Journal of Safety Research 38 (2007) 627 – 641

www.elsevier.com/locate/jsr www.nsc.org

Safety culture: Analysis of the causal relationships between

its key dimensions
Beatriz Fernández-Muñiz ⁎, José Manuel Montes-Peón 1 , Camilo José Vázquez-Ordás 2
University of Oviedo, Facultad de Ciencias Económicas y Empresariales, Departamento de Administración de Empresas y Contabilidad,
Avda. del Cristo, s/n, 33071, Oviedo, Spain

Received 6 April 2005; received in revised form 14 May 2007; accepted 5 September 2007
Available online 5 November 2007

Abstract

Introduction: Several fields are showing increasing interest in safety culture as a means of reducing accidents in the workplace. The literature
shows that safety culture is a multidimensional concept. However, considerable confusion surrounds this concept, about which little
consensus has been reached. Method: This study proposes a model for a positive safety culture and tests this on a sample of 455 Spanish
companies, using the structural equation modeling statistical technique. Results: Results show the important role of managers in the
promotion of employees' safe behavior, both directly, through their attitudes and behaviors, and indirectly, by developing a safety
management system. Impact on Industry: This paper identifies the key dimensions of safety culture. In addition, a measurement scale for the
safety management system is validated. This will assist organizations in defining areas where they need to progress if they wish to improve
their safety. Also, we stress that managers need to be wholly committed to and personally involved in safety activities, thereby conveying the
importance the firm attaches to these issues.
© 2007 National Safety Council and Elsevier Ltd. All rights reserved.

Keywords: Safety culture; Safety management system; Management commitment; Employees' Involvement; Structural equation modeling

