10 1016@j JSR 2019 09 015 PDF

JSR 1622 No.
of Pages 14, Model 5G

12 November 2019
Journal of Safety Research xxx (xxxx) xxx

1
Contents lists available at ScienceDirect
Journal of Safety Research

journal homepage: www.elsevier.com/locate/jsr
5
6
3 A comparative analysis of safety management and safety performance

4 in twelve construction projects
7 Stig Winge a,⇑, Eirik Albrechtsen a, Jan Arnesen b
8 a
Norwegian University of Science and Technology (NTNU), Department of Industrial Economics and Technology Management, NO-7491, Trondheim, Norway
9 b
Statsbygg, NO-0155, Oslo, Norway
10
11
a r t i c l e i n f o a b s t r a c t
1
2 3
9
14 Article history: Introduction: Safety management in construction is complicated due to the complex ‘‘nature” of the con- 30
15 Received 6 February 2019 struction industry. The aim of this research was to identify safety management factors (e.g., risk manage- 31
16 Received in revised form 26 June 2019 ment and site management), contextual factors (e.g., organisational complexity) and combinations of 32
17 Accepted 26 September 2019
such factors connected to safety performance. 33
18 Available online xxxx
Method: Twelve construction projects were selected to compare their safety management and safety per- 34
formance. An analytical framework was developed based on previous research, regulations, and stan- 35
19 Keywords:
dards where each management factor was defined. We employed qualitative comparative analysis 36
20 Occupational health and safety
21 management
(QCA) to produce case knowledge, compare the cases, and identify connections between the factors 37
22 Safety performance and safety performance. The material collected and analyzed included, for example, construction plan- 38
23 Construction safety ning documents, reports from OHS-inspections, safety indicators, and interviews with project leaders 39
24 Construction project and OHS experts. 40
25 Comparative methods Results and conclusions: The research showed that: (a) the average score on 12 safety management factors 41
26 Qualitative comparative analysis was higher among projects with high safety performance compared to projects with low safety 42
27 Causal complexity performance; (b) high safety performance can be achieved with both high and low construction 43
28
complexity and organizational complexity, but these factors complicate coordination of actors and oper- 44
ations; (c) it is possible to achieve high safety performance despite relatively poor performance on many 45
safety management factors; (d) eight safety management factors were found to be ‘‘necessary” for high 46
safety performance, namely roles and responsibilities, project management, OHS management and inte- 47
gration, safety climate, learning, site management, staff management, and operative risk management. 48
Site management, operative risk management, and staff management were the three factors most 49
strongly connected to safety performance. 50
Practical implications: Construction stakeholders should understand that the ability to achieve high safety 51
performance in construction projects is connected to key safety management factors, contextual factors, 52
and combinations of such factors. 53
Ó 2019 National Safety Council and Elsevier Ltd. All rights reserved. 54
55
56
57
58 1. Introduction 2013). There is little research on the effect of safety management 67
systems and programs on safety performance in construction. 68
59 The construction industry in Europe (EU-28) had the highest There is, however, some research that has identified factors poten- 69
60 share of fatal occupational accidents in 2015, with more than one tially connected to safety performance in construction, for example 70
61 in five accidents (Eurostat, 2018). Safety management in construc- management commitment (Loushine, Hoonakker, Carayon, & 71
62 tion is demanding, since construction projects are technologically Smith, 2006), subcontractor selection and management 72
63 and organizationally complex (Lingard, 2013). (Hallowell & Gambatese, 2009), worker involvement (Chen & Yin, 73
64 Previous research on the effectiveness of occupational health 2013), interrelations between various project partners (Terwel & 74
65 and safety management (OHSM) on safety performance is ambigu- Jansen, 2014), site-specific safety plans (Hallowell & Calhoun, 75
66 ous, and the importance of different factors are debated (Zwetsloot, 2011), and safety culture (Choudhry, Fang, & Mohamed, 2007). 76
The aim of this research is to identify how safety management 77
factors, contextual factors, and combinations of such factors influ- 78
⇑ Corresponding author. ence safety performance. To do that, we (a) developed an analytical 79
E-mail address: stig.winge@arbeidstilsynet.no (S. Winge).
https://doi.org/10.1016/j.jsr.2019.09.015
0022-4375/Ó 2019 National Safety Council and Elsevier Ltd. All rights reserved.
Please cite this article as: S. Winge, E. Albrechtsen and J. Arnesen, A comparative analysis of safety management and safety performance in twelve con-
struction projects, Journal of Safety Research, https://doi.org/10.1016/j.jsr.2019.09.015
JSR 1622 No. of Pages 14, Model 5G
12 November 2019
2 S. Winge et al. / Journal of Safety Research xxx (xxxx) xxx
80 framework iteratively based on relevant literature and empirical level; and, (e) feedback and monitoring loops ensuring assessment 143
81 results; (b) analyzed documents (e.g., health and safety plans, against performance indicators at each of the three levels. 144
82 inspection reports), safety indicators, and interviewed project lead- The research on the effectiveness of OHSMSs on safety perfor- 145
83 ers and OHS inspectors from 12 construction projects; and (c) mance is, however, ambiguous. In a literature review, Gallagher, 146
84 assessed factors and combinations of factors connected to safety Rimmer and Underhill (2001) concluded that OHSMSs can deliver 147
85 performance employing Qualitative comparative analysis (QCA) more healthy and safe workplaces under the right circumstances. 148
86 (Ragin, 1987, 2008). The study was performed in cooperation with In another review, Robson et al. (2007) concluded that the body 149
87 Statsbygg, a Norwegian government client organization who build of evidence was insufficient to make recommendations either in 150
88 and rehabilitate public buildings. favor of or against OHSMSs. In a review of the effectiveness of 151
89 Construction contractors have traditionally been held responsi- safety management systems (SMS), Thomas (2011) concluded that 152
90 ble for OHS on construction sites (Lingard & Rowlinson, 2005). In organizations with a certified SMS had significantly lower accident 153
91 1992, the EU Construction Sites Directive (92/57/EEC) (European rates. There was, however, a lack of agreement about which com- 154
92 Commission, 1992, 2011) put heavy responsibility for OHS on the ponents of a safety management system contributed the most to 155
93 client. The motivation for the directive was the recognition that safety performance. Zwetsloot (2013) argue that the difficulties 156
94 many occupational accidents had been attributed to unsatisfactory in demonstrating the effects of OHSMS on safety performance 157
95 architectural and/or organizational options, poor planning, and can be explained that many do not consider ‘‘contextual factors” 158
96 inadequate co-ordination (Berger, 2000). Clients can play a positive like the ambitions and commitment of management, the participa- 159
97 role in safety management during construction projects (Huang & tion of workers, and the continual adaptation to changing circum- 160
98 Hinze, 2006; Lingard, Oswald, & Le, 2018; Spangenberg, 2010). This stances. Zwetsloot (2013) also argues that the system is more than 161
99 paper focuses on safety management in construction projects pri- the sum of its parts and the interactions between the elements are 162
100 marily from a client’s perspective. just as important as the elements. 163
A literature review of 49 studies of safety management and 164
quality management in construction projects supported the use 165
101 2. Literature on safety management of integrated safety and quality management (Loushine et al., 166
2006). The characteristics found to contribute to improved con- 167
102 Safety management can be defined as ‘‘the process to realize struction safety were management commitment, employee 168
103 certain safety functions,” and a safety management system (SMS) involvement, a formal safety management program, training, 169
104 is commonly defined as ‘‘. . . the management procedures, elements audits and observations, continuous improvement, and 170
105 and activities that aim to improve the safety performance of and communication. 171
106 within an organization” (Li & Guldenmund, 2018, p. 96). An occu- Hallowell and Gambatese (2009) identified from previous 172
107 pational health and safety management system (OHSMS) is defined research 13 critical elements of an effective construction safety 173
108 as ‘‘A set of interrelated or interacting elements to establish OSH program: a written and comprehensive safety and health plan, 174
109 policy and objectives, and to achieve those objectives” (ILO, 2001, upper management support, job hazard analyses and hazard com- 175
110 p. 19). The purpose of an OHSMS is to ‘‘provide a framework for munication, safety and health orientation and training, frequent 176
111 managing OH&S risks and opportunities” (ISO, 2019, p. vi). In this worksite inspections, emergency response planning, record keep- 177
112 research we use the term safety management to include both speci- ing and accident analyses, project-specific training and regular 178
113 fic safety management factors as well as general management fac- safety meetings, safety and health committees, substance abuse 179
114 tors that can influence safety performance. programs, safety manager on site, subcontractor selection and 180
115 Tinmannsvik and Hovden (2003) distinguish between safety management, and employee involvement safety and evaluation. 181
116 specific factors (management factors mainly to promote safety) Hallowell and Calhoun (2011) quantified the interrelationships 182
117 and general management factors (management factors to improve between these 13 elements and concluded that the most central 183
118 the production system and organization in general). In this elements in an effective program were the site safety manager, 184
119 research we include both safety specific factors as well as other worker participation and involvement, a site-specific safety plan, 185
120 management factors with a potential of influencing safety perfor- and upper management support and commitment. Another impor- 186
121 mance, like, for example, staff management and project manage- tant conclusion was that many of the strategies found to be effec- 187
122 ment. This is in line with Hale (2003b), who argues that safety tive in isolation also provided a high level of synergistic effects that 188
123 management should be seen as an aspect system, not a subsystem, enhance the effectiveness of other elements. 189
124 of the organization. In a review of 90 papers, Mohammadi, Tavakolan, and Khosravi 190
125 The literature reviewed use different terms – safety manage- (2018) identified 13 factors influencing safety performance in con- 191
126 ment (SM), occupational health and safety management (OHSM), struction: motivation, rules and regulation, competency, safety 192
127 and safety programs. In this research we use the term safety man- investment and costs, financial aspects and productivity, resource 193
128 agement to include all these terms. Risk control and safety manage- and equipment, work pressure, work condition, culture and cli- 194
129 ment is often seen in terms of a hierarchy of system levels (Hale, mate, attitude and behavior, lesson learned from accidents, organi- 195
130 2003a; Rasmussen, 1997; Reason, 1997). The operational work is zation, and safety programs and management systems. They also 196
131 at the lowest level, where the accidents happen, being controlled concluded that safety performance is not only determined by man- 197
132 by technology and human behavior, which is in turn controlled agement activities within project levels, but also by the interac- 198
133 by management provision of resources, information, and instructions among factors at different hierarchical levels. 199
134 tion. These in turn are influenced by policy, regulation, market The ISO 45001 (Occupational health and safety management 200
135 and other societal forces (Hale, Guldenmund, Van Loenhout, & systems, ISO, 2019) also states that effectiveness and the ability 201
136 Oh, 2010). to achieve the outcomes of an OHS system are dependent on a 202
137 Hale (2003a, 2002) argues that we know ‘‘quite securely” the number of key factors, for example top management leadership 203
138 structure of a good safety management system, including: (a) an and commitment, communication, consultation and participation 204
139 anchorage to the specific hazards of the production; (b) a life cycle of workers, allocation of necessary resources, risk management, 205
140 approach; (c) problem-solving at three levels (operational, tactical, continual performance evaluation and monitoring, integration of 206
141 strategic); (d) systems at the tactical level delivering the crucial the OHSM system into the organization’s business processes (ISO, 207
142 resources and controls for safety-critical tasks at the operational 2019). 208
12 November 2019
S. Winge et al. / Journal of Safety Research xxx (xxxx) xxx 3
209 Summarized, the research on safety management in construc- Table 1

