because of the variability of the
environment but also the fluctuations present within the constituent
subsystems (Hollnagel, 2006, p. 13). Based on the above mentioned
results, the following relationship was hypothesized:
Hypothesis 8: For managers and workers “Improving the ergo-
nomics of work methods and/or work equipment” would be posi-
tively related with “Development of the work conditions” in the
joint framework of RE.
4.4.2.4. The independent variable: “Decreasing mental overload” (to be
considered in Hypothesis 9). The tactical control mode corresponds to
situations where performance more or less follows a known and
articulated procedure or rule (Sundström and Hollnagel, 2006, p.
242) and also in this context has effects on the mental stress of
employees. The real challenge for system safety, and therefore also
115
Safety Science 101 (2018) 108–120
Fig. 3. “Equipment and Interfaces”: the sub-ca-
tegories of the taxonomy.
for RE, is to recognize that the processes involved in running complex
systems are dynamic and that a state of dynamic stability sometimes
may change into a state of dynamic instability (Hollnagel, 2006, p. 15)
e.g. into the mental overload. The potential solutions in human factors
issues tend to be more multifaceted (for example involving procedure,
operational process, training, etc.) and require possibly significant
changes in the operational system and the supporting organizational
and individual processes (McDonald, 2006, p. 171). The following
hypothesis was made:
Hypothesis 9: For managers and workers “Decreasing mental
overload at work” would be positively related with “Development of
the work conditions” in the joint framework of RE.
4.4.2.5. The independent variable: “Training and guiding the workers”
(issues which will be evaluated in Hypothesis H10). The challenge is to
make visible the gap between hazards and safety and provide a basis for
learning and adaptation where necessary (Dekker, 2006, p. 89). One
aspect of assessing a system's resilience is whether that system knows if
it is operating near to boundary conditions (Woods and Cook, 2006, p.
72). Assessing the margin is not a simple static state (the distance of an
operating point to a definitive boundary), but a more complex
assessment of adaptive responses to different kinds of disturbances
(Woods and Cook, 2006, p. 72) e.g. for training challenges. For effective
control, the training process models must have knowledge of the three
following factors (Leveson et al., 2006, p. 105): (1) the current state of
the system being controlled, (2) the required relationship between
system variables, and (3) the ways that the process can change the state.
The following hypothesis is put forward:
Hypothesis 10: For managers and workers “Training and guiding
 T. Niskanen

Table 1
For workers’ OSH representatives (Workers, n = 1244) and OSH managers (Managers, n = 1321), the parameter estimates in the regression analysis concerned with “Decreasing accident risks” (dependent variable) with respect to different
independent variables (1)-(5). In this model, for 'Workers’ the value of R-squared is 0.50 and for ‘Managers’ the corresponding value is 0.38. The meanings in the verification of the Hypotheses are as follows: S (=Supported) and NS (=Not
supported).

Dependent variable “Decreasing accident risks”

Independent variables Cronbach’s alpha Parameter Estimate Standard Error t-value p-value Tests

***
Intercept Workers 0.74 0.10 7.34 < 0.0001 S
Managers 0.88 0.12 7.24 < 0.0001*** S
H1 (1) Identification of OSH dangers Workers 0.81 −0.02 0.03 −0.76 0.4499 NS
(1) Systems and organization: work systems, rules and leadership; Managers 0.82 −0.01 0.03 −0.23 0.8217 NS
(2) People and Performance: communication, processes and competencies;
(3) Equipment and Interfaces: technology, procedure and performance.
H2 (2) Carrying out risk assessments Workers 0.80 0.04 0.03 1.42 0.1555 NS
(1) Systems and organization: work systems, rules and leadership; Managers 0.79 0.02 0.03 0.57 0.5695 NS
(2) People and Performance: communication, processes and competencies;

116
(3) Equipment and Interfaces: technology, procedure and performance.