1. Introduction atmosphere in which employees are aware of the risks in their

Over the past two decades, researchers and practitioners
have shown increasing interest in the concept of safety culture
because of its impact on safety outcomes such as injuries,
fatalities, and other incidents (Choudhry, Fang, & Mohamed,
2006). Safety culture can be viewed as a component of the
organizational culture that refers to the individuals, jobs, and
organizational characteristics that affect employees' health and
safety. The aim of a positive safety culture is to create an
⁎ Corresponding author. Tel.: +34 985106215; fax: +34 985103708.
E-mail addresses: beatrizf@uniovi.es (B. Fernández-Muñiz),
jmmontes@uniovi.es (J.M. Montes-Peón), cvordas@uniovi.es
(C.J. Vázquez-Ordás).
1
Tel.: +34 985104975; fax: +34 985103708.
2
Tel.: +34 9853704; fax: +34 985103708.
0022-4375/$ - see front matter © 2007 National Safety Council and Elsevier Ltd. All rights reserved.
doi:10.1016/j.jsr.2007.09.001
workplace, are continually on guard against them (Ostrom,
Wilhelmsen, & Daplan, 1993), and avoid taking any unsafe
actions. Thus, safety culture can be considered an important
management tool helping to control workforce beliefs, atti-
tudes, and behaviors with regard to safety (Beck & Woolfson,
1999).
Recognition of the importance of the safety culture in
preventing occupational accidents has led to numerous
attempts to define and assess it in many organizations
(Arboleda, Morrow, Crum, & Shelley, 2003; Cooper, 2000;
Cox & Cheyne, 2000; Glendon & Stanton, 2000; Gulden-
mund, 2000; Farrington-Darby, Pickup, & Wilson, 2005;
Havold, 2005; Laurence, 2005; Lee & Harrison, 2000;
Richter & Koch, 2004; Sorensen, 2002). But gaps still remain
in the literature, and there is no universal agreement about
the definition or content of the term. Previous studies have
considered safety culture as a multidimensional concept, but
628 B. Fernández-Muñiz et al. / Journal of Safety Research 38 (2007) 627–641
without a clear structure and without consensus on the
specific dimensions. Empirical efforts trying to operationa-
lize the safety culture concept have been scarce (Seo, Torabi,
Blair, & Ellis, 2004). The literature has focused more on
analyzing employees' perceptions, attitudes, and behaviors
(Brown & Holmes, 1986; Cox & Cox, 1991; Coyle, Sleeman,
& Adams, 1995; Dedobbeleer & Beland, 1991; Diaz &
Cabrera, 1997; Melià & Sesé, 1999; Niskanen, 1994; Seo
et al., 2004; Silva, Lima, & Baptista, 2004; Williamson,
Feyer, Cairns, & Biancotti, 1997; Zohar, 1980). Researchers
have conceded rather less importance to measuring the situa-
tional characteristics of the safety management system, which
Cooper (2000) and Mearns, Whitaker, and Flin (2003) con-
sider to be an integral part of the organization's safety culture.
In an attempt to address these gaps in the literature, the
current study aims to provide an integrative definition of the
safety culture concept, identify its main indicators, and
analyze the possible relations between them, proposing a
model of positive safety culture. With this purpose, we carry
out an exhaustive review of the previous studies of safety
management and safety culture, which will provide the basis
for the formulation of the hypotheses. Subsequently, we
develop measurement scales of the concepts used in the
proposed model, in order to operationalize them, and carry
out an empirical study on a sample of 455 Spanish firms to
validate the scales and test the hypotheses. This process
involves the use of the structural equation modeling statistical
technique.
1.1. Structuring the concept of safety culture
Several meanings have been attributed to the term safety
culture, giving rise to numerous definitions (e.g., CBI, 1990;
Cooper, 2000; Cox & Cox, 1991; HSC, 1993; IAEA, 1986;
Pidgeon, 1991; Wallace & Neal, 2000; Wiegmann, Zhang,
Von Thaden, Sharma, & Mitchell, 2002). Nevertheless, most
of these are very broad and implicit (Guldenmund, 2000), and
no unanimously accepted definition exists. Despite this, it is
possible to identify various common aspects allowing us to
propose the following definition of a positive safety culture:
“A set of values, perceptions, attitudes and patterns of
behavior with regard to safety shared by members of the
organization; as well as a set of policies, practices and pro-
cedures relating to the reduction of employees' exposure
to occupational risks, implemented at every level of the
organization, and reflecting a high level of concern and
commitment to the prevention of accidents and illnesses.”
Researchers have carried out empirical studies to determine
the content of safety culture (e.g., Berends, 1996; Geller,
1994; Lee, 1996; Mearns et al., 2003; Ostrom et al., 1993;
Safety Research Unit, 1993). Other work has reviewed these
studies in an attempt to identify similar elements, but several
inconsistencies are apparent, and the authors' idiosyncratic
labeling of safety culture indicators often makes it extremely
difficult to reconcile them (Seo et al., 2004; Wiegmann et al.,
2002). Despite this, two factors — employees' involvement
and management commitment to safety — have been
properly replicated across studies (Cox & Cheyne, 2000;
Dedobbeleer & Beland, 1991; Hofmann & Stetzer, 1996;
Mearns et al., 2003).
Cooper (2000) offered an integrative framework to
analyze this construct, using Bandura's (1986) model of
reciprocal determinism. Cooper considers that the attitudes,
perceptions and beliefs of individuals, their behaviors, and
the safety management systems (objective situational fea-
tures) are elements that combine to form the organization's
safety culture. In this way, the previous literature review
helps identify three key indicators of the organization's
safety culture: managers' commitment to safety, employees'
involvement, and the policies and procedures that form the
safety management system. Likewise, the dependence rela-
tions between these three dimensions constitute the hypotheses
of the study.
1.2. Proposed Model of Safety Culture: Hypotheses
Previous research has shown that management commit-
ment to safety is a determining factor of employees' attitudes
and behaviors with respect to risk (e.g., Cheyne, Cox, Oliver,
& Tomás, 1998; Cohen, 1977; Dedobbeleer & Beland, 1991;
DeJoy, Schaffer, Wilson, Vandenberg, & Butts, 2004;
Donald & Canter, 1994; Eiff, 1999; Hofmann & Stetzer,
1996; Ostrom et al., 1993; O'Toole, 2002; Rundmo, 1996;
Rundmo & Hale, 2003; Seo et al., 2004; Simonds & Shafari-
Sahrai, 1977; Smith, Cohen, Cohen, & Cleveland, 1978;
Vecchio-Sadus & Griffiths, 2004; Zohar, 1980). A commit-
ted manager who is personally involved in safety activities
and who takes an interest in working conditions, conveys to
the employees a sense of the importance of safety for
the organization. As a result, the employees comply with
regulations, take the proper safety measures, and participate
actively in meetings and activities designed to promote
improvements in their workplace. Likewise, the manage-
ment policies, programs, and practices that form the safety
management system are regarded in several studies as a
precursor of the employees' perceptions about the impor-
tance of safety in their organization (DeJoy et al., 2004;
Hofmann, Jacobs, & Landy, 1995), and so contribute to
performing tasks in a safe manner (Cohen, 1977; DeJoy
et al., 2004; Diaz & Cabrera, 1997; Hofmann et al., 1995).
However, the implementation of the safety management
system in the general management of the firm requires the
support of managers, who should allocate financial resources
to the activities that make up such a system and should be
personally involved in them, showing a positive attitude and
total commitment and interest in the issue (Civil Aviation
Safety Authority, 2002; Zohar, 1980). Finally, the safety
management system will have a positive and direct effect
on the firm's accident rates (Petersen, 2000). Thus, this sys-
tem will reduce workers' injuries and illnesses, as well
as material damage. Consequently, it will reduce lost time
caused by interruptions in the productive process, cut labor
 B. Fernández-Muñiz et al. / Journal of Safety Research 38 (2007) 627–641
absenteeism, and at the same time improve workers'
satisfaction and motivation. In addition, safety performance
depends on the workers correctly applying the preventive
measures and on their suggestions for improving working
conditions (Brown, Willis, & Prussia, 2000; Donald &
Young, 1996; Eiff, 1999; O'Toole, 2002). In short, we can
formulate the following hypotheses on the dependence
relationships between the constituents of safety culture:
Hypothesis H1. Managers' commitment, through their
positive attitudes to prevention of accidents at work and
their behaviors, positively influences employees' involve-
ment and active participation in safety activities.
Hypothesis H2. A high degree of development of the safety
management system positively influences employees' invol-
vement and participation in safety activities.
Hypothesis H3. Managers' commitment conditions the
development of the safety management system by providing
financial, human and material resources.
Hypothesis H4. The level of development of the firm's
safety management system positively influences its safety
performance.
Hypothesis H5. Employees' participation in activities
relating to their health and safety positively affects the
firm's safety performance.
2. Method
2.1. Sample
In order to test the above hypotheses, we carried out
an empirical study. The majority of empirical studies limit
their sample to one organization from a very specific sector,
casting some doubt about their external validity (Silva et al.,
2004). Thus, with the aim of achieving as high a degree of
generalization as possible for the results of this work, we
considered the target population made up of companies located
in Spain, both national and foreign, from the construction,
industrial, and services sectors, employing more than 10
workers. With these criteria, and using the SABI database
(which holds accounting data on Spanish companies), we
obtained a population size of 62,146 firms, excluding non-
profit organizations. Subsequently, we calculated the number
of responses that would be necessary to ensure that our sample
would be representative of the total population, assuming a
maximum error of 5% for a confidence level of 95%. A total of
382 respondents were necessary. Assuming a similar response
rate to that obtained in other survey-based studies carried out in
Spain (Ordiz & Fernández, 2003; Ordoñez, 2002; Prieto &
Revilla, 2004) — 10% — we concluded that we needed to
contact and request the collaboration of 3,820 companies. We
then divided the population firms into groups according to firm
size (small: b50 workers; medium-sized: 50–250; large:
N 250) and sector of activity (industrial, construction, and
629
services). Finally, we selected 3,820 firms systematically at
random from each of these groups in the same proportion as in
the population as a whole.
We devised a questionnaire to compile the information. As
a sample unit we chose the safety officer, since this individual
can be expected to have most information about the specific
practices and procedures that are being carried out within
the firm, and be familiar with the difficulties involved in
implementing the system. This officer can also be expected to
have access to all types of information concerning potential
dangers to the health and safety of the workers. Likewise, these
officers occupy an intermediate position between management
and the employees, so we consider that their information will
be less biased and more accurate. We contacted this particular
officer in each firm by telephone in order to send them the
questionnaire together with a presentation letter. In this way,
we eliminated firms not wishing to take part in the study, and
guaranteed that the appropriate person would respond to the
questionnaire. When the firm lacked a safety officer and had
outsourced its occupational safety activities, which was the
case in some small firms, we contacted the safety coordinator
responsible for liaising between the firm and its external
service provider, in order not to exclude this type of firm from
the study. Finally, we received a total of 474 questionnaires.
Some of these had been incorrectly completed in some respect,
so we were obliged to contact the firms concerned again
requesting data that was either missing or erroneous. But 19 of
these questionnaires still had serious internal inconsistencies or
more than 80% of their values lost, so we decided to eliminate
them, following the complete case approach described in Hair,
Anderson, Tatham, and Black (1998). In this way, we obtained
a total of 455 valid responses. The definitive response rate was
12%. The sampling error was 4.57% for the most unfavorable
case, and a confidence level of 95%. Taking into account only
the population that we contacted, the error was 4.18%.
Although the response rate is not as high as in U.S. or UK
studies, the sample is representative because the sampling
error, which reflects the maximum difference between the
sample estimator and the corresponding population parameter,
is low.
Table 1 reports the firms' distribution by activity and firm
size for both the population and the sample, as well as the
response rates by activity and size. The differences between
the population and the sample are due to the different re-
sponse rates of the segments of population. We find that the
proportion of construction-sector firms is very similar in both
the sample and the total population. But the proportion of
industrial firms is higher in the sample than in the population
as a whole, while the proportion of service-sector firms is
lower in the sample than in the population. This circumstance
is due to the fact that the response rate is higher in the
industrial sector than in the service sector, this latter sector
not traditionally being associated with occupational risks.
Nevertheless, we observe an extremely diversified sectorial
structure, which can be interpreted as a significant reflection
of the Spanish productive structure. Similarly, the sample
630 B. Fernández-Muñiz et al. / Journal of Safety Research 38 (2007) 627–641