210 tion shows several factors connected to safety performance. Some Operational definitions of outcome and factors.
211 studies also show that certain combinations of factors increase the Name Description
212 effect on safety performance. Based on a literature review, OUTCOME:
213 Mohammadi et al. (2018) suggested that more research is needed Safety performance (SP) Assessment of the overall safety at site based on:
214 to investigate the interaction between the identified factors and (1) interviews with OHS-inspectors about their
215 determine how they are able to affect safety performance, which assessments of the relative extent of hazards and
dangerous situations; (2) interviews with the
216 is one of the aims of this research. client project leaders about their assessments of
the relative extent of hazards and dangerous
situations; (3) reports from audits/inspections;
217 3. Material and methods (4) analysis of all registered dangerous situations
(RUOs and SDs); and (5) the total recordable
218 3.1. Analytical framework injury rate (TRI-rate)
CONDITIONS /FACTORS:
219 In qualitative comparative analysis (QCA), explanatory models 1. Construction The characteristics and inherent complexity of
220 are developed in an iterative manner to facilitate a dialogue complexity (CC) the project, the structure being constructed
(buildability), location, and physical restrictions
221 between theory and evidence as described by Ragin (2014). The of the site.
222 analytical framework (Table 1) was therefore developed in an iter- 2. Organisational The extent of use of subcontractors, other
223 ative manner and employed for measuring and comparing safety complexity (OC) companies and hired workers relative to project
224 performance and safety management in 12 construction projects. size
3. Time (TI) Progress plans, time pressure and delays
225 A preliminary framework was developed based on previous
4. Economy (EC) Whether the project was on budget, and whether
226 research, regulations, and standards. Important sources were: (1) contractors made money
227 The ConAC framework (Haslam et al., 2005), which we had used 5. Contract management Contracting strategy, contract type, cooperation
228 to identify deficiencies in management factors (and other factors) (CO) between client and contractors, and the
229 in construction accidents (Winge & Albrechtsen, 2018); (2) Hale, contractor’s commitment to OHS
6. OHS-planning (PL) Whether OHS was part of project planning and
230 Walker, Walters, and Bolt (2012), who developed an analytical activities: Adequate SH-plan communicated to all
231 framework for understanding underlying causes of construction actors and regularly updated; assessment of risks
232 fatal accidents; (3) Törner and Pousette (2009) who carried out in advance with specific measures; and progress
233 an inductive, qualitative interview study of experienced workers plans
7. Roles & responsibilities The presence, clarity and performance of roles
234 (worker safety representatives) and first-line supervisors in con-
(RO) central to OHS (client, principal enterprise,
235 struction to identify ‘‘preconditions and components of high safety coordinators for the planning stage and execution
236 standards” in the construction industry; (4) The Directive 92/57/ stage, HS staff).
237 EEC (European Commission, 1992) and the Norwegian version, 8. Project management Coordination, cooperation, communication and
238 the Construction Client Regulations (Directorate of Labour (PM) follow-up of actors on OHS
9. Management Commitment to OHS by managers (client and
239 Inspection, 2009), which specify key elements for OHS manage- commitment to OHS contractors) and emphasis on and integration of
240 ment systems in construction; (5) The OHSM system standards (MC) safety management with project management.
241 by ILO (2001) and ISO (2019). Many of the detailed analytical ques- 10. Safety climate (SC) Attitudes, communication, openness and trust
242 tions underlying the relatively broadly defined factors in the regarding OHS
11. Learning (LE) Learning from incidents, accidents and deviations
243 framework were adapted from the safety management and organi-
through reporting, safety walks and inspections.
244 zation review technique (SMORT), which was originally based on 12. Performance Ability to evaluate OHS performance and
245 the management oversight and risk tree MORT (Johnson, 1980). evaluation (PE) implement measures
246 SMORT was originally published by Kjellén, Tinmannsvik, 13. Operative risk Operative risk management by people in direct
247 Ulleberg, Olsen, and Saxvik (1987) and is later revised based on management (RM) control of the risk at the operational level
(planning of operations to reduce risk)
248 more recent experiences and standards (Kjellén & Albrechtsen, 14. Site management (SI) Site organisation, storage, logistics, housekeeping
249 2017). and provision of physical barriers
250 The preliminary analytical framework consisted of 18 main cat- 15. Staff management Planning to ensure the availability of sufficient
251 egories and 83 subcategories. The framework was then tested on (SM) workers with adequate capacity that is
competent and suitable. Supervision and follow
252 the documentation collected (see Sections 3.4 and 3.5) from eight
up of safety behaviour (short-cuts and
253 projects and revised. Then the framework was tested as an inter- compliance) on site
254 view guide for semi structured ‘‘pilot interviews” with three pro- 16. Hardware Availability, condition, usability and suitability of
255 jects leaders (for client) and revised to the final version with the management (HA) materials and equipment
256 16 categories displayed in Table 1.
257 The factors in the framework can be divided into different cat-
258 egories. Since we focus primarily on the execution stage of con-
p. 192). We employed Qualitative comparative analysis (QCA) 269
259 struction projects, factors 1–4 are treated as ‘‘contextual factors.”
(Ragin, 1987, 2008) to identify conditions and combinations of 270
260 They are to some extent ‘‘contextual” factors and/or decisions
conditions connected to safety performance. QCA uses the terms 271
261 made at an early stage. Factor 5 (contract management) can be
‘‘condition” (causal factor), outcome, and connections (associa- 272
262 seen as both as a contextual factor and a safety management factor.
tions). In this research we use the term condition when using 273
263 Factors 6–16 represent the safety management process of the pro-
QCA, and otherwise use the term factor since that is a term used 274
264 ject. Factors 13–16 to large extent manage the workplace – the
in most of the safety literature studied. 275
265 ‘‘sharp end.”
QCA is a methodological approach for comparing cases, produc- 276
ing case knowledge, and identifying associations between condi- 277
266 3.2. Qualitative comparative analysis (QCA) tions and the outcome. QCA strives to meet two apparently 278
contradictory goals of in-depth insight into cases and complexity, 279
267 Hale (2003a) argued that we need to do comparative studies of and the production of generalizations (Ragin, 1987). Comparative 280
268 good and bad companies to see ‘‘. . . what features are crucial” (ibid. studies of ‘‘good” and ‘‘bad” construction projects is also an 281
12 November 2019
282 opportunity to study both what goes right (safety I) and what goes are therefore not ‘‘representative” for construction projects, and 346
283 wrong (safety II) in safety management in construction projects the projects are not representative of Statsbygg’s projects, since 347
284 (Hollnagel, 2014). most Statsbygg projects have a high safety performance. 348
285 QCA is a set-theoretic approach where concepts are understood
286 as sets in which cases have membership. There are two types of
287 sets. Crisp sets allow only full membership (1) and full non-
3.4. Measuring safety performance 349
288 membership (0). Fuzzy sets allow for partial membership in addi-
289 tion to full membership and non-membership where the point of
Oswald, Zhang, and Lingard (2018) argue that great care needs 350
290 maximum ambiguity (fuzziness) is 0.5. A fuzzy set can be seen as
to be taken when using safety indicators to evaluate organizational 351
291 a continuous variable that has been calibrated to indicate degree
safety policy and practices. Common health and safety indicators 352
292 of membership. Researchers must use substantive and theoretical
can, for example, be subject to manipulation and misinterpreta- 353
293 knowledge to calibrate membership.
tion. In this research we found indications that some injuries that 354
294 The approach is based on a notion of causal complexity, where
should have been registered as lost time injuries (LTIs) were regis- 355
295 outcomes are produced by combinations of conditions (configura-
tered as medical treatment injuries (MTIs). LTI-rate was therefore 356
296 tions), and that different configurations can produce a similar out-
assessed to be a relatively unreliable indicator since systems of 357
297 come (equifinality). QCA is very well suited to researching
registration were different across the projects studied, in addition 358
298 complexity (Gerrits & Verveij, 2018). When QCAs are undertaken,
to the other weaknesses of the LTI-rate (see Kjellén & 359
299 we look for conditions that are necessary parts of a combination
Albrechtsen, 2017). The total recordable injury rate (TRI-rate) 360
300 of conditions (configurations). A condition (X) is necessary if,
was assessed to be the only reliable quantitative injury indicator 361
301 whenever the outcome (Y) is present across cases, the condition
with which to indicate safety performance. TRI-rate is more robust 362
302 (X) is also present. We also look for configurations that are suffi-
than LTI-rate since the number of injuries is higher. TRI-rate 363
303 cient to explain the outcome. A condition or configuration is suffi-
includes mostly less severe injuries and is therefore primarily an 364
304 cient if, whenever it is present across cases, the outcome is also
indicator of the presence of less severe hazards and occurrences. 365
305 present (If X, then Y). There can, however, be several configurations
The data collection took place after the projects were finished, 366
306 that are sufficient for the outcome (equifinality). Sufficient condi-
or in some cases, more than halfway, making it possible to assess 367
307 tions or configurations can produce the outcome alone, but there
results from most of the construction period. It was therefore not 368
308 can also be other conditions/configurations with this ability.
practicable to use safety climate surveys or other leading indica- 369
309 QCA has gained increased in popularity in recent decades, espe-
tors. Because of limitations in the safety indicators, and difficulties 370
310 cially in the disciplines of comparative politics, business and econ-
using more leading indicators, we chose to do a researcher-based 371
311 omy, sociology, and management and organization (Roig-Tierno,
assessment of safety performance based on five sources 372
312 Gonzalez-Cruz, & Llopis-Martinez, 2017). As QCA is a relatively
(triangulation): 373
313 new technique, we explain its basic logic and steps (for a detailed
314 treatment, see Ragin, 2008; Schneider & Wagemann, 2012). Data
1. The total recordable injury rate (TRI-rate). 374
315 analysis was performed using the fsQCA 3.0 for Windows software
2. Analysis of all registered dangerous situations (RUOs and SDs). 375
316 (Ragin & Davey, 2017) and its software manual (Ragin, 2017).
3. Reports from OHS audits/inspections (see 3.5.2). 376
4. Interviews with client project leaders about their assessments 377
317 3.3. Case selection
of the extent of hazards and dangerous situations relative to 378
the project size. 379
318 The study was performed in cooperation with Statsbygg, a gov-
5. Interviews with OHS-inspectors about their assessments of the 380
319 ernment client organization responsible to the Norwegian Ministry
extent of hazards and dangerous situations relative to the pro- 381
320 of Local Government and Modernisation (KMD). Statsbygg build
ject size. 382
321 and rehabilitate state public buildings, such as court buildings,
383
322 prisons, museums and university buildings. Statsbygg is actively
Table 2 describes the indicators used. We interpret the number 384
323 involved in safety management in projects with project staff pre-
of ‘‘registered dangerous situations” (RUO & SD) primarily as an 385
324 sent at the site and following up production and OHS regularly.
indicator of willingness to report and tackle safety issues, not as 386
325 The sampling was carried out in dialogue with OHS experts in
an indicator of high levels of danger (Hale et al., 2010). 387
326 Statsbygg based on their familiarity with projects. The cases were
327 selected based on three criteria: (1) Projects initially assessed to
328 have relatively high or low safety performance were selected, Table 2
329 because it is advantageous to include cases with a ‘‘positive” or a Materials, numbers and indicators used to assess safety performance and willingness
330 ‘‘negative” outcome in comparative methods (Berg-Schlosser, De to report.
331 Meur, Ragin, & Rihoux, 2009); (2) Projects relatively similar in size Abbreviations Description
332 (working hours), building type, and contractual arrangements WH Working hours registered by main contractor, subcontractors
333 were selected to keep these conditions as constant as possible; and hired workers. Working hours by designers not included
334 (3) Projects that were finished or more than halfway finished were LTI-rate Lost time injuries (LTI) per 1 million working hours. LTIs are
335 selected making it possible to compare safety performance. injuries resulting in more sick leave than just the day of
injury as reported by the contractors to the client
336 Materials from 11 different projects were collected. One project
MTI-rate Medical treatment injuries (MTI) per 1 million working
337 was much larger than the others. This project experienced many hours. MTIs are reported from the contractors to the client
338 problems in the first part of the executions stage regarding project TRI-rate Total recordable injuries (TRI = LTI + MTI) per 1 million
339 management and safety management. It was paused for some working hours. Reported from the contractors to the client.
340 weeks and several measures were implemented. Since the two RUOs & SDs Registered unwanted occurrences (RUO) include accidents
and near accidents. Site deviations (SD) include deviations
341 parts of the execution stage were very different as regards safety from regulations registered by contractors and client on
342 management, it was decided to analyze it as two cases. The num- mostly safe job analysis (SJAs), working instructions, lack of
343 ber of cases analyzed is therefore 12. personal protective equipment, failure of scaffolding, and
344 Statsbygg is one of Norway’s largest clients, with a top danger zones not defined
WTR Willingness to report: RUO&SD per 1 000 working hours
345 management strongly committed to OHS. The client and projects
12 November 2019
388 3.5. Measuring safety management factors and contextual factors all the projects and knew all projects well. He also participated 449
in the analysis, which was a crucial advantage for data collection 450
389 Different materials were used and triangulated to assess the and analysis. 451
390 factors in Table 1: safety and health plans, inspection and audit All in all, 22 interviews were carried out. Eleven interviews with 452
391 reports, logs of OHS-related information, and interviews with client project leaders (one assistant project leader) and 11 inter- 453
392 OHS coordinators and project managers. views with OHS inspectors for each project were carried out. The 454
interviews were semi-structured, using the analytical framework 455
393 3.5.1. The client’s safety and health plan (SH plan) as an interview guide with both project leaders and OHS inspec- 456
394 Directive 92/57/EEC (European Commission, 1992) and the Nor- tors. The interviews were conducted as a dialogue, exploring how 457
395 wegian version, the Construction Client Regulations (Directorate of different factors affected the situations on site, and safety perfor- 458
396 Labour Inspection, 2009), require the client to produce a plan for mance. The dialogue enabled the quality of different aspects of pro- 459
397 safety and health plan (SH plan), which must be communicated ject performance to be compared. 460
398 to all actors in the construction project. The SH plan is the client’s
399 documentation that the work is planned and that risks are assessed
4. Results 461
400 in advance, and a tool for the client ensuring that the work is car-
401 ried out without health and safety risks. The SH plan must describe
4.1. General characteristics of the projects 462
402 (a): the organization of the project, including roles and responsibil-
403 ities for OHS; (b) a progress plan describing when and where the
We do not present detailed information about the cases, and the 463
404 various work operations are to be carried out, for example, coordi-
projects are anonymized for reasons of sensitivity regarding the 464