H3 (3) The commitment of upper management Workers 0.94 0.10 0.03 3.48 0.0005 *** S
(1) Systems and organization: work systems, rules and leadership; Managers 0.93 0.13 0.03 4.30 < 0.0001 *** S
(2) People and Performance: communication, processes and competencies;
(3) Equipment and Interfaces: technology, procedure and performance.
***
H4 (4) Cooperation between employers and their workers Workers 0.94 0.12 0.03 3.70 0.0002 S
*
(1) Systems and organization: work systems, rules and leadership; Managers 0.93 0.07 0.03 2.17 0.0299 S
(2) People and Performance: communication, processes and competencies;
(3) Equipment and Interfaces: technology, procedure and performance.
**
H5 (5) Systematic improvement of OSH Workers 0.94 0.11 0.03 3.17 0.0016 S
*
(1) Systems and organization: work systems, rules and leadership; Managers 0.93 0.07 0.04 2.05 0.0403 S
(2) People and Performance: communication, processes and competencies;
(3) Equipment and Interfaces: technology, procedure and performance.

* p < 0.05;
** p < 0.01;
*** p < 0.001.
Safety Science 101 (2018) 108–120
T. Niskanen
the workers” would be positively related with “Development of the
work conditions” in the joint framework of RE.
5. Results
For workers’ OSH representatives and OSH managers, the parameter
estimates in the regression analysis are presented in Table 1 concerned
with the “Decreasing accident risks” (dependent variable) with respect
to the following independent variables: “(1) Identification of OSH
dangers”; “(2) Carrying out risk assessments”; “(3) The commitment of
upper management”; “(4) Cooperation between employers and their
workers”; and “(5) Systematic improvement of OSH Decreasing acci-
dent risks”. In this qualitative systems thinking analysis, the in-
vestigator selected in Table 1 the most important individual category
(bold), which has the most important influence in this taxonomy (see
the Section 4.3). In this context the sub-categories of the taxonomy in
Figs. 1–3 (Section 4.3.4) and the reasons (Sections 4.3.1–4.3.3) were
taken into account.
For example, with respect to both managers and workers,
“Systematic improvement of OSH” displayed a statistically positive
influence on “Decreasing accident risks” (Table 1). For “(1) Systems and
organization”, the “work systems” are evaluated by the investigator as
being the most important aspect; in this context three most important
sub-classes (see Fig. 1) are as follows: Socio-technical systems; Goal-
oriented; and Proactive. Furthermore, for “(2) People and Perfor-
mance”, the “processes” are of major importance; in this context three
most important sub-classes (see Fig. 2) are as follows: Procedures; Re-
lationships; and Control and interactions. Finally, for “(3) Equipment
and Interfaces”, the importance of “technology” is emphasized; in this
context three most important sub-classes (see Fig. 3) are as follows:
Socio-technical system; Expertise; and Flexibility.
For the statistical consistency of the responses, Cronbach’s alpha
was estimated as a coefficient of reliability or consistency. For the
“Decreasing accident risks” it is for managers 0.82 and for workers
0.80. For the “Independent variables” they are included in Table 1.
For workers’ OSH representatives and OSH managers, the parameter
estimates in the regression analysis are presented in Table 2 concerned
with the “Development of the work condition” (dependent variable)
with respect to the following independent variables: “(1) Obligations of
the employers in risk assessments”; “(2) Decreasing accident risks”; “(3)
Improving the ergonomics of work methods and/or work equipment”;
“(4) Decreasing mental overload ”; and “(5) Training and guiding the
workers”. In this qualitative systems thinking approach, the in-
vestigator has selected the most important individual category for use
in this taxonomy (see the Section 4.3). In this qualitative systems
thinking analysis, the investigator selected in Table 2 the most im-
portant individual category (bold and underlined), which has the most
important influence in this taxonomy (see the Section 4.3).In this
context the sub-categories of the taxonomy in Figs. 1–3 (Section 4.3.4)
and the reasons (Sections 4.3.1–4.3.3) were taken into account.