Table 1
Distribution of Population and Sample by Activity and Size
ACTIVITY SECTORS (CNAE 93) POPULATION CONTACTS QUESTIONNAIRES RESPONSE RATE
WITH FIRMS MADE RECEIVED (SAMPLE)
Industry 26,566 (42.7%) 1,538 (40.2%) 271 (60.0%) 17.6%
Food and drink (15) 35 (7.7%)
Textile, footwear and derivatives (17, 18 and 19) 21 (4.6%)
Wood industry (20 and 36) 23 (5.1%)
Chemical, paper and plastics industry (21, 24 and 25) 42 (9.2%)
Metallurgy industry (27 and 28) 95 (20.9%)
Machinery, vehicle and various equipment (29, 30, 33 and 34) 18 (4.0%)
Other manufacturers (22, 26, 31, 32 and 37) 39 (8.6%)
Services 28,257 (45.5%) 1,700 (44.5%) 107 (20.2%) 6.3%
Vehicle sales and repairs (50) 15 (3.3%)
Wholesale business (51) 25 (5.5%)
Retail business (52) 11 (2.4%)
Various services (55, 70, 71, 72, 92 and 93) 41 (9.0%)
Construction (45) 9,710 (15.6%) 583 (15.3%) 77 (19.8%) 13.2%
TOTAL COMPANIES 62,146 (100%) 3,820 (100%) 455 (100%) 12%

SIZE (no. employees) POPULATION CONTACTS WITH FIRMS MADE QUESTIONNAIRES RECEIVED (SAMPLE) RESPONSE RATE
10–49 48,702 (78.4%) 2,994 (78.4%) 241 (53%) 8%
50–249 10,985 (17.7%) 675 (17.7%) 160 (35.2%) 23.7%
More than 250 2,459 (3.9%) 151 (3.9%) 54 (11.8%) 35.7%
TOTAL COMPANIES 62,146 (100%) 3,820 (100%) 455 (100%) 12%
Source: SABI Database (Sistema de Análisis de Balances Ibéricos / System of Analysis of Spanish Accounts).