405 nating various work operations; (c) specific measures connected to
companies and persons involved. The buildings were museums 465
406 activities that may involve risks to life and health with specific
and university buildings, mostly new buildings and rehabilitations 466
407 measures for work involving risk; and (d) procedures for handling
of old buildings. Some projects also included groundworks and 467
408 deviations. During the execution phase the SH plan must be
demolition. The number of working hours varied from some 468
409 updated after any changes that may affect health and safety. There
17,000 to 1,150,000, with an average of 305,000. 469
410 can thus be many versions of the SH plan during a project. Between
411 two and four SH plans from different phases for each project were
412 analyzed. 4.2. Employing qualitative comparative analysis (QCA) 470
413 3.5.2. Reports from OHS audits/inspections The results of the calibration of the outcome and the conditions 471
414 OHS audits/inspections of projects are regularly conducted and (factors) are presented in Table 3. Recall that the calibration of the 472
415 led by OHS experts from the OHS department at Statsbygg. Other outcome and the condition scores are based on assessments of dif- 473
416 participants include representatives of project management in ferent types of data (triangulation) (see Section 3). The choice of 474
417 Statsbygg and representatives from the main contractor (e.g., pro- the number of values for the outcome and each condition was 475
418 ject leader, supervisor, OHS expert). The OHS inspection reports based on the characteristics of each condition. For the outcome 476
419 consist of: (a) document reviews (SH plan, progress plans, SJAs, list (safety performance) we used the six-value set with the values very 477
420 of workers, companies, RUOs, etc.); (b) interviews with central per- good (1.0), relatively good (0.8), adequately (0.6) etc. We used the 478
421 sons (e.g., managers, coordinators, safety representatives); (c) crisp set (0 or 1) for five conditions and the four-value set (0, 479
422 description of site inspection focusing on deviations; and (d) 0.33, 0.67, 1) for 11 conditions. 480
423 requirements for following up deviations. The reports also include The second column in Table 3 shows the assessment of the out- 481
424 descriptions and pictures from the sites. come safety performance (SP). Half of the projects were considered 482
to have relatively high safety performance (HSP) (0.5 <) and the 483
425 3.5.3. Log on OHS related information other half relatively low safety performance (LSP) (0.5 >). This 484
426 OHS related occurrences and activities were registered continu- was not a coincidence since we selected projects that were initially 485
427 ally during the projects, including results from safety rounds, SJAs, assessed to have relatively high or low safety performance (see 486
428 deviations, RUOs, dangerous operations, accidents, near accidents, Section 3.3). The average score for the safety management factors 487
429 OHS inspections and updates of the SH plan. More information was (factors 5–16 in Table 1) was 0.53., for all projects, 0.77 for the 488
430 registered in the ‘‘poor” projects than the ‘‘good” projects. The HSPs, and 0.30 for the LSPs. Hardware management (HA) was 489
431 descriptions gave a good overview of the projects and their assessed as good in all projects. The opinion of the client project 490
432 development. leaders and OHS inspectors was that poor material and equipment 491
is not tolerated, and when identified, measures are taken immedi- 492
433 3.5.4. Interviews ately. This means that hardware management cannot be included 493
434 After analyzing indicators, SH plans, OHS inspection reports and in QCA because it is a constant, and not a variable. 494
435 OHS logs, interviews with client project leaders (PL) were under-
436 taken. The interviews lasted between 60 and 90 minutes using 4.3. Set coincidence 495
437 videoconferencing or telephone. Interviews with OHS inspectors/-
438 experts were undertaken after all the documents and PL interviews Set coincidence is the degree to which two or more sets overlap, 496
439 were analyzed. Four OHS experts who had each been inspecting or, in other words, the extent to which they constitute one and the 497
440 several of the projects were interviewed about each of the projects. same set (Borgna, 2013). Fuzzy-set coincidence is ‘‘. . . a special case 498
441 OHS experts who had inspected a project at least twice were of correlation” (Ragin, 2008, p. 59). A set coincidence score close to 499
442 selected for interview for that particular project. Before the inter- 1 indicates that most of our cases share exactly the same degree of 500
443 views, the OHS experts read summaries of the preliminary analy- membership in two sets. If two conditions have the same values in 501
444 sis. All interviews were undertaken to supplement the all projects, they can be merged into one, if it is theoretically advis- 502
445 documented information and to triangulate (verify or contradict) able. The procedure was used to assess which conditions strongly 503
446 information from documents. The interviews were recorded and coincide (0.75 ), and whether they should be included in further 504
447 transcribed by the first author. One of the OHS experts and co- analysis. Table 4 shows that staff management, operative risk man- 505
448 author of this article (Arnesen), had participated at inspections of agement, roles and responsibilities, site management, project man- 506
12 November 2019
Table 3
Raw data for 12 construction projects.
Case SP CC OC TI EC CO PL RO PM MC SC LE PE RM SI SM HA
A 1 1 1 1 1 1 0.67 1 1 1 1 1 1 1 0.67 1 1
B 0.8 0.33 0 1 1 1 0.67 1 1 1 0.67 0.67 1 1 0.67 0.67 1
C 0.8 0.33 0.67 1 1 1 0.33 0.67 1 1 0.67 0.33 1 0.67 0.67 0.67 1
D 0.8 1 0.67 0 0 0 0.33 0.33 0.33 0.33 0.33 0.33 1 0.67 0.67 0.33 1
E 0.6 0.33 1 1 1 1 0.33 0.67 1 1 1 1 1 0.67 0.67 1 1
F 0.6 0 0 1 1 1 0.33 1 0.67 0.67 0.67 1 1 0.67 0.67 0.33 1
G 0.4 0.33 0.33 1 1 0 0.33 0.33 0.33 0.33 0.33 0.67 0 0.33 0.33 0.33 1
H 0.4 0.33 0 0 0 0 0 0.33 0.33 0.33 0 0.33 0 0.33 0.33 0.33 1
I 0.4 0 0 0 0 0 0 0.33 0.33 0 0.33 0.33 0 0.33 0.33 0.33 1
J 0.2 0.33 0.33 1 0 1 0.33 0.67 0.33 0.33 0.33 0.33 0 0.33 0.33 0.33 1
K 0.2 0.67 0.67 0 0 0 0.33 0.33 0.33 0.33 0.33 0.33 0 0.33 0.33 0.33 1
L 0 0 0.67 1 0 0 0 0 0 0 0 0 0 0 0 0 1
Note: Safety Performance (SP), Construction Complexity (CC), Organisational Complexity (OC), Time (TI), Economy (EC), Contract Management (CO), OHS-Planning (PL), Roles
and responsibilities (RO), Project Management (PM), Management Commitment (MC), Safety Climate (SC), Learning (LE), Performance Evaluation (PE), Operative Risk
Management (RM), Site (SI), Staff management (SM), Hardware management (HA).
Table 4
Set coincidence matrix for 16 conditions in 12 construction projects.
Conditions 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1. Construction complexity
2. Organisational complexity 0.58
3. Time 0.27 0.43
4. Economy 0.28 0.36 0.75
5. Contract management 0.28 0.36 0.48 0.71
6. OHS-planning 0.56 0.42 0.35 0.38 0.38
7. Roles and responsibilities 0.47 0.44 0.57 0.58 0.65 0.55
8. Project management 0.48 0.50 0.57 0.65 0.65 0.55 0.82
9. Management commitment 0.50 0.52 0.59 0.68 0.68 0.58 0.77 0.95
10. Safety climate 0.47 0.57 0.52 0.59 0.59 0.64 0.76 0.85 0.80
11. Learning 0.50 0.46 0.53 0.61 0.54 0.58 0.77 0.77 0.73 0.80
12. Performance evaluation 0.40 0.42 0.56 0.71 0.71 0.38 0.58 0.65 0.68 0.59 0.54
13. Operative risk management 0.57 0.52 0.48 0.54 0.54 0.58 0.86 0.86 0.81 0.80 0.73 0.61
14. Site management 0.55 0.50 0.42 0.46 0.46 0.64 0.76 0.76 0.72 0.79 0.71 0.53 0.90
15. Staff management 0.55 0.57 0.46 0.52 0.52 0.65 0.76 0.85 0.80 0.89 0.80 0.52 0.80 0.79
16. Hardware management 0.39 0.45 0.67 0.50 0.50 0.30 0.56 0.55 0.53 0.47 0.53 0.50 0.53 0.47 0.47
507 agement, and safety climate strongly coincide with several condi- ditions should be higher than 0.75 (Schneider & Wagemann, 2012). 514
508 tions (Table 5). If a relationship is established to be consistent, the coverage should 515
be calculated. Coverage assesses the degree to which a condition 516
accounts for instances of an outcome, or empirical relevance 517
509 4.4. Necessary conditions (Ragin, 2008). With a consistency threshold of 0.80, the results 518
indicate that eight of the OHSM conditions are ‘‘necessary” for high 519
510 Consistency and coverage are parameters used to assess how safety performance, indicating that high safety performance can- 520
511 well the cases in a data set fit a relation. Consistency resembles sig- not be achieved without high performance on that specific condi- 521
512 nificance in statistical approaches where 0 indicates no consistency tion (e.g., operative risk management) (if Y, then X). The coverage 522
513 and 1 indicates perfect consistency. The consistency value for con- (empirical relevance) is also high for these eight conditions. Hard- 523
ware was previously identified as a constant and therefore cannot 524
Table 5
be included in the analysis. 525
Necessary conditions for the outcome safety performance. N = 12. Consistency and
coverage. 4.5. Conditions and models for QCA analysis 526
Condition Consistency Coverage
1. Construction complexity 0.62 –
The aim of QCA analysis is to identify combinations of condi- 527
2. Organisational complexity 0.59 – tions that are sufficient for the outcome (if X, then Y). This process 528
3. Time 0.71 – requires four steps: (a) presenting the data in a truth table; (b) 529
4. Economy 0.68 – minimizing the truth table; (c); reporting the parameters of fit 530
5. Contract management 0.64 –
for the solution formula; and (d) interpreting the results. 531
6. OHS planning 0.55 –
7. Roles and responsibilities 0.87 0.81 The number of conditions in QCA must be kept quiet low; three 532
8. Project management 0.89 0.83 to eight conditions are recommended (Ragin, 2008). The problem is 533
9. Management commitment 0.84 0.82 that as the number of binary conditions increases, the number of 534
10. Safety climate 0.80 0.87 possible combinations of these variables increases exponentially, 535
11. Learning 0.80 0.79
12. Performance evaluation 0.74 –
so-called limited diversity (Ragin, 1987). One strategy for including 536
13. Operative risk management 0.92 0.91 more conditions is to conduct separate QCAs for different sets of 537
14. Site management 0.85 0.93 conditions (Gerrits & Verweij, 2018). We decided to conduct two 538
15. Staff management 0.81 0.89 QCAs, one including ‘‘contextual conditions” and one for ‘‘OHSM 539
16. Hardware management – –
conditions.” 540
12 November 2019
541 4.6. QCA for contextual factors combined with operative risk logical minimization (Quine-McCluskey Algorithm): ‘‘If two Boo- 590
542 management lean expressions differ in only one causal condition yet produce 591
the same outcome, then the causal condition that distinguishes 592
543 The aim of this first QCA was to detect combinations of ‘‘contex- the two expressions can be considered irrelevant and can be 593
544 tual conditions” (project complexity, organizational complexity, removed to create a simpler, combined expression” (Ragin, 1987, 594
545 and contract management) combined with OHSM. Operative risk p. 93). 595
546 management is included in the analysis to represent OHSM. Oper- Table 8 shows the results of the minimization process. The table 596
547 ative risk management was selected because it coincides (overlaps) shows two configurations sufficient for high safety performance 597
548 strongly with many of the other OHSM conditions (see Table 4). covering 80% of the outcome, and two configurations sufficient 598
for low safety performance covering 78% of the outcome. Results 599
using QCA are often presented using circles. Black circles (d) rep- 600
549 4.6.1. Truth table
resent present conditions, white circles (s) represent absent con- 601
550 A truth table lists all possible logically possible configurations.
ditions, and empty cells represent redundant conditions, that is, 602
551 Each row in the truth table represents one logically possible config-
conditions that have been minimized away through pairwise com- 603
552 uration (Ragin, 1987). Each case in the raw data matrix (Table 3) is
parison. Consistency measures the degree to which solution terms 604
553 assigned to the respective truth table row which it belongs. The
and the solution as a whole are sufficient for the outcome (suffi- 605
554 fuzzy conditions need to be dichotomized to match calibrated
ciency). Raw coverage calculates how much of the outcome is 606
555 fuzzy cases into a truth table. A fuzzy set score below 0.5 is dichot-
explained by each solution separately. Unique coverage measures 607
556 omized to 0, and a fuzzy set score above 0.5 is dichotomized to 1.
the proportion of memberships in the outcome explained solely 608
557 The truth table shows the configurations as dichotomies, but the
by each individual solution term and indicates the unique contri- 609
558 calibrated cases remain fuzzy. Since we include four conditions
butions to covering the outcome. 610
559 that can score either 0 or 1, there are 16 logically possible combi-
560 nations of the four conditions. The truth table for the outcome high
561 safety performance (Table 6) shows that there are cases in eight
4.6.3. Interpretation 611
562 out of the 16 possible configurations. Column number 2–5 indicate
The part of QCA using the software is only one step in the QCA 612
563 the qualitative status of the four conditions (present 1 vs. not pre-
process. Interpreting means returning to the cases and asking more 613
564 sent 0). Column ‘‘SP” indicates whether the given row is sufficient
focused causal questions about the mechanisms producing the 614
565 for the outcome ‘‘high safety performance” (score of 1) or not suf-
outcome (Rihoux & Lobe, 2009). There are two ‘‘paths” that pro- 615
566 ficient (0). The decision about sufficiency is based on each row’s
duce high safety performance. Configuration HP1 combines high 616
567 consistency score (‘‘Cons.”). Consistency expresses the degree to
safety performance with high project complexity, good contract 617
568 which empirical evidence supports the claim that a set-theoretic
management, and good operative risk management. Since the pro- 618
569 relation exists. Values below 0.80 indicate substantial inconsis-
jects differ in organizational complexity yet produce high safety 619
570 tency. PRI is a consistency score that is more sensitive to the pos-
performance, organizational complexity is considered redundant 620
571 sibility that one row can be a subset of the outcome as well as its
(irrelevant) for this configuration according to the logical mini- 621
572 negation. This yields four rows that include six cases that are con-
mization process. Configuration HP2 includes two projects with 622
573 sidered sufficient for high safety performance, and four rows
low construction complexity and organizational complexity, and 623
574 including six cases that are considered not sufficient for high safety
with good operative risk analysis. Contract management is consid- 624
575 performance. The ‘‘Cases” column shows the labels of cases that are
ered redundant since one project has good contract management 625
576 members of a given row.
and the other projects poor contract management. Configuration 626
577 The occurrence and non-occurrence of the outcome require sep-
LP1 has high project complexity and organizational complexity, 627
578 arate analysis because the concepts often contain various qualita-
combined with poor operative risk management. Contract man- 628
579 tively different notions, so-called causal asymmetry (Ragin,
agement is considered redundant since one project has poor con- 629
580 2008). Table 7 shows the truth table for the outcome low safety
tract management while the others have good contract 630
581 performance. It shows four rows that include six cases that are
management. Configuration LP2 has low project complexity, high 631
582 considered sufficient for low safety performance, and four rows that
organizational complexity, poor contract management, and poor 632
583 include six cases that are considered not sufficient for low safety
operative risk analysis. 633
584 performance.
The results indicate that high safety performance can be 634
achieved with both high and low construction complexity and 635
585 4.6.2. Minimizing the truth table into a solution formula organizational complexity, showing that the conditions are not 636
586 Each of the four first rows in Table 6 has been identified as suf- ‘‘necessary” for high safety performance. What seems to be impor- 637
587 ficient for high safety performance, and each of the four first rows tant is how project complexity and organizational complexity are 638
588 in Table 7 has been identified as sufficient for low safety perfor- handled by OHSM, here represented by operative risk manage- 639
589 mance. The rows of each truth table can be made simpler using ment. The results also indicate that high safety performance can 640
Table 6
Truth table for contextual conditions with high safety performance.
Row CC OC CO RM SP Cons. PRI. Cases