For example, with respect to both managers and workers, “Training
and guiding the workers” displayed a statistically positive influence on
“Development of the work conditions” (Table 2). For “(1) Systems and
organization”, the “rules” are evaluated by the investigator as being the
most important aspect; in this context three most important sub-classes
(see Fig. 1) are as follows: Legislation; Guidelines; and Collaboration.
Furthermore, for “(2) People and Performance”, the “competencies” are
of major importance; in this context three most important sub-classes
(see Fig. 2) are as follows: Knowledge; Application; and Exchange. Fi-
nally, for “(3) Equipment and Interfaces”, the importance of “perfor-
mance” is emphasized; in this context three most important sub-classes
(see Fig. 3) are as follows: Cognitive performance; Experience: and
Anticipation.
For the statistical consistency of the responses, Cronbach’s alpha
was estimated as a coefficient of reliability or consistency. For the
“Development of the work conditions” it is for managers 0.93 and for
117
Safety Science 101 (2018) 108–120
workers 0.94. For the “Independent variables” they are included in
Table 2.
6. Discussion
6.1. Theoretical implications
In this study, a combination of qualitative and quantitative research
methods was applied and a taxonomy was developed based on the RE
approach. For both managers and workers, H3–H5 were supported,
meaning that “Management” (H3), “Cooperation” (H4) and “Systems
thinking in OSH (H5)” decrease substantially accident risks. For
workers, H6–H10 were supported, meaning that “Obligations” (H6),
“Decreasing accident risks” (H7), “Improving Ergonomics” (H8),
“Decreasing mental overload (H9)” and “Training” (H10) develop
substantially work conditions; the corresponding Hypotheses supported
were for managers H7, H8 and H10.
With respect to data gathering, this study supports the findings of
Stringfellow (2010) who concluded that by conducting a “Taxonomy
assessment”, one can develop a causal factor taxonomy which can be
used as a framework for understanding how best to prevent accidents as
well as anticipating how system behavior may be adequately con-
trolled. Better understanding of systemic structures can then facilitate
the design of more effective safety culture interventions (Goh et al.,
2010). In addition, Stringfellow (2010, p. 110) found that a process that
can lessen the gap between the classification of human and organiza-
tional errors to prevent them from occurring is needed and even al-
though applying a taxonomy may be useful in accident analysis and
investigation, it is only part of the solution. Furthermore, Stringfellow
(2010) concluded that useful safety engineering methods must identify
how flaws in what he called the “mental models” of humans or orga-
nizations can occur and what will lead to mismatches between reality
and the mental model that adversely affects safety.
With respect to data analysis, this study is in accordance of a con-
structionist approach presented by several other investigators (e.g. Lord
and Dinh, 2012; Cunliffe, 2008; Fairhurst and Grant, 2010; Uhl-Bien
et al., 2012). Consistent with the social construction of reality (Lord and
Dinh, 2012), the constructionist approaches of “Taxonomy assessment”
emphasize the interactions themselves as well as the shared patterns of
meaning-making, conjoint agency, and coordinated behavior involved
in the work and interactions of managers and workers in organizations.
In this respect, the social interaction produces an understanding of
collective efficacy and creates a different construct, rather than a simple
aggregation of the organizational and individual-level attributes
(Cunliffe, 2008). Thus, the conscious experience is hypothesized to
exert a synergistic effect that can integrate constructs like current
perceptions, goals, self-structures, and experienced impacts to produce
a coherent entirety (applied from Fairhurst and Grant, 2010). Con-
structionism assumes that organizational reality is not separate for
different individuals but it is intimately interwoven and shaped by
everyday socio-technical interactions (applied from Uhl-Bien et al.,
2012).
With respect to applications of RE, the present results of “Taxonomy
assessment” support the conclusion of other investigators (e.g.
Hollnagel, 2006) that RE represents a valid cognitive engineering ap-
proach to modelling socio-technical system and an organization’s de-
cision making process. Furthermore, the present results support the
results of Tjeirhom and Aase (2011, p.161) that within “Taxonomy
assessment” upward resilience is also important because local micro-
level actors can create resilience in a system by applying their experi-
ence, flexibility, and professionalism to bridge the gap between rules
and procedures, and the actions required to adapt to new circum-
stances.