firms are fundamentally small and medium-sized firms
(88.2%) — a proportion in line with the Spanish economy.
But we find small firms with less than 50 employees to be
more reluctant to respond to the questionnaire, which leads to
a lower response rate from this type of firm. As a con-
sequence, some differences exist in the distribution by size
between the sample and the population.
2.2. Measurement Scales
The measurement scales of the concepts used in this work
were constructed following a multiple indicator approach.
Thus, each concept was measured using various items or
variables. This process for generating items involved succes-
sive stages. First, we conducted an exhaustive review of
previous studies on safety management and safety culture, both
theoretical (e.g., Guldenmund, 2000; HSE, 1997; Wiegmann
et al., 2002) and empirical (e.g., Grote & Künzler, 2000; Hayes,
Perander, Smecko, & Trask, 1998; McDonald, Corrigan, Daly,
& Cromie, 2000; Ostrom et al., 1993; Glendon & Litherland,
2001; Williamson et al., 1997; Zohar, 1980). This process
provided a pool of items to measure some of the concepts
considered in this research. But we observed that the evalua-
tion of the measurement scales' psychometric properties
was deficient in some studies, fundamentally with respect to
the safety management system. This is why we opted here to
build specific measurement scales, adapted to the theoretical
model developed in this study, taking into account not only
existing scales but also international standards and guidelines
in safety management. We now outline the sources of each
scale used.
2.2.1. Safety Management System (SMS)
This system is considered as the set of integrated mech-
anisms in the organization, comprising policies, strategies, and
procedures (Kirwan, 1998), designed to control the risks that
may affect employees' health and safety. Previous research has
debated the content of this management system (Grote &
Künzler, 2000; Hale, Heming, Carthey, & Kirwan, 1997;
Hurst, 1997; McDonald et al., 2000; Mitchison & Papadakis,
1999; Santos-Reyes & Beard, 2002), and many authors have
stressed its importance (Hale et al., 1997). But few empirical
works evaluate the psychometric properties of the scales
proposed to measure the SMS concept (Carder & Ragan, 2003;
Grote & Künzler, 2000), and no consensus has been reached
about the specific dimensions making up the system. The
concept is non-operationalized, which makes it difficult to
study its antecedents and consequences. Thus, the first ob-
jective of this work is to develop and validate a scale
measuring and evaluating Spanish firms' safety management,
which we can subsequently use to test the hypotheses for-
mulated above. So we combine characteristics from the
management systems proposed in the international standards
and guidelines (BSI, 1996, 1999; HSE, 1997; ILO, 2001). We
also take into account studies analyzing the practices that
discriminate between companies with high and low accident
rates (Bentley & Haslam, 2001; Cohen, 1977; Cohen, Smith,
& Cohen, 1975; Mearns et al., 2003; Shafai-Sahrai, 1971;
Shannon, Mayr, & Haines, 1997; Smith et al., 1978; Tam,
Zeng, & Deng, 2004; Vredenburgh, 2002), as well as studies of
safety culture (e.g., Cox & Cheyne, 2000; Farrington-Darby
et al., 2005; Guldenmund, 2000; Havold, 2005; Lee &
Harrison, 2000; Olive, O'Connor, & Mannan, 2006). Thus, we
 B. Fernández-Muñiz et al. / Journal of Safety Research 38 (2007) 627–641
consider that an adequate safety management system must
contain six key dimensions: safety policy, incentives for
employee participation, training, communication, planning,
and control. Apart from the theoretical contributions from the
abovementioned literature, in the process of developing the
measurement scales of each dimension we used as reference
the scales used in the works mentioned below.
2.2.1.1. safety policy. This concept captures the existence
of a written declaration reflecting the organization's commit-
ment to safety, as well as its integration with the firm's other
policies (HSE, 1997; National Safety Council, 2001). Four
items measure this variable, adapted from Mearns et al. (2003)
and Melià and Sesé (1999).
2.2.1.2. incentives. This variable measures the extent to
which the firm encourages its workers to participate in ac-
tivities relating to their safety. Five items measure this variable,
selected from the measurement scales developed by Carder
and Ragan (2003), Glendon and Litherland (2001), Hayes
et al. (1998), Melià and Sesé (1999), and Vredenburgh (2002).
2.2.1.3. training. This construct captures the existence of
training plans to develop employee competences and skills in
safety. Nine items measure this concept, adapted from Carder
and Ragan (2003), Coyle et al. (1995), Donald and Canter
(1994), Glendon, Stanton, and Harrison (1994), Grote and
Künzler (2000), and Vredenburgh (2002).
2.2.1.4. communication. This concept refers to the transfer
of information to employees about the possible risks in the
workplace and the correct way to combat them. Four items
measure this construct, adapted from the scales developed by
Bentley and Haslam (2001), Cox and Cheyne (2000), Donald
and Canter (1994), Glendon and Litherland (2001), Mearns
et al. (2003), and Melià and Sesé (1999).
2.2.1.5. planning. This variable is measured using two
constructs called preventive planning and emergency plan-
ning (BSI, 1999; HSE, 1997). The first captures the existence
of procedures to evaluate the risks and establish the necessary
safety measures for avoiding accidents, and includes seven
items. Emergency planning captures the existence of an
organized plan in case of emergency, and includes four items.
The items making up each subscale are adapted from the
works of Bentley and Haslam (2001), Coyle et al. (1995),
Glendon and Litherland (2001), Grote and Künzler (2000),
and Wu and Lee (2003).
2.2.1.6. control. This concept is measured using two
constructs called internal control and benchmarking techni-
ques. Internal control captures the existence of mechanisms
to verify the extent to which objectives are fulfilled, as well as
the degree of compliance with the internal norms or work
procedures (Hurst, Young, Donald, Gibson, & Muyselaar,
1996; HSE, 1997). The benchmarking techniques construct
631
measures the extent to which the firm compares its safety
actions and performance with other firms (Fuller, 2000). Six
and two items, respectively, are used to measure these con-
structs, adapted from the scales developed by Bentley and
Haslam (2001), Melià and Sesé (1999), and Vredenburgh
(2002).
2.2.2. Managers' commitment
This variable measures the extent to which the firm's
managers are committed to their workers' safety. This com-
mitment can be manifested in the positive attitudes toward the
activities relating to safety management and in the behaviors
visible to the workers. Thus, the managers' commitment con-
cept is measured using two constructs: attitudes and behaviors.
These factors are measured using nine items, adapted from the
scales developed by Carder and Ragan (2003), Cox and
Cheyne (2000), Donald and Canter (1994), Grote and Künzler
(2000), Mearns et al. (2003), Rundmo and Hale (2003),
Vredenburgh (2002), and Wu and Lee (2003).
2.2.3. Employees' involvement
The four items of this construct, which reflects the degree
of workers' compliance with the safety procedures and the
extent to which they participate in improving working
conditions, are adapted from the works of Carder and Ragan
(2003), Cox and Cheyne (2000), Cox and Cox (1991), Hayes
et al. (1998), Mearns et al. (2003), Vredenburgh (2002), and
Wu and Lee (2003).
All the items from the previous scales were measured on a
5-point Likert-type scale, the respondents being asked to
indicate their perceptions of their company on each item
ranging from “5 = strongly agree” to “1 = strongly disagree.”
2.2.4. Safety performance
This construct captures the direct consequences of the
safety culture model proposed in this work. Previous research
has not produced a consensus about the outcome or con-
sequences of safety culture (Cooper, 2000). Indeed, Glendon
and Litherland (2001) point to the lack of an adequate measure
of safety performance as one limitation associated with
evaluating the effectiveness of different safety programs.
Some authors argue that the reduction in accident and incident
rates provides the best results measure of the safety culture
(Clarke, 1998), and accident or injury data of various forms
have been used in a number of studies (Mearns et al., 2003;
Niskanen, 1994; O'Toole, 2002; Silva et al., 2004; Vreden-
burgh, 2002; Zohar, 2000). Nevertheless, using objective
accident data to measure safety performance is notoriously
problematic (Cooper & Phillips, 1994), because such data are
insufficiently sensitive, of dubious accuracy, retrospective,
ignore risk exposure (Glendon and Litherland, 2001), and tend
to be very unstable (DeJoy et al., 2004; Havold, 2005). More-
over, accident rates can be reduced for a number of reasons
having little to do with the safety culture, such as, for example,
not declaring incidents due to the existence of incentive
schemes. Objective accident data are not very reliable, since:
632 B. Fernández-Muñiz et al. / Journal of Safety Research 38 (2007) 627–641
near misses are difficult to collect; staff are reticent about
reporting accidents; and an accident free period can suppress
accident reporting (Havold, 2005). In addition, implementing
a safety management system, such as the one described in this
study, implies an improvement in the process of recording and
analyzing any accidents or incidents that occur. Hence the
firm's performance in this area may improve, while at the same
time the objective indicators of the phenomenon are rising. To
get around these limitations, researchers often resort to
surrogate measures to assess safety-related outcomes (DeJoy
et al., 2004). Other studies have used self-reporting compo-
nents of safety performance as outcome variables, or these
have been estimated subjectively. In this work, the construct
was measured, like in various previous works (DeJoy et al.,
2004; Lu & Shang, 2005), perceptually, using self-reported
data. Respondents were asked to provide information relating
to their safety performance in terms of their degree of satis-
faction with: (a) the number of personal injuries; (b) the
material damage; (c) the employees' motivation; and (d) the
absenteeism or lost time; following Lu and Shang's (2005)
procedure. Satisfaction was measured considering the evolu-
tion of these indicators in previous years and their relative
position compared to the sector average, using a 5-point Likert
scale ranging from “5 = extremely satisfied” to “1 = extremely
dissatisfied.” Thus, the better the evolution of the indicators,
the more satisfied the firm is with them, and hence the closer
the response is to five.
Having obtained an initial structure of items, we elaborated
a first draft of the questionnaire that would be later used to
collect the information. In a second phase, the draft ques-
tionnaire was submitted to a refinement process in order to
eliminate items that were highly redundant because of
analogous meanings. For this, we considered the opinion of
various doctors in business management, and carried out
four in-depth interviews with experts on occupational safety
from the Institute for the Prevention of Occupational Risks
of the region of Asturias (Spain). These experts were
engineers with considerable experience identifying and
controlling risks in the workplace, having spent more than
20 years advising and supporting companies and monitoring
compliance with the current legislation in the area of
occupational safety. Participants were provided with a list of
the refined scale items, and were requested to comment on
the relevance of each statement to companies. These
interviews revealed the need to modify the starting items.
Specifically, we improved the wording of some questions to
aid understanding in the target audience, introduced two
variables that had not been initially considered, and elimi-
nated three items that did not add any information. Finally,
after selecting the definitive items we subjected the ques-
tionnaire to a pilot test with safety officers from eight
companies. The officers were asked whether they under-
stood the specific terms used, and asked about the intended
meaning of each statement. The questionnaire finally
consisted of a total of 57 items. The final composition of
the scales can be seen in the Appendix.
3. Results
3.1. Estimation of Measurement Model
The scales proposed were subjected to a process of
evaluation, focusing on the study of their psychometric
properties. Specifically, we analyzed the dimensionality,
examined the reliability of their composition, and evaluated
the content, convergent, and discriminant validity of each
subscale, following the original proposals of Churchill
(1979) and Anderson and Gerbing (1988). For this purpose,
we conducted principal components exploratory factor
analyses with varimax rotation, and confirmatory factor
analyses using structural equation modeling, employing the
statistics programs SPSS/PC version 11.5 for Windows and
EQS version 5.7a for Windows, respectively.
3.1.1. Dimensionality Study
The evaluation process began by carrying out a principal
components exploratory factor analysis, considering all the
items proposed for each dimension. This analysis reflected a
composition of the scales in accordance with the starting
assumptions, in other words, it revealed the unidimensional
nature of the factors policy, incentives, training, communica-
tion, employee involvement and safety performance, and the
bidimensional nature of planning (preventive and emergency),
control (internal and benchmarking), and managers' commit-
ment (attitudes and behaviors). Subsequently, structural
equation modeling (SEM) was used to perform a confirmatory
factor analysis on the proposed model. The goodness-of-fit
measures of the initial safety management system model
showed that the model fits the data poorly. Consequently, an
iterative process was applied, whereby items contributing to
lack of fit, that is, items not converging sufficiently on their
corresponding latent variable (Anderson & Gerbing, 1988;
Steenkamp & Van Trijp, 1991), were systematically dropped.
The best model fit possible was achieved after reducing the
proposed safety management system scale from 40 to 29
explanatory variables structured in eight subscales. The items
finally eliminated appear in italics in the Appendix. Table 2
reports the results obtained in the estimation of these re-
specified models. The models clearly present satisfactory
goodness-of-fit indices (Bentler, 1995; Hair et al., 1998). We
then proceeded to evaluate the reliability and validity of this
final proposal of items for each of the subscales making up the
safety management system, managers' commitment, employ-
ee involvement, and safety performance.
3.1.2. Reliability Analysis
The reliability study indicates the degree of internal
consistency between the multiple variables that make up the
scale, and represents the extent to which the indicators or
items of the scale are measuring the same concepts. For the
purpose of guaranteeing the maximum reliability of the
scales proposed, Cronbach's α coefficient (Cronbach, 1951)
and the Composite Reliability Index were calculated for each
 B. Fernández-Muñiz et al. / Journal of Safety Research 38 (2007) 627–641 633