1 1 1 1 1 1 0.93 0.90 A
2 1 1 0 1 1 0.90 0.78 D
3 0 0 1 1 1 0.90 0.80 B, F
4 0 1 1 1 1 0.88 0.77 C, E
5 1 1 0 0 0 0.65 0.22 K
6 0 0 1 0 0 0.65 0.22 J
7 0 0 0 0 0 0.52 0 G, H, I
8 0 1 0 0 0 0.40 0 L
Note: Construction Complexity (CC), Organisational Complexity (OC), Contract Management (CO), Operative Risk Management (RM), and Safety Performance (SP).
12 November 2019
Table 7
Truth table for contextual conditions with low safety performance.
Row CC OC CO RM SP Cons. PRI. Cases

1 0 1 0 0 1 1 1 L
2 0 0 0 0 1 0.92 0.83 G, H, I
3 1 1 0 0 1 0.90 0.78 K
4 0 0 1 0 1 0.90 0.78 J
5 1 1 0 1 0 0.65 0.22 D
6 0 0 1 1 0 0.57 0.13 B, F
7 0 1 1 1 0 0.56 0.15 C, E
8 1 1 1 1 0 0.43 0.10 A
Table 8
Consistent configurations for high and low safety performance (Complex solution). (Algorithm: Quine-McCluskey).
High safety performance configurations Low safety performance configurations

(frequency cut-off; 1, consistency cut- (frequency cut-off; 1, consistency cut-
off; 0.90) off; 0.87)
Conditions HP1 HP2 LP1 LP2
Construction complexity s d s s
Organisational complexity d s d
Contract management d s
Operative risk analysis d d s s
Consistency 0.91 0.92 0.92 0.94
Raw coverage 0.44 0.49 0.63 0.32
Unique coverage 0.30 0.35 0.46 0.15
Cases B, C, E, F A, D G, H, I, J K, L
Overall solution consistency 0.93 0.91
Overall solution coverage 0.80 0.78
d, core condition (present); s, core condition (negated); empty cells, redundant conditions.
641 be achieved with both good and poor contract management, show- Table 10 shows the truth table for OHMS configurations con- 662
642 ing that contract management is not ‘‘necessary” for high safety nected to low safety performance. There are two rows that include 663
643 performance. What seems to be more important than contract six cases that are considered sufficient for low safety performance, 664
644 management is operative risk management. Operative risk man- and four rows including six cases that are considered not sufficient 665
645 agement is good in all high safety performance projects and poor for high safety performance. 666
646 in all low safety performance projects, suggesting that it is ‘‘neces-
647 sary” for high safety performance. 4.7.2. Minimizing the truth table into a solution formula 667
648 Similar explanations are also found for the time and economy The next step is to minimize the configurations from the truth 668
649 conditions. Poor economy and time makes it more challenging to tables into a solution formula. The four configurations producing 669
650 achieve high safety performance but can be handled by good high safety performance in Table 9 are reduced to three configura- 670
651 OHSM. The results are interpreted in more depth in the next sections in Table 11. The three configurations are sufficient for high 671
652 tion, jointly with the QCA of OHSM conditions. safety performance covering 90% of the outcome. The two configu- 672
rations producing low safety performance in Table 10 are reduced 673
653 4.7. QCA for safety management factors to one configuration in Table 11. The configuration is sufficient for 674
low safety performance covering 93% of the outcome. 675
654 The aim of the second QCA was to detect combinations of safety
655 management factors connected to safety performance. 4.7.3. Interpretation 676
4.7.3.1. High safety performance configuration 1 (HP1). HP1 in 677
656 4.7.1. Truth tables Table 11 includes four cases with high safety performance. Project 678
657 Table 9 shows the truth table for OHMS configurations con- C and E are also part of HP2 and interpreted with HP2. Projects A 679
658 nected to high safety performance. There are four rows that include and B performed relatively well on all OHSM conditions. One dif- 680
659 six cases that are considered sufficient for high safety performance, ference between project A and B was that A had low project com- 681
660 and two rows including six cases that are considered not sufficient plexity and organizational complexity, while B had relatively high 682
661 for high safety performance. project complexity and organizational complexity. This was ana- 683
Table 9
Truth table for OHSM conditions with high safety performance.
Row RM PL RO MC SM SP Cons. PRI Cases