 T. Niskanen

Table 2
For workers’ OSH representatives (observations used for Workers, n = 1244) and OSH managers (Managers, n = 1321), the parameter estimates in the regression analysis concerned with the “Development of the work conditions” (dependent
variable) with respect to the independent variables (1)-(5). In this model, for ‘Workers’ the value of R-squared is 0.56 and for ‘Managers’ 0.44. The meanings in the verification of the Hypotheses are as follows: S (=Supported) and NS (=Not
supported).

Dependent: the variable “Development of the work conditions”

Independent variables Cronbach’s alpha Parameter Estimate Standard Error t-value p-value Tests

***
Intercept Workers 0.48 0.10 4.94 < 0.0001 S
Managers 1.05 0.10 10.06 < 0.0001*** S
H6 (1) Obligations of the employers in risk assessments Workers 0.94 0.06 0.02 2.29 0.0222* S
(1) Systems and organization: work systems, rules and leadership; Managers 0.93 0.07 0.03 2.91 0.0037** S
(2) People and Performance: communication, processes and competencies;
(3) Equipment and Interfaces: technology, procedure and performance.
H7 (2) Decreasing accident risks Workers 0.94 0.22 0.03 7.67 < 0.0001*** S
(1) Systems and organization: work systems, rules and leadership; Managers 0.93 0.13 0.03 4.68 < 0.0001*** S
(2) People and Performance: communication, processes and competencies;

118
(3) Equipment and Interfaces: technology, procedure and performance.
H8 (3) Improving the ergonomics of work methods and/or work equipment Workers 0.94 0.08 0.03 2.19 0.0285* S
(1) Systems and organization: work systems, rules and leadership; Managers 0.93 0.18 0.04 5.12 < 0.0001*** S
(2) People and Performance: communication, processes and competencies;
(3) Equipment and Interfaces: technology, procedure and performance.
H9 (4) Decreasing mental overload Workers 0.94 0.05 0.02 2.19 0.0285* S
(1) Systems and organization: work systems, rules and leadership; Managers 0.93 0.02 0.02 1.02 0.3087 NS
(2) People and Performance: communication, processes and competencies;
(3) Equipment and Interfaces: technology, procedure and performance.
H10 (5) Training and guiding the workers Workers 0.94 0.20 0.03 7.82 < 0.0001*** S
(1) Systems and organization: work systems, rules and leadership; Managers 0.93 0.14 0.03 5.07 < 0.0001 *** S
(2) People and Performance: communication, processes and competencies;
(3) Equipment and Interfaces: technology, procedure and performance.

* p < 0.05;
** p < 0.01;
*** p < 0.001.
Safety Science 101 (2018) 108–120
T. Niskanen
6.2. Practical implications
The present results indicated that “Systematic improvement of
OSH” was statistically significantly associated with “Decreasing acci-
dent risks” which emphasize the importance of applying RE approach in
the qualitative assessment of the interactions between work systems,
processes and technology. If one views the results from a systems’
perspective, this study supports the results of Malakis and Kontogiannis
(2011, p. 110) that monitoring strategies should enable the organiza-
tion, not only to 'catch' errors, but also to provide feedback for error
correction without hindering the flow of tasks in the organization.
The present results indicated that the quality of identification of
OSH dangers and carrying out risk assessments should be developed in
order that they have more influences on decreasing accident risks. The
present results are in accordance with Hollnagel (2006) that if an in-
vestigator models a resilient system, this should allow us to obtain data
about how well the system has adapted and how well operational sys-
tems are prepared in advance to handle different kinds of challenging
events. This study supports the findings of Righi et al. (2015) that the
classification of resilience helps to uncover where resilience is and how
it looks like and in this context helps to develop safety management.
This study supports the results of Righi et al. (2015) that key con-
structs found a deeper understanding of the commonalities, differences,
and relationships in the sub-categories of the developed taxonomy of
the RE approach. With respect to leadership and cooperation in “Tax-
onomy assessment”, this study supports the results of DeJoy et al.