Table 2
First-order Confirmatory Factor Analysis (1st Order CFA)
1st Order CFA for Safety Management System
Dimension Variables Cronbach Composite Standardized t Values Dimension- Correlation Confidence Interval
α Reliability Indexes Lambda Parameters Dimension
Prevention Policy (PP) 0.7131 0.717 PP-IN 0.577 (0.477–0.677)
Polic1 0.65 13.114 PP-TR 0.649 (0.549–0.749)
Polic2 0.63 14.160 PP-CO 0.631 (0.531–0.731)
Polic4 0.75 15.611 PP-PrP 0.721 (0.625–0.817)
Incentives to participation (IN) 0.7312 0.747 PP-EP 0.465 (0.357–0.573)
Incent1 0.57 13.189 PP-IC 0.658 (0.568–0.748)
Incent3 0.52 11.216 PP-BE 0.458 (0.352–0.564)
Incent4 0.80 21.183 IN-TR 0.755 (0.677–0.833)
Incent5 0.69 17.359 IN-CO 0.808 (0.746–0.870)
Training (TR) 0.7818 0.785 IN-PrP 0.523 (0.421–0.625)
Train1 0.71 16.803 IN-EP 0.386 (0.288–0.484)
Train3 0.74 18.552 IN-IC 0.629 (0.051–0.707)
Train6 0.62 15.337 IN-BE 0.450 (0.342–0.558)
Train8 0.59 13.395 TR-CO 0.899 (0.847–0.951)
Train9 0.58 11.776 TR-PrP 0.650 (0.560–0.740)
Communication (CO) 0.8046 0.808 TR-EP 0.463 (0.367–0.559)
Commun1 0.80 22.704 TR-IC 0.809 (0.743–0.875)
Commun2 0.76 20.474 TR-BE 0.470 (0.368–0.572)
Commun4 0.74 20.465 CO-PrP 0.676 (0.594–0.758)
Preventive Planning (PrP) 0.7636 0.770 CO-EP 0.552 (0.464–0.640)
Planning3 0.66 11.816 CO-IC 0.830 (0.772–0.888)
Planning6 0.83 19.326 CO-BE 0.543 (0.449–0.637)
Planning7 0.69 14.531 PrP-EP 0.532 (0.432–0.632)
Emergency Planning (EP) 0.8496 0.856 PrP-IC 0.752 (0.682–0.822)
Planning8 0.82 18.288 PrP-BE 0.397 (0.295–0.499)
Planning9 0.80 24.075 EP-IC 0.535 (0.451–0.619)
Planning10 0.86 23.118 EP-BE 0.379 (0.279–0.479)
Planning11 0.59 16.448 IC-BE 0.580 (0.496–0.664)
Internal Control (IC) 0.8543 0.859
Contro1 0.79 20.189 Results of Model Fit:
Contro2 0.81 23.773 S-Bχ2 (349) = 855.561
Contro3 0.71 20.167 p = 0.001
Contro4 0.81 23.522 RMSEA = 0.057 IFI = 0.920
Contro5 0.58 10.613 BNNFI = 0.901 GFI = 0.879
Benchmarking (BE) 0.8231 0.825 CFI = 0.920 AGFI = 0.849
Contro6 0.79 19.823
Contro7 0.88 19.922