1 1 1 1 1 1 1 0,93 0,84 A, B
2 1 0 1 1 1 1 0,92 0,79 C, E
3 1 0 1 1 0 1 0,89 0,64 F
4 1 0 0 0 0 1 0,89 0,64 D
5 0 0 1 0 0 0 0,77 0,18 J
6 0 0 0 0 0 0 0,44 0,05 G, H, I, K, L
Note: Operative Risk Management (RM), OHS planning (PL), Roles and Responsibilities (RO), Management Commitment (MC), Staff management (SM) and Safety Performance
(SP).
12 November 2019
Table 10
Truth table for OHSM conditions with low safety performance.
Row RM PL RO MC SM SP Cons. PRI Cases

1 0 0 1 0 0 1 0.95 0.82 J
2 0 0 0 0 0 1 0.92 0.87 G, H, I, K, L
3 1 0 0 0 0 0 0.80 0.36 D
4 1 0 1 1 0 0 0.78 0.28 F
5 1 0 1 1 1 0 0.67 0.16 C, E
6 1 1 1 1 1 0 0.61 0.16 A, B
Table 11
Consistent configurations for high and low safety performance (Complex solution). (Algorithm: Quine-McCluskey).
High safety performance configurations (frequency Low safety performance configuration

cut-off; 1, consistency cut-off; 0.89) (frequency cut-off; 1, consistency cut-off; 0.92)
Conditions HP1 HP2 HP3 LP1
Staff management d s s
Operative risk management d d d s
OHS planning s s s
Roles and responsibilities d d s
Management commitment d d s s
Consistency 0.93 0.91 0.89 0.94
Raw coverage 0.75 0.63 0.33 0.70
Unique coverage 0.16 0.04 0.11 0.06
Cases A, B, C, E C, E, F D G, H, I, J, K, L
Overall solution consistency 0.93 0.93
Overall solution coverage 0.90 0.93
684 lyzed in the first QCA (Table 8). Project A had a main contractor 4.7.3.4. Low safety performance configuration (LP1). The combina- 720
685 directly employing many workers themselves, using few subcon- tion of conditions connected to high safety performance are not 721
686 tractors and hired workers. Project B had many subcontractors necessarily the same as those that produce low safety performance, 722
687 and had to coordinate and follow all subcontractors and work so-called causal asymmetry. The same conditions can play differ- 723
688 operations closely (project management). Project B was good at ent roles in different contexts. A separate truth table analysis and 724
689 following up the contractors to ensure that each contractor was minimization process was therefore performed for poor safety 725
690 responsible for the risks they brought into the project. The good performance. 726
691 project management had an impact on operative risk management The projects with poor safety performance were more similar to 727
692 and safety performance. OHS planning was relatively good for pro- each other than the high safety performance projects, and the min- 728
693 ject A and B, compared to the other 10 projects. imization process produced only one solution for low safety perfor- 729
mance. The solution combines poor staff management, poor 730
694 4.7.3.2. High safety performance configuration 2 (HP2). The three operative risk management, poor OHS planning, and poor safety 731
695 cases in HP2 performed relatively well on most OHS conditions management commitment, while ‘‘roles” is redundant. 732
696 and achieved relatively high safety performance. One difference Project J performed relatively well on many factors (roles and 733
697 between HP2 and HP1 was that OHS planning was relatively poor responsibilities, contract management, and parts of the project 734
698 in HP2, indicating that good OHS planning might be ‘‘necessary” for management), but still had low safety performance. The main defi- 735
699 the highest safety performance. ciency was that the focus on, and commitment to, OHS as a process 736
was low and poorly integrated into production management. This 737
700 4.7.3.3. High safety performance configuration 3 (HP3). Configuration implied, for instance, that the involvement and supervision of 738
701 3 includes one project (D) that is different from the other high workers (staff management) was poor, and that planning of oper- 739
702 safety performance projects. Despite many deficiencies regarding ations to reduce risk (operative risk management) was poor. As a 740
703 project management and OHS management (e.g., planning, orga- consequence, safety behavior was poor and there was a high TRI- 741
704 nizing, roles, and staff management), the project managed to rate. The project indicates that it is not sufficient to have a rela- 742
705 respond to some of the deficiencies relatively early and implement tively good production and project management, it is also neces- 743
706 a few decisive measures. A new project leader and a coordinator for sary to emphasize the OHSM management process on its own in 744
707 the execution stage established a strong control regime, following order to achieve high safety performance. 745
708 up contractors and workers, handling hazards and potentially dan- Project L stands out as a project where almost all OHSM condi- 746
709 gerous operations on site. The measures were time-consuming and tions and safety performance were very poor and is the total oppo- 747
710 labor-intensive for the client, taking over many of the responsibil- site of Project A where all OHMS conditions were good. Project L 748
711 ities from the contractors. However, the measures worked as an had many similarities with Project D, where most OHSM condi- 749
712 improvised solution in the most safety-critical phase of the project tions were poor. Both projects experienced major problems during 750
713 to achieve high safety performance, showing the importance of the execution stage, but Project D managed to implement mea- 751
714 operative risk management. The case demonstrates that OHSM is sures to improve OHSM and achieve high safety performance while 752
715 not deterministic: It is possible to achieve high safety performance Project L did not. One reason that Project D managed to implement 753
716 despite many problems and deficiencies in OHSM. Contributory sufficient changes and Project L did not, was that Project L was 754
717 conditions were that this project (D) (like project A) combined more complex regarding both organization and the building and 755
718 low construction complexity and low organizational complexity site (see Table 8). 756
719 that reduced the difficulties of managing safety.
12 November 2019
757 4.8. Analysis of the 16 factors availability of suitable contractors was a contributory factor to 818
poor safety performance in some projects. Like Hinze and Gambat- 819
758 This section analyses and discuss the influence of each of the 16 ese (2003), we also found that subcontractor safety performance 820
759 factors. The operational definitions of the factors are described in was affected to a large extent by the actions of the general contrac- 821
760 Table 1. tor and construction management. Hale et al. (2012) also found 822
that contracting strategy (competitive tendering and contractor- 823
761 4.8.1. Construction complexity ization) was an important causal factor in fatal accidents. 824
762 Construction complexity was not found to be ‘‘necessary” for
763 high safety performance. The results indicate that high safety per- 4.8.5. OHS-planning 825
764 formance can be achieved with both high and low construction OHS-planning was not found to be ‘‘necessary” for high safety 826
765 complexity. What seemed to be important is how project complex- performance. Most projects did not have an adequate SH-plan 827
766 ity and organizational complexity were handled by operative risk including assessment of risks in advance with specific measures. 828
767 management, and probably other management factors. The result The results indicate that much residual risk was left for the front- 829
768 is consistent with Törner and Pousette (2009), who found that pro- line workers to handle. Only two projects had an adequate OHS- 830
769 ject characteristics and nature of the work are ‘‘. . . the limiting con- planning including adequate risk assessments and specific mea- 831
770 ditions to which safety management must be adjusted”, are sures. These two projects also had the highest safety performance, 832
771 difficult to change, and ‘‘establish the outer limits of safety man- indicating that good OHS planning might be ‘‘necessary” for high 833
772 agement” (ibid. p. 404). safety performance. The research also showed that assessment of 834
risks in advance with specific measures is very demanding because 835
773 4.8.2. Organizational complexity of the dynamic nature and new risks being produced consecu- 836
774 Organizational complexity was treated as a ‘‘contextual” in this tively. Conventional OHS risk management methods, assuming 837
775 research, even though it is also the result of management deci- that that work can be decomposed into its parts, is of limited value 838
776 sions. The results indicate that low organizational complexity is in construction because system elements are in constant dynamic 839
777 not ‘‘necessary” for high safety performance. The results do not interaction with one another (Cooke-Davis et al., 2007). Hallowell 840
778 contradict that there is an association between increased on-site and Gambatese (2009) found that a written and comprehensive 841
779 subcontracting and increased risks of injuries (Azari-Rad, Philips, safety and health plan was an essential safety program element, 842
780 & Thompson-Dawson, 2003). The results do, however, indicate that and Hallowell and Calhoun (2011) found that a site-specific safety 843
781 high organizational complexity complicates the coordination of plan is one of the most central elements in an effective safety 844
782 actors and operations and achieving high safety performance. program. 845
783 What seems to be important is how the organizational complexity
784 is handled by safety management, particularly operative risk 4.8.6. Roles and responsibilities 846
785 management. Roles and responsibilities was found to be ‘‘necessary” for high 847
safety performance. The results indicated that two types of roles 848
786 4.8.3. Time and economy were important for high safety performance. First, that OHS was 849
787 Time and economy are tightly connected and are therefore to a large degree a management responsibility with active project 850
788 described jointly. The hypothesis was that ‘‘poor” time and econ- leaders. Second, that at least one of the roles with specific respon- 851
789 omy can bring about reduced effort on safety, which can lead to sibilities for OHS (coordinator for the execution stage, OHS-leader, 852
790 poor safety performance. Adequate time and economy were not OHS-coordinator etc.) was very active in the OHS activities and 853
791 found to be ‘‘necessary” for safety performance. In most projects, coordination. These results are consistent with results from litera- 854
792 there was a connection between time/economy and safety perfor- ture reviews about the importance of top management generally 855
793 mance, but there was also one project with poor ‘‘time/economy” (e.g., Hale, 1998; Shannon, Mayr, & Haines, 1997) and in construc- 856
794 with high safety performance, and one project with good ‘‘time/ tion (Hallowell, Hinze, Baud, & Wehle, 2013; Tam, Zeng, & Deng, 857
795 economy” and poor safety performance. The results indicate, like 2004). The results are also similar to Hale et al. (2010) who found 858
796 organizational complexity, that poor contextual factors can be han- that the amount of energy and creativity injected by top managers 859
797 dled by good safety management. Previous studies have found con- and coordinator (safety professional) appeared to be a distinguish- 860
798 nections between time/economy and safety performance (Han, ing factor. 861
799 Saba, Lee, Mohamed, & Peña-Mora, 2014; Holmes, Lingard,
800 Yesilyurt, & De Munk, 1999; Mullen, 2004). Mullen (2004) found 4.8.7. Project management 862
801 that when resources (i.e., time and money) were inadequate, there Project management was found to be ‘‘necessary” for high 863
802 was pressure from both managers and co-workers to prioritize per- safety performance. Projects with adequate project management 864
803 formance over safety, and that such pressure swiftly socialized managed to follow up OHS, coordinate the activities, and ensure 865
804 individuals to adapt and consider unsafe practices as normal. adequate communication between the actors. Project management 866
is defined differently in different studies and it is hence problem- 867
805 4.8.4. Contract management atic to compare to many studies. Our definition is similar to the 868
806 Except for two projects, the contracts were different variants of ConAC model (Haslam et al., 2005; Winge, Albrechtsen, & 869
807 design and build (turn-key) contracts where a main contractor is Mostue, 2019). In different accident studies using the ConAC 870
808 given a performance specification by the client and must under- framework, deficiencies in project management was found to one 871
809 take the project from design to construction, and to a completed of the most frequent ‘‘originating” factors (Gibb, Lingard, Behm, 872
810 building. What seemed to be most important in achieving a high & Cooke, 2014; Winge et al., 2019). 873
811 safety performance was how well the client and contractors were
812 able to cooperate, communicate, and avoid conflicts, not the formal 4.8.8. Management commitment to OHS 874
813 contract management and contract. The result is broadly consis- Management commitment to OHS was found to be necessary 875
814 tent with Bolt, Haslam, Gibb, and Waterson (2012) who found that for high safety performance. In projects with high safety perfor- 876
815 a key factor was that systems (contracts, processes, systems and mance and adequate management commitment to OHS, the man- 877
816 equipment, etc.) and people work in tandem. The choice of main agers expressed clearly that safety was prioritized before 878
817 contractor was also important for safety performance. Limited production, and actively participated in OHS and other OHS- 879
12 November 2019
880 related management factors like project management, safety cli- 4.8.12. Operative risk management 942
881 mate, planning, and staff management. The results are consistent Operative risk management was found to be ‘‘necessary” for 943
882 with literature reviews (Mohammadi et al., 2018; Shannon et al., high safety performance. Operative risk management was good 944
883 1997). Hallowell et al. (2013) concluded that safety performance in all high safety performance projects and poor in all low safety 945
884 is exceptionally strong when top management is visibly involved performance projects. One project illustrates the importance of 946
885 in safety. operative risk management. The project performed poorly on most 947
safety management factors but had good operative risk manage- 948
886 4.8.9. Safety climate ment and high safety performance (0.8). The results also indicate 949
887 The aim of including safety climate as a factor, was to assess the that projects with high project complexity and organizational com- 950
888 ‘‘informal aspects” of safety management. Antonsen (2009, p. 17) plexity can achieve high safety performance, if the operative risk 951
889 describes a ‘‘good” safety climate as ‘‘. . . one where managers at management is good. Operative risk management does not, how- 952
890 all levels are highly committed to safety; where the workforce ever, operate in isolation. An analysis indicated that project man- 953
891 express satisfaction with and adherence to the organization’s agement and roles and responsibilities were consistent 954
892 safety system; where everyone is risk adverse; where there is no ‘‘necessary” factors for operative risk management. 955
893 pressure towards maximizing profits at the expense of safety; One central explanation why operative risk management is 956
894 and where operators as well as managers are highly qualified important, is probably that many risks were not handled in early 957
895 and competent.” Safety climate was found to be ‘‘necessary” for stages, and that many residual risks had to be handled consecu- 958
896 safety performance. Table 4 shows that safety climate coincides tively. The factor OHS-planning (Section 4.8.5) showed that most 959
897 with several other factors, and the analysis suggests that safety cli- projects did not have an adequate OHS-planning, including assess- 960
898 mate both influence, and is influenced, by several other factors. ment of risks in advance with specific measures, and that many 961
899 Safety climate coincide with staff management, which indicates risks therefore were left for the frontline workers to handle. These 962