(2010) who found that the sharpening focus on social exchange reflects
the recognition that workplace safety is part of the dynamic interaction
between employees and employers, and that employees react to both
the substance and the perceived intent of the employer’s actions (DeJoy
et al., 2010).
The present results of “Taxonomy assessment” support the results of
Woods and Branlat (2011, p. 129) that the capacity to respond to
challenging events resides in the expertise, strategies, tools, and plans
that managers and workers in various roles can deploy to prepare for
and respond to specific criteria of challenges. In addition, this study
supports the results of Tjeirhom and Aase (2011, p.160) that the con-
cept of downward resilience includes macro-level directions and solu-
tions conferring resilience e.g. through clear goal structures, infra-
structure, and procedures to handle tradeoffs. Safety culture is created
as a by-product of the systemic structure put in place unknowingly by
the joint actions of different parties in the system (Goh et al., 2010).
Thé present results of “Taxonomy assessment” support the results of
Lundberg and Johansson (2015) that with precise monitoring, the
system may exhibit resilience by the effective deployment of a ready-
built response strategy or through devising and deploying rapid and
flexible responses to counteract the event. In addition, a system may be
resilient with respect to all the facets of the model, alternatively the
evaluation may not only identify undeniable strengths in some areas
but also areas of vulnerability in others (Lundberg and Johansson,
2015).
6.3. Limitations
The current survey has several limitations that should be noted. The
qualitative method was based on the classification of the investigators.
Therefore, the present results may be subject to potential inaccuracies
related to the inability of the investigator to recall information cor-
rectly. Therefore, the results may be subject to potential inaccuracies
related to the inability of the investigator to recall information cor-
rectly. This study recognized that the RE approach represents a feasible
method. The epistemological and methodological choices make it much
more likely that other investigators scholars will “observe” the socio-
technical work environment possibly in a particular way. An in-
vestigator may omit important sources in a causal explanation of some
phases of these models, or it is possible to introduce an “omitted
119
Safety Science 101 (2018) 108–120
variable bias,” such that a misleading account is produced (Thies,
2002). The “bias” is a systematic error that is expected to occur in a
given context of research with respect to the present taxonomy models,
whereas “error” is generally taken to mean any difference between an
estimated value and the “true” value of a variable (applied from Collier
and Mahoney, 1996). Selection bias is commonly understood as oc-
curring when some form of selection process in either the design of the
study or the real-world phenomena under investigation results in in-
ferences that suffer from a systematic error (Collier and Mahoney,
1996). The present study makes the arguments explicit and allows the
readers to consider the generalization claim. On the basis of the de-
tailed contextual descriptions of the qualitative results, the reader has
to judge for him/herself whether the findings may be generalized to
novel situations. Kvale and Brinkmann (2009) stated that if the findings
of the qualitative study are judged to be reasonably reliable and valid,
the question still remains about whether the results are primarily of
local interest, or whether they may be transferable to other subjects and
different situations.
6.4. Suggestions for future research
Notwithstanding the above limitations, the present results have
implications for future research.
The present study introduced a new method for handling multibody
system data by applying a taxonomy data representation and has de-
monstrated its benefits in an example case. The study has left many
open questions for further research. The concept of using taxonomy
data representation was shown to be useful, but in future studies, a
questionnaire survey specifically designed for use with an RE system
model could be better designed by incorporating built-in structured
data. This study agrees with the other investigators (e.g. van der Beek
and Schraagen, 2015) that the research methods are an important phase
of any RE evaluation and future studies could evaluate resilience abil-
ities by creating a dedicated questionnaire. This RE-based ontology
would make it possible to have a hierarchical structuring of knowledge
about different aspects and factors by subcategorizing them according
to their essential qualities in processing and sharing meaning. The RE-
based ontology would consist not only of activities and properties (in-
cluding socio-technical activities and properties) but also interactions
(including actions and relations) between individuals, it could even
contain qualities related to the team’s and organization’s attributes
referred to in (a) Conceptual understanding of actions within RE, and
(b) Explanation of actions within RE. With respect to the RE research
design, this study will encourage other investigators to seek opportu-
nities for undertaking longitudinal research. For example, by under-
taking longitudinal designs and performing experimental RE interven-
tions, it should be possible to draw more convincing causal inferences
about RE. One would predict that a measurement of the RE excellence
could provide predictive validity.