1st Order CFA for Managers' Commitment and Employees' Involvement

Dimension Variable Cronbach Composite Standardized t Values Dimension-Dimension Correlation
α Reliability Indexes Lambda Parameters (95% Confidence Interval)
Managers' Attitudes (MA) .8449 0.844 MA-MB
Commit1 0.69 10.018 0.823
Commit2 0.78 15.405 (0.759–0.887)
Commit3 0.77 15.756 MA-EI
Commit4 0.79 16.705 0.566
Managers' Behaviors (MB) 0.8334 0.840 (0.484–0.648)
Commit5 0.74 14.982 MB-EI
Commit6 0.82 19.842 0.526
Commit7 0.65 12.081 (0.442–0.610)
Commit8 0.72 15.112
Commit9 0.64 15.685 Results of Model Fit:
Employees' Involvement (EI) 0.7610 0.780 S-Bχ2 (62) = 130.73
Involv1 0.82 21.691 p = 0.000
Involv2 0.88 23.126 RMSEA = 0.059 IFI = 0.963
Involv3 0.51 9.950 BBNNFI = 0.953 GFI = 0.948
Involv4 0.50 11.368 CFI = 0.963 AGFI = 0.924
(continued on next page)
634 B. Fernández-Muñiz et al. / Journal of Safety Research 38 (2007) 627–641
Table 2 (continued )
1st Order CFA for Safety Performance
Dimension Variable Cronbach Composite
α Reliability Indexes
Safety Performance 0.7433 0.746
Safety1
Safety2
Safety3
Safety4
Note: values above 1.96 indicate significance at 95% confidence level.
one-dimensional critical factor identified in the previous
section. As Table 2 shows, all the unidimensional subscales
present Cronbach α coefficients greater than 0.7, considered
to be an adequate level of reliability to test causal relations
(Nunnally, 1978). Also, the Composite Reliability Index in
all cases exceeds the minimum level of 0.6 recommended by
Bagozzi and Yi (1988).
3.1.3. Validity Analysis
The validity of the scales was verified by considering the
content validity, convergent validity, and discriminant validity.
The content validity is confirmed bearing in mind that the
proposed scales were designed following an exhaustive review
of the literature and subjected to a process of revision
involving in-depth interviews with safety experts. The con-
vergent validity of a concept evaluates the extent to which two
measurements of the concept may be correlated (Hair et al.,
1998). Convergent validity can be analyzed by means of
standardized factorial regression coefficients relating each
variable observed with the latent one (Anderson & Gerbing,
1988), in other words, by means of standardized lambda
parameters. A strong condition of convergent validity is that
those coefficients are over 0.5 and significant at a confidence
level of 95%, which requires t values greater than 1.96. The
values of the coefficients and the t values appear in Table 2.
This table shows that all subscales fulfill both conditions,
therefore confirming the convergent validity of the proposed
scales. The discriminant validity indicates the extent to which
two conceptually similar concepts differ, and was verified by
Anderson and Gerbing's (1988) methodology, which involves
estimating the confidence interval around the parameters that
indicate the correlation between the eight unidimensional
factors of safety management system on the one hand, and
the two unidimensional factors of managers' commitment
(attitudes and behaviors) and employees' involvement on the
other, bearing in mind the value of these parameters and the
corresponding standardized errors, to check that no interval
includes one.
In order to confirm the bidimensional nature of planning,
control and managers' commitment, we ran a second-order
Confirmatory Factor Analysis, and noted that in all cases the
existence of two factors or dimensions improves the model fit.
Standardized t Values Dimension-Dimension Correlation
Lambda Parameters (95% Confidence Interval)
Results of Model Fit:
S-Bχ2 (2) = 18.037
0.760 14.404 p = 0.001
0.716 11.900 RMSEA = 0.13 IFI = 0.962
0.592 10.683 BBNNFI = 0.885 GFI = 0.981
0.519 9.795 CFI = 0.962 AGFI = 0.903
In addition, the resulting standardized lambda parameters
were evaluated, as well as the confidence intervals of the
correlations between the factors, in order to re-check the
convergent and discriminant validity. Finally, we ran a third-
order Confirmatory Factor Analysis to confirm that the six
dimensions of policy, incentives, training, communication,
planning, and control converge on a single latent variable
termed safety management system. Fig. 1 shows that the
model fit is satisfactory. Additionally, the regression coeffi-
cients of the safety management system factor are much
greater than 0.5 and significant at the 95% confidence level,
confirming the convergent validity of the concept. Thus, the
reliability and validity of the safety management system scale
have been demonstrated, and it is possible to structure its 29
items in six dimensions that refer to the practices to which they
are linked. Likewise, the reliability and validity of the mea-
suring scales of managers' commitment, employee involve-
ment, and safety performance are also confirmed, meaning
they can be used to test the hypotheses.
3.2. Estimation of Proposed Causal Model
To test the proposed model, we use the structural equation
modeling statistical technique. This methodology, which is
frequently used in the literature (Brown et al., 2000; Huang,
Ho, Smith, & Chen, 2006; Seo, 2005; Silva et al., 2004; Siu,
Phillips, & Leung, 2004), allows us to test complex models of
relations between variables considering all the model relations
simultaneously. The goodness-of-fit indices of the suggested
model shown may be considered satisfactory (Fig. 2), since
they are very close to the recommended values.
With regard to testing the hypotheses put forward, the
coefficients reflected in the model confirm that managers'
commitment has a direct, positive, and statistically significant
influence on both employees' involvement and safety man-
agement system, corroborating hypotheses H1 and H3,
respectively. Hypothesis H2 is also accepted, since the results
show that safety management system exerts a significant,
positive influence on employees' involvement. Finally, safety
management system and employees' involvement both have a
significant, positive influence on safety performance, provid-
ing support for hypotheses H4 and H5, respectively.
 B. Fernández-Muñiz et al. / Journal of Safety Research 38 (2007) 627–641 635

Fig. 1. Occupational Safety Management System Measurement Scale. (Third-order Confirmatory Factor Model). Note: t-values in paretheses.