900 that safety climate is connected to the selection of personnel, residual risks hence had to be handled by operative risk manage- 963
901 and the safety climate they bring with them. What characterizes ment. Poor risk management was also found to be a dominant 964
902 the projects with a high score on safety climate was that the pro- organizational factor in accident analyses in construction (Behm 965
903 ject management and OHS coordinators followed up the frontline & Schneller, 2013, Haslam, 2005, Winge & Albrechtsen, 2019). 966
904 workers closely regarding safety behaviors, and that they had sev-
905 eral social arrangements focusing on OHS where both managers 4.8.13. Site management 967
906 and frontline workers participated. The results are consistent with Site management was found to be ‘‘necessary” for high safety 968
907 Törner and Pousette (2009) who found that ‘‘interaction and coop- performance. Projects that had adequate site management were 969
908 eration and conditions supporting cooperation through empower- well organized, had clearly defined danger zones, pathways, areas 970
909 ment, mutual trust, and having a keen ear were important in for storage, good housekeeping, and few hazards. The importance 971
910 relation to safety” (p. 405). Based on a literature review, of site condition is evident in falls from height. In a literature 972
911 Mohammadi et al. (2018) concluded that an adequate safety cli- review of 75 studies about falls from height, Nadhim, Hon, Xia, 973
912 mate is a key aspect to prevent accidents and illnesses. Stewart, and Fang (2016) found that site condition was one of 974
the most common factors. Falls from height could occur when 975
there were, for example, unprotected walkways, improper guard- 976
913 4.8.10. Learning rails, slippery or sloped surfaces. 977
914 Learning was found to be ‘‘necessary” for high safety perfor-
915 mance. Projects with high performance on learning had regular 4.8.14. Staff management 978
916 inspections and safety walks, and safety representatives that were Staff management was found ‘‘necessary” for high safety perfor- 979
917 active and participated on safety walks. These projects had a high mance. In projects with adequate staff management, the share of 980
918 willingness to report unwanted occurrences (see Table 2) and risks skilled (trained) workers was high, the companies and workers 981
919 were mostly handled consecutively. Mohammadi et al. (2018) had often worked together in previous projects, the safety climate 982
920 identified ‘‘lesson learned from accidents” as one of the factors was good, and supervision and safety behavior were good. Staff 983
921 influencing safety performance in construction. Hallowell and management coincide strongly with safety climate and project 984
922 Gambatese (2009) found in their literature review that ‘‘record- management (Table 4). The results are consistent with Choudhry 985
923 keeping and accident analyses” was a central program element, and Fang (2008), who concluded that management behavior plays 986
924 but one of the least effective. an important role in improving workers’ behavioral safety perfor- 987
mance. They also concluded that management can help workers 988
925 4.8.11. Performance evaluation to improve safety behaviors through the influence of rules and reg- 989
926 Performance evaluation was not found to be ‘‘necessary” for ulations, training, and increased communication. Several studies 990
927 high safety performance. All projects had some problems regarding also show the importance of the supervisors in enhancing good 991
928 OHS management early in the execution stage, all tried to solve safety behavior (Fang, Wu, & Wu, 2015; Kines et al., 2010; 992
929 them, but not all succeeded. What seems to be important was Mohamed, 2002; Rowlinson, Mohamed, & Lam, 2003; Winge & 993
930 how early problems were identified, which types of measures were Albrechtsen, 2019). 994
931 implemented, and how extensive the measures were. It is also
932 clear that construction complexity and organizational complexity 5. Discussion and conclusion 995
933 described above influenced the opportunities for implementing
934 the measures successfully. The result is similar to a literature The aim of this research was to identify how safety manage- 996
935 review by Loushine et al. (2006) who concluded that continuous ment factors, contextual factors, and combinations of factors influ- 997
936 improvement requires continuous monitoring of work or collection ence safety performance. 998
937 of data, analysis, and changes in the work processes to ensure that
938 work is progressing towards goals. Performance evaluation (con- 5.1. Combinations of factors 999
939 tinual performance evaluation and monitoring of the OH&S man-
940 agement system) is also a key factor in ISO 45001 (ISO, 2019) for The results showed that the average score for the 12 safety 1000
941 continuous improvement. management factors was far better among the high safety perfor- 1001
12 November 2019
1002 mance projects compared to the low safety management projects. agement, (7) staff management, and (8) operative risk 1066
1003 The result is broadly consistent with literature reviews that that management. Site management, operative risk management, and 1067
1004 safety management systems can deliver more healthy and safe staff management were the three factors most strongly connected 1068
1005 workplaces (Gallagher, Rimmer, & Underhill, 2001) and lower acci- to safety performance. This is probably because the factors are the 1069
1006 dent rates (Thomas, 2011). The results are also broadly consistent most proximal, with most direct influence on what is going on at 1070
1007 with Loushine et al. (2006) who in a literature review on safety the sharp end, and essential in the daily control of the safety at site. 1071
1008 management and quality management found that construction
1009 projects with integrated safety and quality management systems/
5.3. Limitations, contributions and future research 1072
1010 programs have better safety performance.
1011 The results showed that high safety performance can be
One contribution of this research is the comparative approach 1073
1012 achieved with both high and low construction complexity and
studying ‘‘good” and ‘‘bad” construction projects employing QCA 1074
1013 organizational complexity. The results indicated, however, that
and the QCA software, which gives an opportunity to study what 1075
1014 high construction complexity and organizational complexity com-
goes right and what goes wrong (Hollnagel, 2014). Our experience 1076
1015 plicate safety management. What seemed to be important was
was that the approach and software helped to identify the patterns 1077
1016 how project complexity and organizational complexity was han-
of causal complexity producing high and low safety performance. 1078
1017 dled by operative risk management. Regarding construction com-
Employing QCA helped us identify complexity in combinations of 1079
1018 plexity, the results are broadly consistent with Törner and
factors (configurations) and different paths producing high and 1080
1019 Pousette (2009) who concluded that the inherent complexity of
low safety performance (equifinality). There are, however, some 1081
1020 construction work restricts and complicates safety management
methodological and empirical limitations in this research that we 1082
1021 and demands comprehensive safety management. Regarding orga-
recommend is followed up by future research. 1083
1022 nizational complexity, the results do not contradict that there is a
First, the results are not based on hard facts but on the research- 1084
1023 statistical association between increased on-site subcontracting
ers’ assessment of each causal factor and safety performance. 1085
1024 and increased risks of injuries (Azari-Rad et al., 2003). The results
Regarding measurement of safety performance, there is much evi- 1086
1025 do, however, indicate that high organizational complexity compli-
dence of under-reporting of workplace injuries (Shannon et al., 1087
1026 cates the coordination of actors and operations and the production
2001) and safety indicators can be subject to manipulation and 1088
1027 of a high safety performance. In projects with high organizational
misinterpretation (Oswald, Zhang, Lingard, Pirzadeh, & Le, 2018). 1089
1028 complexity and high safety performance, the organizational com-
Several data sources and methods were therefore combined by tri- 1090
1029 plexity was adequately planned for, and extensively managed by,
angulation (Denzin, 1970) and mixed methods (Tashakkori & 1091
1030 for example, involvement, cooperation and follow-up of the con-
Teddlie, 2010) to increase internal validity. Second, the assessment 1092
1031 tractors. The results are consistent with Hallowell and
of safety performance and safety management performance was 1093
1032 Gambatese (2009) who found that strategic subcontractor selec-
problematic in some of the construction projects since perfor- 1094
1033 tion and management is among the most effective elements in
mance often varies during a project. Third the study focuses on 1095
1034 safety programs.
safety management in construction projects primarily from a cli- 1096
1035 The results also showed that it is possible to achieve high safety
ent’s perspective, and the cases could have been studied more in 1097
1036 performance despite many relatively poor safety management fac-
depth from other actors’ perspective, for example from contractor’s 1098
1037 tors, and that it is possible to produce low safety performance
perspective or frontline workers’ perspective. It would have been 1099
1038 despite many relatively good safety management factors. The
preferable to study more documents and interview more managers 1100
1039 results indicate that it is not sufficient to have a relatively good
and safety representatives at the sharp end. There is, however, 1101
1040 production and project management, it is also necessary to empha-
always a trade-off between depth and width. Forth, it was chal- 1102
1041 size the safety management process on its own to achieve high
lenging to compare the results to previous research due to different 1103
1042 safety performance. The result showing that single necessary fac-
definitions and poorly described definitions of factors in many 1104
1043 tors can be jointly sufficient to produce high and low safety perfor-
studies. It is important for the accumulation of knowledge that 1105
1044 mance is broadly consistent with the understanding of causality in
key factors are clearly described. Fifth, the number of factors using 1106
1045 many accident models (Hopkins, 2014; Reason, Hollnagel, Paries
QCA must be kept relatively low. The problem is that as the num- 1107
1046 (2006); Winge et al., 2019). This research also showed that there
ber of binary factors increases, the number of possible combina- 1108
1047 can be different combinations of factors producing high and low
tions of these variables increases exponentially – so-called 1109
1048 safety performance, so-called equifinality. The results support that
limited diversity (Ragin, 1987). Therefore, two different QCAs 1110
1049 the combination of many factors play a key role in safety manage-
involving different combinations of factors were employed. Sixth, 1111
1050 ment (e.g., Dyreborg, Nielsen, Kines, Dziekanska, Frydendall,
aiming for both in-depth insight into cases and complexity, and 1112
1051 Bengtsen, & Rasmussen, 2013; Hale et al., 2005; Hallowell &
to produce generalizations, might seem to be contradictory goals 1113
1052 Calhoun, 2011; Shannon, Robson, & Sale, 2001). Similarly, ISO
(see Ragin, 1987). Twelve cases are relatively few to produce gen- 1114
1053 45001 (ISO, 2019) states that the effectiveness and ability to
eric results, and the results must therefore be treated with caution. 1115
1054 achieve outcomes of an OHS system are dependent on several
The research is therefore seen as a building-block type of research 1116
1055 key elements.
(George & Bennett, 2005), and we encourage researchers to under- 1117
take similar studies with larger N, which makes it easier to include 1118
1056 5.2. Single factors
more factors and study how they operate together. We also 1119
encourage more case studies and intermediate-N studies to 1120
1057 Safety performance is the result of a complex interplay between
increase our knowledge of associations between safety manage- 1121
1058 different factors as demonstrated by the QCA analysis. At the same
ment factors, combinations of factors, and safety performance in 1122
1059 time, single factors have special characteristics and can have a
construction projects. 1123
1060 specific causal influence. All factors are analyzed and discussed
1061 in the results section. The analysis showed how each factor influ-
1062 enced safety performance. Eight safety management factors were Uncited references 1124
1063 found to be ‘‘necessary” for high safety performance: (1) roles
1064 and responsibilities, (2) project management, (3) OHS manage- Abad et al. (2013), Conchie et al. (2011), Cooke and Lingard 1125
1065 ment and integration, (4) safety climate, (5) learning, (6) site man- (2011), Deming and Edwards (1982), European Union (1989), 1126
12 November 2019
1127 Hallowell (2010), Hardison et al. (2014), Haslam et al. (2003), ISO eksisterende vitenskapelig litteratur om virkningen av ulike typer 1197
sikkerhetsforanstaltninger for å forebygge ulykker på jobben. file:///C:/Users/ 1198
1128 (2015), Khosravi et al. (2014), López-Alonso et al. (2013), Pinto 1199
winge_s/Downloads/Dyreborg_Review_af_ulykkesforebyggelsen_SIPAW_2013.
1129 et al. (2011), Reason (2016), Ringen et al. (1995), Swuste et al. pdf (extracted January 3, 2019) (2013). 1200
1130 (2012), Zhou et al. (2015). European Commission (1992). Council Directive 92/57/EEC of 24 June 1992 on the 1201
implementation of minimum safety and health requirements at temporary or 1202
mobile construction sites. Brussels: European Commission. 1203
1131 Declaration of Competing Interest European Commission., Non-binding guide to good practice for understanding and 1204
implementing Directive 92/57/EEC on the implementation of minimum safety 1205
and health requirements at temporary or mobile construction sites. http:// 1206
1132 The authors declare that they have no known competing finan- www.mrms.hr/wp-content/uploads/2013/03/non-binding-guide-for- 1207
1133 cial interests or personal relationships that could have appeared construction-sites.pdf (extracted January 10, 2019) (2011). 1208
European Union., Council Directive of 12 June 1989 on the introduction of measures 1209
1134 to influence the work reported in this paper.
to encourage improvements in the safety and health of workers at work (89/ 1210
391/EEC). Official Journal of the European Communities, No. L 183, 29 June 1989 1211
1135 Acknowledgements (1989). 1212
Eurostat., Accidents at work statistics. https://ec.europa.eu/eurostat/statistics- 1213
explained/index.php/Accidents_at_work_statistics#Incidence_rates (extracted 1214
1136 This study is a part of a research project about construction 24 January 2019) (2018). 1215
1137 safety funded by the Research Council of Norway and the Norwe- Fang, D., Wu, C., & Wu, H. (2015). Impact of the supervisor on worker safety 1216
behavior in construction projects. Journal of Management in Engineering, 31(6), 1217
1138 gian Labour Inspection Authority. The funding sources had no 1218
04015001.
1139 involvement in study design, in the collection, analysis and inter- Gallagher, C.; Rimmer, M.; Underhill, E., Occupational Health and Safety 1219
1140 pretation of data, in the writing of the report, and in the decision Management Systems [electronic Resource]: A Review of Their Effectiveness 1220
in Securing Healthy and Safe Workplaces. National Occupational Health and 1221
1141 to submit the article for publication. Study’s content, conclusions, 1222
Safety Commission (2001).
1142 and the opinions expressed are those of the authors alone. The George, A. L., & Bennett, A. (2005). Case studies and theory development in the social 1223
1143 authors are grateful to Statsbygg and Tanja Dugstad for letting us sciences. MIT Press. 1224
Gerrits, L., & Verweij, S. (2018). The evaluation of complex infrastructure projects: A 1225
1144 study their construction projects, showing an extraordinary will-
guide to qualitative comparative analysis. Cheltenham: Edward Elgar Publishing 1226