Acknowledgements
The author is grateful for help with the online questionnaire to
Paula Naumanen, Hannu Kallio, Mika Liuhamo, Jorma Lappalainen,
Jarmo Sillanpää, Erkki Nykyri, Antti Zitting and Matti Hakkola of the
Finnish Institute of Occupational Health (FIOH) with respect the survey
regarding the different workplaces (i.e. the Finnish private business,
municipal and state sectors). I would like to thank also Maria L.
Hirvonen from the FIOH for her assistance in data analysis.
The author is very grateful to the anonymous reviewer for the im-
portant comments, which improved substantially the quality of the
paper.
References
Amalberti, R., 2006. Optimum system safety and optimum system resilience: agonistic or
T. Niskanen
antagonistic concepts? In: Hollnagel, E., Woods, D.D., Leveson, N. (Eds.), Resilience
Engineering: Concepts and Precepts. Ashgate Publishing Limited, Aldershot, pp.
253–274.
Bergström, J., Dahlström, N., Dekker, S., Petersen, P., 2011. Training organizational re-
silience in escalating situations. In: Hollnagel, E., Paries, J., Woods, D., Wreathall, J.
(Eds.), Resilience Engineering in Practice: A Guidebook. Ashgate Publishing Limited,
Surrey, pp. 45–57.
Collier, D., Mahoney, J., 1996. Insights and pitfalls: selection bias in qualitative research.
World Politics 49 (1), 56–91. http://www.jamesmahoney.org/articles/Insights
%20and%20Pitfalls.pdf (accessed 17.01.17).
Cook, R.I., Nemeth, C.P., 2006. Taking things in one's stride: cognitive features of two
resilient performances. In: Hollnagel, E., Woods, D.D., Leveson, N. (Eds.), Resilience
Engineering: Concepts and Precepts. Ashgate Publishing Limited, Aldershot, pp.
205–221.
Creswell, J.W., Plano Clark, V.L., 2007. Designing and conducting mixed methods re-
search. Sage Publications, Thousand Oaks, CA.
Cronbach, L.J., Shavelson, R.J., 2004. My current thoughts on coefficient alpha and
successor procedures. Educ. Psychol. Measur. 64 (3), 391–418.
Cunliffe, A., 2008. Orientations to social constructionism: relationally responsive social
constructionism and its implications for knowledge and Learning. Manage. Learn. 39
(2), 123–139.
Dejoy, D.M., Della, L.J., Vandenberg, R.J., Wilson, M.G., 2010. Making work safer: testing
a model of social exchange and safety management. J. Safety Res. 41, 163–171.
Dekker, S.W.A., 2006. Resilience engineering: chronicling the emergence of confused
consensus. In: Hollnagel, E., Woods, D.D., Leveson, N. (Eds.), Resilience Engineering:
Concepts and Precepts. Ashgate Publishing Limited, Aldershot, pp. 77–92.
Fairhurst, G., Grant, D., 2010. The social construction of leadership: a sailing guide.
Manage. Commun. Quarterly 24 (2), 171–210.
Goh, Y.M., Brown, H., Spickett, J., 2010. Applying systems thinking concepts in the
analysis of major incidents and safety culture. Saf. Sci. 45 (3), 302–309.
Hale, A., Guldenmund, F., Goossens, L., 2006. Auditing resilience in risk control and
safety management systems. In: Hollnagel, E., Woods, D.D., Leveson, N. (Eds.),
Resilience Engineering: Concepts and Precepts. Ashgate Publishing Limited,
Aldershot, pp. 289–314.