Fig. 2. Results of Estimation of Proposed Model. Note: ⁎⁎significant at 95% level. ⁎significant at 90% level.
636 B. Fernández-Muñiz et al. / Journal of Safety Research 38 (2007) 627–641
4. Discussion
Nowadays, there is wide recognition in the literature that
safety culture is an essential element in the organization's
efforts to prevent accidents in the workplace. But authors have
not reached a consensus about its content, or how this culture is
reflected within the organization. This study has reviewed
the main papers relating to safety culture, and on this basis
has formulated an integrative definition of the concept and
identified three key indicators of the organization's safety
culture: the managers' commitment, the employees' involve-
ment, and the safety management system. Thus, this work
structures the safety culture concept, contributing to the existing
knowledge about it. These findings mean that we can say that
an organization possesses a safety culture when it has adopted
an adequate safety management system, in other words,
when the firm: (a) has defined a safety policy that reflects the
organization's principles and values in this area; (b) has
established incentives for the workers to become involved in
safety activities; (c) provides its employees with continuous
training, so that they can carry out their jobs in the healthiest and
safest way possible; (d) provides fluid information about the
risks to which workers are exposed, and the correct way of
combating them; (e) plans the actions to carry out in order to
avoid accidents occurring (preventive planning), and to be able
to react quickly in case of emergency (emergency planning);
and finally, (f) has in place a control or feedback on actions
taken in the organization, through an analysis of the working
conditions and events occurring inside the firm (internal
control), and by comparing the firm with other companies
(benchmarking techniques). Likewise, the managers manifest a
strong commitment toward safety, showing a continuing
interest in their employees' working conditions, and are per-
sonally involved in the safety activities. Finally, the workers are
conscious of the importance of working safely, so that they
comply with the regulations and working procedures, partic-
ipate actively in meetings, and offer suggestions about how to
improve safety in their workplaces. The dimensions of safety
culture identified are not independent but are interrelated,
allowing us to build a model leading to a positive safety culture.
This model is tested, with the following conclusions.
This study, in line with Donald and Young (1996), suggests
that employees' involvement is fundamental for the proper
development of the firm's technical system, since human
factors play an important role in the organization's safety
performance. Employees constitute the final barrier to risk, and
their behavior is crucial to avoid both personal and material
damage (Eiff, 1999; Hofmann & Stetzer, 1996). Similarly,
employees' involvement leads to a reduction in absenteeism
and an improvement in workers' satisfaction and motivation,
as they feel they are an important part of the organization, and
that their managers value their opinions and contributions. In
this way, the employees are more strongly identified with and
committed to their organization, contributing to its interests
(Vecchio-Sadus & Griffiths, 2004). Likewise, we find that
employees' behavior and involvement in safety activities is
positively influenced by the managers' commitment and by
the safety management system implemented in the organiza-
tion, which, in turn, is conditioned by managers' commitment.
Unsafe acts or human failures are frequently a product of latent
conditions, in other words, of a combination of managerial
and organizational factors and local workplace factors that
predispose the worker to act unsafely (Reason, 1997, 1998),
such as, for example, a lack of awareness of risks, lack of
adequate training, perception of the scant importance attached
by the firm to risk prevention, poor preventive organization, or
insufficient management controls. All these factors form part
of the safety management system, and at the same time
illustrate the managers' commitment to safety. As a result, it
should be borne in mind that employees cannot be blamed for
causing accidents, since human error is a consequence rather
than a cause (Reason, 1997). Thus, managers play an essential
role in reducing occupational accident rates, since they have a
dual influence on employees' attitudes and behaviors. On the
one hand, managers have a direct influence through their
positive attitude to safety and thorough their behaviors. On the
other, the managers have an indirect influence through their
support and funding for the implementation and development
of the safety management system. If employees perceive a
high degree of managerial commitment, and, moreover,
this commitment is supported by the implementation of an
adequate safety management system, the employees will tend
to be positive in their attitudes toward safety, they will be less
inclined to commit unsafe acts, and more likely to make
suggestions and comments on improving work conditions.
These findings are consistent with Michael, Evans, Jansen, and
Haight's (2005) study.
Consequently, this work not only identifies the important
dimensions of safety culture, but also indicates the steps to
follow in order to achieve a positive safety culture. Given the
chain of relations found, the process of improving the safety
outcomes must start with managers' commitment. Managers
should be strongly committed to safety and should be extremely
sensitive to the consequences of accidents and serious injuries
and their effects on the victims. This commitment should be
reflected in managers' awareness of problems, in their concern
for safety, in their conviction that high levels of safety can be
attained (Cohen, 1977; Simonds & Shafari-Sahrai, 1977; Smith
et al., 1978), in their ability to demonstrate a continuous
positive attitude to safety, in their ability to promote safety
actively at all levels of the organization, in their low tolerance of
failure to comply with safety norms, and in their prioritization
of safety in the production. But attitudes and positive words
contained in a formal declaration of principles (Kletz, 1985) are
insufficient to modify employee behavior and reduce accident
rates. Good intentions have to be translated into daily
experience, and management decisions should reflect an
interest in investing, not only for further profit, but also for
people. Consequently, managers should allocate human and
financial resources to the safety management system and
become personally involved in it. This involvement at the
managerial level is reflected, for example, in frequent and
 B. Fernández-Muñiz et al. / Journal of Safety Research 38 (2007) 627–641
informal communications between employees and managers,
in managers' presence and contribution in safety meetings,
safety committees and training courses, or in regular visits to the
workplace to assess working conditions. Additionally, man-
agers should be prepared to delegate decision-making to the
workforce (empowerment), so that these latter can be actively
involved in the development and intervention of safety policies,
instead of assuming a passive role as the targeted recipients, as
Williamson et al. (1997) suggest.
The findings of this study are generalizable to firms from
various sectors and of various sizes, given the sample selection
criteria. Thus, the proposed model can be used as a manage-
ment tool to guide organizations' safety activities and improve
safety performance across diverse sectors. Managers can use
this model to compare their safety management practices with
those of other organizations, and to identify deficiencies in
their current safety practices. This will enable organizations to
improve the overall effectiveness of their safety activities and
reduce their accident rates.
Empirically testing the hypotheses proposed in this work
required developing measurement scales of the concepts
contained in the model. These scales were the result of the
literature review and of a rigorous process of refinement that
demonstrated the reliability and validity of the scales. Thus, this
work is a step forward in the operationalization process of the
safety culture concept, and helps to fill a gap in the literature.