1145 ingness to be open and to learn from successes and non- Limited. 1227
1146 successes. The authors are also grateful to the interviewees Gibb, A., Lingard, H., Behm, M., & Cooke, T. (2014). Construction accident causality: 1228
Learning from different countries and differing consequences. Construction 1229
1147 (OHS-experts and project leaders at Statsbygg) for taking time to
Management and Economics, 32(5), 446–459. 1230
1148 contribute their professional insight into the causes of safety per- Hale, A. R. (2003a). Safety management in production. Human Factors and 1231
1149 formance in the projects studied. The authors are also grateful to Ergonomics in Manufacturing & Service Industries, 13(3), 185–201. 1232
1150 Urban Kjellén and Trond Kongsvik for discussions and comments. Hale, A. R. (2003b). Management of industrial safety. Netherlands: Delft University of 1233
Technology. 1234
Hale, A. (2005). Safety management, what do we know, what do we believe we 1235
1151 References know, and what do we overlook. Tijdschrift voor toegepaste Arbowetenschap, 18 1236
(3), 58–66. 1237
Hale, A. R., & Hovden, J. (1998). Management and culture: The third age of safety. A 1238
1152 ILO (International Labour Organization) (2001). Guidelines on occupational safety and
review of approaches to organizational aspects of safety, health and 1239
1153 health management systems. ILO-OSH 2001. Geneva: International Labour Office.
environment. Occupational Injury: Risk, Prevention and Intervention, 129–165. 1240
1154 Abad, J., Lafuente, E., & Vilajosana, J. (2013). An assessment of the OHSAS 18001
Hale, A. R., Guldenmund, F. W., Van Loenhout, P. L. C. H., & Oh, J. I. H. (2010). 1241
1155 certification process: Objective drivers and consequences on safety
Evaluating safety management and culture interventions to improve safety: 1242
1156 performance and labour productivity. Safety Science, 60, 47–56.
Effective intervention strategies. Safety Science, 48(8), 1026–1035. 1243
1157 Antonsen, S. (2009). Safety culture: Theory, method and improvement. CRC Press.
Hale, A. R., Walker, D., Walters, N., & Bolt, H. (2012). Developing the understanding 1244
1158 Azari-Rad, H.; Philips, P.; Thompson-Dawson, W., Subcontracting and injury rates in
of underlying causes of construction fatal accidents. Safety Science, 50(10), 1245
1159 construction. In Industrial Relations Research Association Series (January).
2020–2027. 1246
1160 www.http://lerachapters.org/OJS/ojs-2.4.4-1/index.php/PFL/article/view/555
Hallowell, M. (2010). Cost-effectiveness of construction safety programme 1247
1161 (extracted January 9. 2019) (2003).
elements. Construction Management and Economics, 28(1), 25–34. 1248
1162 Behm, M., & Schneller, A. (2013). Application of the Loughborough Construction
Hallowell, M. R., & Calhoun, M. E. (2011). Interrelationships among highly effective 1249
1163 Accident Causation model: A framework for organizational learning.
construction injury prevention strategies. Journal of Construction Engineering 1250
1164 Construction Management and Economics, 31(6), 580–595.
and Management, 137(11), 985–993. 1251
1165 Berger, J., The Health and Safety Protection Plan and the file containing features of
Hallowell, M. R., & Gambatese, J. A. (2009). Construction safety risk mitigation. 1252
1166 the building according to EEC Directive (92/57). The management of
Journal of Construction Engineering and Management, 135(12), 1316–1323. 1253
1167 construction safety and health, Rotterdam (2000).
Hallowell, M. R., Hinze, J. W., Baud, K. C., & Wehle, A. (2013). Proactive construction 1254
1168 Berg-Schlosser, D., De Meur, G., Ragin, C., & Rihoux, B. (2009). Configurational
safety control: Measuring, monitoring, and responding to safety leading 1255
1169 comparative methods. In B. Rihoux & C. Ragin (Eds.), Qualitative comparative
indicators. Journal of Construction Engineering and Management, 139(10), 1256
1170 analysis (QCA) and related techniques (pp. 1–18). Thousand Oaks California: Sage.
04013010. 1257
1171 Bolt, H.; Haslam, R.; Gibb, A. G.; Waterson, P., Pre-conditioning for success:
Han, S., Saba, F., Lee, S., Mohamed, Y., & Peña-Mora, F. (2014). Toward an 1258
1172 characteristics and factors ensuring a safe build for the Olympic Park. London:
understanding of the impact of production pressure on safety performance in 1259
1173 Health and Safety Executive. Research Report RR955. https://dspace.lboro.ac.
construction operations. Accident Analysis & Prevention, 68, 106–116. 1260
1174 uk/dspace-jspui/bitstream/2134/12247/2/RR955%202012.pdf (extracted
Hardison, D., Behm, M., Hallowell, M. R., & Fonooni, H. (2014). Identifying 1261
1175 January 8, 2019) (2012).
construction supervisor competencies for effective site safety. Safety Science, 1262
1176 Borgna, C., Fuzzy-set coincidence analysis: The hidden asymmetries. COMPASSS
65, 45–53. 1263
1177 Working Paper 2013-72. http://www.compasss.org/wpseries/Borgna2013.pdf
Haslam, R. A., Hide, S. A., Gibb, A. G. F., Gyi, D. E., Atkinson, S., Pavitt, T. C., ... Suraji, A. 1264
1178 (extracted January 10, 2019) (2013).
(2003). Causal factors in construction accidents, HSE Report RR156. Norwich: 1265
1179 Choudhry, R., Fang, D., & Mohamed, S. (2007). Developing a model of construction
HMSO. 1266
1180 safety culture. Journal of Management in Engineering, 23(4), 207–212.
Haslam, R., Hide, S., Gibb, A., Gyi, D., Pavitt, T., Atkinson, S., & Duff, A. (2005). 1267
1181 Conchie, S. M., Taylor, P. J., & Charlton, A. (2011). Trust and distrust in safety
Contributing factors in construction accidents. Applied Ergonomics, 36(3), 1268
1182 leadership: Mirror reflections? Safety Science, 49(8–9), 1208–1214.
401–451. 1269
1183 Cooke, T., & Lingard, H. (2011). A retrospective analysis of work-related deaths in
Hollnagel, E. (2014). Safety-I and safety-II: The past and future of safety management. 1270
1184 the Australian construction industry. In C. Egbu & E. C. W. Lou (Eds.), Proceedings
CRC Press. 1271
1185 27th Annual ARCOM Conference, Bristol, 5–7 September, Association of Researchers
Holmes, N., Lingard, H., Yesilyurt, Z., & De Munk, F. (1999). An exploratory study of 1272
1186 in Construction Management (pp. 279–288).
meanings of risk control for long term and acute effect occupational health and 1273
1187 Deming, W. E., & Edwards, D. W. (1982). Quality, productivity, and competitive
safety risks in small business construction firms. Journal of Safety Research, 30 1274
1188 position. Cambridge, MA: Massachusetts Institute of Technology, Center for
(4), 251–261. 1275
1189 Advanced Engineering Study.
Hopkins, A. (2014). Issues in safety science. Safety Science, 67, 6–14. 1276
1190 Denzin, N. K. (1970). The research act in sociology: A theoretical introduction to
Huang, X., & Hinze, J. (2006). Owner’s role in construction safety. Journal of 1277
1191 sociological methods. London: Butterworths.
Construction Engineering and Management, 132(2), 164–173. 1278
1192 Directorate of Labour Inspection., Construction client regulations. Regulations
ISO, International Standards Organization., ISO 9001: Quality Management 1279
1193 concerning safety, health and working environment at construction sites.
Systems-Requirements. https://www.iso.org/standard/62085.html (extracted 1280
1194 Unofficial English translation, issued 2015. Order No. 599-ENG (2009).
January 8, 2019) (2015). 1281
1195 Dyreborg, J.; Nielsen, K.; Kines, P.; Dziekanska, A.; Frydendall, K.; Bengtsen, E.;
1196 Rasmussen, K., Gjennomgang av ulykkesforebygging: gjennomgang av
12 November 2019
1282 ISO, International Standards Organization., ISO 45001: Occupational health and Ringen, K., Englund, A., Welch, L., Weeks, J. L., & Seegal, J. L. (1995). Why 1347
1283 safety management systems - Requirements with guidance for use. https:// construction is different. Occupational Medicine (Philadelphia, Pa.), 10(2), 1348
1284 www.iso.org/standard/62085.html (extracted January 8, 2019) (2018). 255–259. 1349
1285 Johnson, W. G. (1980). MORT safety assurance systems. Marcel Dekker Inc.. Robson, L. S., Clarke, J. A., Cullen, K., Bielecky, A., Severin, C., Bigelow, P. L., & 1350
1286 Khosravi, Y., Asilian-Mahabadi, H., Hajizadeh, E., Hassanzadeh-Rangi, N., Bastani, H., Mahood, Q. (2007). The effectiveness of occupational health and safety 1351
1287 & Behzadan, A. H. (2014). Factors influencing unsafe behaviors and accidents on management system interventions: A systematic review. Safety Science, 45(3), 1352
1288 construction sites: A review. International Journal of Occupational Safety and 329–353. 1353
1289 Ergonomics, 20(1), 111–125. Roig-Tierno, N., Gonzalez-Cruz, T. F., & Llopis-Martinez, J. (2017). An overview of 1354
1290 Kines, P., Andersen, L. P., Spangenberg, S., Mikkelsen, K. L., Dyreborg, J., & Zohar, D. qualitative comparative analysis: A bibliometric analysis. Journal of Innovation 1355
1291 (2010). Improving construction site safety through leader-based verbal safety & Knowledge, 2(1), 15–23. 1356
1292 communication. Journal of Safety Research, 41(5), 399–406. Rowlinson, S., Mohamed, S., & Lam, S. W. (2003). Hong Kong construction foremen’s 1357
1293 Kjellén, U., & Albrechtsen, E. (2017). Prevention of accidents and unwanted safety responsibilities: A case study of management oversight. Engineering, 1358
1294 occurrences: Theory, methods, and tools in safety management. Boca Raton: CRC Construction and Architectural Management, 10(1), 27–35. 1359
1295 Press. Schneider, C. Q., & Wagemann, C. (2012). Set-theoretic methods for the social sciences: 1360
1296 Kjellén, U., Tinmannsvik, R. K., Ulleberg, T., Olsen, P. E., & Saxvik, B. (1987). SMORT A guide to qualitative comparative analysis. Cambridge University Press. 1361
1297 Sikkerhetsanalyse av industriell organisasjon. Oslo: Yrkeslitteratur. Shannon, H. S., Mayr, J., & Haines, T. (1997). Overview of the relationship between 1362
1298 Li, Y., & Guldenmund, F. W. (2018). Safety management systems: A broad overview organizational and workplace factors and injury rates. Safety Science, 26(3), 1363
1299 of the literature. Safety Science, 103, 94–123. 201–217. 1364
1300 Lingard, H. (2013). Occupational health and safety in the construction industry. Shannon, H. S., Robson, L. S., & Sale, J. E. (2001). Creating safer and healthier 1365
1301 Construction Management and Economics, 31(6), 505–514. workplaces: Role of organizational factors and job characteristics. American 1366
1302 Lingard, H., Oswald, D., & Le, T. (2018). Embedding occupational health and safety in Journal of Industrial Medicine, 40(3), 319–334. 1367
1303 the procurement and management of infrastructure projects: Institutional Spangenberg, S. (2010). Large construction projects and injury prevention. Doctoral 1368
1304 logics at play in the context of new public management. Construction dissertation (Dr.Techn). Denmark: National Research Centre for the Working 1369
1305 Management and Economics, 1–17. Environment, Denmark & University of Aalborg. 1370
1306 Lingard, H., & Rowlinson, S. (2005). Occupational health and safety in construction Swuste, P., Frijters, A., & Guldenmund, F. (2012). Is it possible to influence safety in 1371
1307 project management. London and New York: Taylor & Francis. the building sector? A literature review extending from 1980 until the present. 1372
1308 López-Alonso, M., Ibarrondo-Dávila, M. P., Rubio-Gámez, M. C., & Munoz, T. G. Safety Science, 50(2), 1333–1343. 1373
1309 (2013). The impact of health and safety investment on construction company Tam, C. M., Zeng, S. X., & Deng, Z. M. (2004). Identifying elements of poor 1374
1310 costs. Safety Science, 60, 151–159. construction safety management in China. Safety science, 42(7), 569–586. 1375
1311 Loushine, T. W., Hoonakker, P. L., Carayon, P., & Smith, M. J. (2006). Quality and Tashakkori, A., & Teddlie, C. (Eds.). (2010). Sage handbook of mixed methods in social 1376
1312 safety management in construction. Total Quality Management and Business & behavioral research. Sage. 1377
1313 Excellence, 17(9), 1171–1212. Terwel, K. C., & Jansen, S. J. (2014). Critical factors for structural safety in the design 1378
1314 Mohamed, S. (2002). Safety climate in construction site environments. Journal of and construction phase. Journal of Performance of Constructed Facilities, 29(3), 1379
1315 Construction Engineering and Management, 128(5), 375–384. 04014068. 1380
1316 Mohammadi, A., Tavakolan, M., & Khosravi, Y. (2018). Factors influencing safety Tinmannsvik, R. K., & Hovden, J. (2003). Safety diagnosis criteria—Development and 1381
1317 performance on construction projects: A review. Safety Science, 109, 382–397. testing. Safety Science, 41(7), 575–590. 1382
1318 Mullen, J. (2004). Investigating factors that influence individual safety behavior at Törner, M., & Pousette, A. (2009). Safety in construction–A comprehensive 1383
1319 work. Journal of Safety Research, 35(3), 275–285. description of the characteristics of high safety standards in construction 1384
1320 Nadhim, E., Hon, C., Xia, B., Stewart, I., & Fang, D. (2016). Falls from height in the work, from the combined perspective of supervisors and experienced workers. 1385
1321 construction industry: A critical review of the scientific literature. International Journal of Safety Research, 40(6), 399–409. 1386
1322 journal of Environmental Research and Public Health, 13(7), 638. Winge, S., Albrechtsen, E., & Mostue, B. A. (2019). Causal factors and connections in 1387
1323 Oswald, D., Zhang, R. P., Lingard, H., Pirzadeh, P., & Le, T. (2018). The use and abuse construction accidents. Safety Science, 112, 130–141. 1388
1324 of safety indicators in construction. Engineering, Construction and Architectural Zhou, Z., Goh, Y. M., & Li, Q. (2015). Overview and analysis of safety management 1389
1325 Management, 25(9), 1188–1209. studies in the construction industry. Safety Science, 72, 337–350. 1390
1326 Pinto, A., Nunes, I. L., & Ribeiro, R. A. (2011). Occupational risk assessment in Zwetsloot, G. I. J. M., What are occupational safety and health management systems 1391
1327 construction industry - Overview and reflection. Safety Science, 49(5), 616–624. and why do companies implement them? (May 7, 2019): https://oshwiki.eu/ 1392
1328 Ragin, C. (1987). The comparative method: Moving beyond qualitative and quantitative wiki/What_are_occupational_safety_and_health_management_systems_and_ 1393
1329 methods. Berkeley: University of California. why_do_companies_implement_them%3F (2013). 1394
1330 Ragin, C. C. (2008). Redesigning social inquiry: Fuzzy sets and beyond. Illinois: 1395
1331 University of Chicago Press. 1396
Stig Winge is a PhD-student in safety management at the Norwegian University of
1332 Ragin, C. C. (2014). The comparative method: Moving beyond qualitative and 1397
Science and Technology in Trondheim. He is involved in research on accidents and
1333 quantitative strategies. Illinois: University of California Press. 1398
safety management. He also works for the Norwegian Labour Inspection Authority.
1334 Ragin, C. C. (2017). User’s guide to fuzzy set/qualitative comparative analysis 3.0. 1399
1335 Irvine, California: Department of Sociology, University of California. 1400
Eirik Albrechtsen is an associate professor in safety management at the Norwegian
1336 Ragin, C. C., & Davey, S. (2017). Fuzzy-set/qualitative comparative analysis 3.0. Irvine, 1401
University of Science and Technology in Trondheim. He has several publications and
1337 California: Department of Sociology, University of California. 1402
research activities on safety management in the construction industry and other
1338 Reason, J. (1997). Managing the risks of organizational accidents. Surrey: Ashgate 1403
industries.
1339 Publishing. 1404
1340 Reason, J. (2016). Organizational accidents revisited. Boca Raton: CRC Press. 1405
Jan Arnesen is an engineer and works with health, safety and environment in
1341 Reason, J.; Hollnagel, E.; Paries, J., Revisiting the Swiss cheese model of accidents, 1406
Statsbygg. He has worked within the building and construction industry for the last
1342 Project Safbuild. EEC Note No. 13/06, Eurocontrol: European Organisation for 1407
30 years. The last 13 years he has been dedicated to work with health, safety and
1343 the Safety of Air Navigation (2006). 1408
environment.
1344 Rihoux, B., & Lobe, B. (2009). The case for qualitative comparative analysis (QCA): 1409
1345 Adding leverage for thick cross-case comparison. In D. Byrne & C. C. Ragin (Eds.),
1346 The Sage Handbook of Case-based Methods (pp. 222–242). Sage Publications.