Hale, A., Heijer, T., 2006. Defining resilience. In: Hollnagel, E., Woods, D.D., Leveson, N.
(Eds.), Resilience Engineering: Concepts and Precepts. Ashgate Publishing Limited,
Aldershot, pp. 35–40.
Hollnagel, E., 2006. Resilience: the challenge of the unstable. In: Hollnagel, E., Woods,
D.D., Leveson, N. (Eds.), Resilience Engineering: Concepts and Precepts. Ashgate
Publishing Limited, Aldershot, pp. 9–17.
Hollnagel, E., Woods, D.D., 2006. Epilogue: resilience engineering precepts. In: Hollnagel,
E., Woods, D.D., Leveson, N. (Eds.), Resilience Engineering: Concepts and Precepts.
Ashgate Publishing Limited, Aldershot, pp. 347–358.
Hollnagel, E., 2011. Epilogue. In: Hollnagel, E., Paries, J., Woods, D., Wreathall, J. (Eds.),
Resilience engineering in practice: a guidebook. Ashgate Publishing Limited, Surrey,
pp. 275–296.
Kvale, S., Brinkmann, S., 2009. Interviews: learning the craft of qualitative research in-
terviewing, second ed. Sage Publications, London.
Leveson, N., Dulac, N., Zipkin, D., Cutcher-Gershenfeld, J., Carroll, J., Barrett, B., 2006.
Engineering resilience into safety-critical systems. In: Hollnagel, E., Woods, D.D.,
Leveson, N. (Eds.), Resilience Engineering: Concepts and Precepts. Ashgate
Publishing Limited, Aldershot, pp. 96–123.
Lord, R.G., Dinh, J.E., 2012. Aggregation processes and levels of analysis as organizing
structures for leadership theory. In: Day, D.V., Antonakis, J. (Eds.), The Nature of
Leadership, second edition. Sage Publications Inc., Thousand Oaks, CA, pp. 29–65.
Lundberg, J., Johansson, B.J.E., 2015. Systemic resilience model. Reliab. Eng. Syst.
Safety. 41(Special Issue on Resilience Engineering, September 2015), 22–32.
Malakis, S., Kontogiannis, T., 2011. Cognitive strategies in emergency and abnormal si-
tuations training – implications for resilience in air traffic control. In: Hollnagel, E.,
Paries, J., Woods, D., Wreathall, J. (Eds.), Resilience engineering in practice: a
guidebook. Ashgate Publishing Limited, Surrey, pp. 101–116.
McDonald, N., 2006. Organizational resilience and industrial risk. In: Hollnagel, E.,
120
Safety Science 101 (2018) 108–120
Woods, D.D., Leveson, N. (Eds.), Resilience Engineering: Concepts and Precepts.
Ashgate Publishing Limited, Aldershot, pp. 155–180.
Onwuegbuzie, A.J., Leech, N.L., 2005. On becoming a pragmatic researcher: the im-
portance of combining quantitative and qualitative research methodologies. Int. J.
Soc. Res. Methodol. 8 (5), 375–387.
Onwuegbuzie, A.J., Teddlie, C., 2003. A framework for analyzing data in mixed methods
research. In: Tashakkori, A., Teddlie, C. (Eds.), Handbook of Mixed Methods in Social
and Behavioral Research. Sage, Thousand Oaks, CA, pp. 351–383.
Paries, J., 2011. Lessons from the Hudson. In: Hollnagel, E., Paries, J., Woods, D.,
Wreathall, J. (Eds.), Resilience engineering in practice: a guidebook. Ashgate
Publishing Limited, Surrey, pp. 9–27.
Reynaldo, J., Santos, A., 1999. Cronbach’s alpha: a tool for assessing the reliability of
scales. J. Extension 37 (2), 11–14.
Righi, A.W., Saurin, T.A., Wachs, P., 2015. A systematic literature review of resilience
engineering: Research areas and a research agenda proposal. Reliab. Eng. Syst. Safety
41(Special Issue on Resilience Engineering, September 2015), 142–152.
SAS Institute Inc., 2005. SAS user’s guide. Version 9. Cary: SAS Institute Inc.
Sundström, G., Hollnagel, E., 2006. Learning how to create resilience in business systems.
In: Hollnagel, E., Woods, D.D., Leveson, N. (Eds.), Resilience Engineering: Concepts
and Precepts. Ashgate Publishing Limited, Aldershot, pp. 235–252.
Stringfellow, M.V., 2010. Accident analysis and hazard analysis for human and organi-
zational factors. Dissertation. Thesis for the degree of Doctor of Philosophy.
Massachusetts Institute of Technology. http://sunnyday.mit.edu/safer-world/
MaggieStringfellowDissertation.pdf (accessed 9.12.16).
Thies, C.G., 2002. A pragmatic guide to qualitative historical analysis in the study of
international relations. Int. Stud. Perspect. 3 (4), 351–372.
Tjeirhom, B., Aase, K., 2011. The art of balance: using upward resilience traits to deal
with conflicting goals. In: Hollnagel, E., Paries, J., Woods, D., Wreathall, J. (Eds.),
Resilience Engineering in Practice: A Guidebook. Ashgate Publishing Limited, Surrey,
pp. 157–170.
Uhl-Bien, M., Maslyn, J., Ospina, S., 2012. The nature of relational leadership: a multi-
theoretical lens on leadership relationships and processes. In: Day, D.V., Antonakis, J.
(Eds.), The Nature of Leadership, second ed. Sage Publications Inc., Thousand Oaks,
California, pp. 289–330.
van der Beek, D., Schraagen, J.M., 2015. ADAPTER: Analysing and developing adapt-
ability and performance in teams to enhance resilience. Reliab. Eng. Syst. Safety.
41(Special Issue on Resilience Engineering, September 2015), 33–44.
Wilson, J.R., 2014. Fundamentals of systems ergonomics/human factors. Appl. Ergonom.
45 (1), 5–13 (Special Issue: Systems Ergonomics/Human Factors).
Woods, D.D., 2015. Four concepts for resilience and the implications for the future of
resilience engineering. Reliability Engineering & System Safety 41(Special Issue on
Resilience Engineering, September 2015), 5–9.
Woods, D.D., 2009. Escaping failures of foresight. Saf. Sci. 47 (4), 498–501.
Woods, D.D., 2006a. Essential characteristics of resilience. In: Hollnagel, E., Woods, D.D.,
Leveson, N. (Eds.), Resilience Engineering: Concepts and Precepts. Ashgate
Publishing Limited, Aldershot, pp. 21–33.
Woods, D.D., 2006b. How to design a safety organization: Test case for resilience en-
gineering. In: Hollnagel, E., Woods, D.D., Leveson, N. (Eds.), Resilience Engineering:
Concepts and Precepts. Ashgate Publishing Limited, Aldershot, pp. 315–324.
Woods, D.D., Branlat, M., 2011. Basic patterns in how adaptive systems fail. In: Hollnagel,
E., Paries, J., Woods, D., Wreathall, J. (Eds.), Resilience Engineering in Practice: A
Guidebook. Ashgate Publishing Limited, Surrey, pp. 127–143.
Woods, D.D., Cook, R.I., 2006. Incidents - markers of resilience or brittleness? In:
Hollnagel, E., Woods, D.D., Leveson, N. (Eds.), Resilience Engineering: Concepts and
Precepts. Ashgate Publishing Limited, Aldershot, pp. 69–76.
Wreathall, J., 2006. Properties of resilient organizations: an initial view. In: Hollnagel, E.,
Woods, D.D., Leveson, N. (Eds.), Resilience Engineering: Concepts and Precepts.
Ashgate Publishing Limited, Aldershot, pp. 275–285.
Wreathall, J., 2011. Monitoring – a critical ability in resilience engineering. In: Hollnagel,
E., Paries, J., Woods, D., Wreathall, J. (Eds.), Resilience Engineering in Practice: A
Guidebook. Ashgate Publishing Limited, Surrey, pp. 61–68.