10 1016@j JSR 2019 09 015 PDF

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

10 1016@j JSR 2019 09 015 PDF

Încărcat de

Drepturi de autor:

Formate disponibile

JSR 1622 No.

of Pages 14, Model 5G

Journal of Safety Research xxx (xxxx) xxx

Contents lists available at ScienceDirect

Journal of Safety Research

3 A comparative analysis of safety management and safety performance

2 S. Winge et al. / Journal of Safety Research xxx (xxxx) xxx

S. Winge et al. / Journal of Safety Research xxx (xxxx) xxx 3

209 Summarized, the research on safety management in construc- Table 1

4 S. Winge et al. / Journal of Safety Research xxx (xxxx) xxx

S. Winge et al. / Journal of Safety Research xxx (xxxx) xxx 5

6 S. Winge et al. / Journal of Safety Research xxx (xxxx) xxx

S. Winge et al. / Journal of Safety Research xxx (xxxx) xxx 7

Row CC OC CO RM SP Cons. PRI. Cases

8 S. Winge et al. / Journal of Safety Research xxx (xxxx) xxx

Row CC OC CO RM SP Cons. PRI. Cases

High safety performance configurations Low safety performance configurations

Row RM PL RO MC SM SP Cons. PRI Cases

S. Winge et al. / Journal of Safety Research xxx (xxxx) xxx 9

Row RM PL RO MC SM SP Cons. PRI Cases

High safety performance configurations (frequency Low safety performance configuration

10 S. Winge et al. / Journal of Safety Research xxx (xxxx) xxx

S. Winge et al. / Journal of Safety Research xxx (xxxx) xxx 11

12 S. Winge et al. / Journal of Safety Research xxx (xxxx) xxx

S. Winge et al. / Journal of Safety Research xxx (xxxx) xxx 13

14 S. Winge et al. / Journal of Safety Research xxx (xxxx) xxx

S-ar putea să vă placă și