4.1. Limitations of Study
First, although the constructs used in this research have
been defined as precisely as possible, based on the available
relevant literature, clearly articulating our conceptual frame-
work and carrying out a meticulous process of generation
and revision of items, the measures developed should be
understood as an approximation to latent phenomena, which
cannot be measured in full. Moreover, it should be borne in
mind that the relations have been evaluated from the viewpoint
of the firm safety officer. Employees' opinions have not been
taken into account, an aspect that would give a more complete
picture of the results. Finally, the study was carried out at a
specific moment in time, and therefore it is a cross-sectional
survey. Thus, conclusions of a causal type may be debatable.
But the methodology used does allow us to reject causal
models that do not confirm to the patterns of association
detected in the data, and accept models that do confirm (Melià,
2004). In our case, the results indicate that the data are con-
sistent with the hypothesized causal relations.
4.2. Future Lines of Research
The limitations detected in this study indicate possible
future lines of research. Thus, it would be extremely valuable
to obtain information on the employees, to evaluate the effect
that some safe working conditions have on their satisfaction
and motivation, and to confirm the effect of managers' com-
mitment on employees' attitudes and behaviors with regard to
637
safety. Likewise, and due to the time lag between improving
working conditions and obtaining superior performance, it
would be interesting to obtain a data panel allowing us to
confirm the causal relations detected. Finally, further study of
various aspects is needed, such as the organizational factors
that encourage or hinder the creation of the safety culture and
implementation of the safety management system, and com-
panies involved in different types of management should be
compared. It would also be useful to relate this model of safety
culture with economic and financial performance, providing
an incentive for managers' commitment.
5. Impact on Industry
The above conclusions lead us to deduce a wealth of im-
plications for business management, with a view to highlight-
ing aspects that management should address. The efficiency
of a safety management system depends on its capacity to
encourage the involvement of the workforce and decentralize
decision-making in this matter. This provides the organization
with flexibility in its risk management. Problems are quickly
resolved, since any hazards are remedied or corrected by
the workers affected. This situation will lead to autonomy in
control, reducing supervisory and monitoring costs, which
enables continuous improvement of working conditions with-
out increasing prevention costs. A considerable amount of
training is needed for this. Also, firm managers are highly
committed and personally involved in safety activities, con-
veying the importance accorded to work safety by the
organization, given their considerable influence on employee
behavior.
6. Summary
A number of studies highlight the important role of safety
culture in determining the frequency of accidents in the work-
place. Nevertheless, this is not a clear concept, there is no
unanimously accepted definition, and there is scant agreement
on its indicators. This study has devised a model of positive
safety culture. Thus, it is considered that the safety culture
within the firm comprises organization members' attitudes and
behaviors with respect to safety and the safety management
system. We also tested a model of causal relations between the
key dimensions of safety culture, finding that firm managers
play a fundamental role in reducing the number of unsafe acts
by employees, and hence in reducing accident rates.
Acknowledgements
This study has been supported by funds from the following
research projects:
1. “Integration of prevention of occupational risk in the
strategic management of the company,” financed by the
Regional Ministry of Industry and Employment of the
Principality of Asturias (Consejería de Industria y Empleo
638 B. Fernández-Muñiz et al. / Journal of Safety Research 38 (2007) 627–641
del Principado de Asturias, Ref: SV-PA-03-15), the
Regional Ministry of Education and Science of the
Principality of Asturias (Consejería de Educación y Ciencia
del Principado de Asturias), and the European Union,
through FEDER funds (Ref: FC-03-PB02-068).
2. “Determining factors in the development of safety manage-
ment of Spanish companies,” financed by the University of
Oviedo (MB-03-512-2).
References
Anderson, J. C., & Gerbing, D. W. (1988). Structural Equation Modeling in
practice: A review and recommend Two-Step Approach. Psychological
Bulletin, 103, 411−423.
Arboleda, A., Morrow, P., Crum, M., & Shelley, M. (2003). Management
practices as antecedents of safety culture within the trucking industry:
similarities and differences by hierarchical level. Journal of Safety
Research, 34, 189−197.
Bagozzi, R. P., & Yi, Y. (1988). On the evaluation of Structural Equation
Models. Journal of the Academic of Marketing Science, 16, 74−94.
Bandura, A. (1986). Social Foundations of Thought and Action: A Social
Cognitive Theory. Englewood Cliffs, NJ: Prentice-Hall.
Beck, M., & Woolfson, C. (1999, January). Safety Culture: a concept too
many? The Safety & Health Practitioner, 14−16.
Bentler, P. M. (1995). EQS Structural Equations Program Manual.
Multivariate Software. Encina, CA.
Bentley, T. A., & Haslam, R. A. (2001). A comparison of safety practices
used by managers of high and low accident rate postal delivery offices.
Safety Science, 37, 19−37.
Berends, J. J. (1996). On the Measurement of Safety Culture (Unpublished
graduation report). Eindhoven: Eindhoven University of Technology.
Brown, R. L., & Holmes, H. (1986). The use of a Factor-Analytic Procedure
for assessing the validity of an employee safety climate model. Accident
Analysis and Prevention, 18, 455−470.
Brown, K. A., Willis, P. G., & Prussia, G. E. (2000). Predicting safe employee
behavior in the steel Industry: Development and test of a Sociotechnical
Model. Journal of Operations Management, 18, 445−465.
BSI. (1996). Occupational Health and Safety Management Systems. British
Standards 8800. London: Author.
BSI. (1999). OHSAS 18001:1999, Occupational health and safety
management systems – specification. London: Author.
Carder, B., & Ragan, P. W. (2003). A survey-based system for safety
measurement and improvement. Journal of Safety Research, 34, 157−165.
CBI. (1990). Developing a Safety Culture — Business for Safety. London:
Confederation of British Industry.
Cheyne, A., Cox, S., Oliver, A., & Tomás, J. M. (1998). Modeling employee
attitudes to safety. Work and Stress, 12, 255−271.
Choudhry, R. M., Fang, D., & Mohamed, S. (2006). The nature of safety
culture: A survey of the state-of-the-art. Safety science. doi:10.1016/j.
ssci.2006.09.003.
Churchill, G. A. (1979). A paradigm for developing better measures for
marketing constructs. Journal of Marketing Research, 16, 64−73.
Civil Aviation Safety Authority. (2002). Safety Management Systems:
Getting Started. Australia: PMP.
Clarke, S. (1998). The Measurement of safety culture. Paper presented at the
24th International Congress of Applied Psychology, Safety Culture
Symposium, San Francisco, CA, USA.
Cohen, A. (1977). Factors in successful occupational safety programs.
Journal of Safety Research, 9, 168−178.
Cohen, A., Smith, M., & Cohen, H. H. (1975). Safety Programme Practices
in High vs Low Accident Rate Companies: An interim Report.Cincinnati,
OH: National Institute for Occupational Safety and Health Publication
No. 75-185.
Cooper, M. D. (2000). Toward a model of Safety Culture. Safety Science, 36,
111−136.
Cooper, M. D. & Phillips, R. A. (1994). Validation of a Safety Climate
measure. Paper presented at the British Psychological Society, Annual
Occupational Psychology Conference, Birmingham.
Cox, S., & Cheyne, J. (2000). Assessing safety culture in offshore environments.
Safety Science, 34, 111−129.
Cox, S., & Cox, T. (1991). The structure of employee attitudes to safety: An
European example. Work and Stress, 12, 189−201.
Coyle, I. R., Sleeman, S. D., & Adams, N. (1995). Safety Climate. Journal
of Safety Research, 26, 247−254.
Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests.
Psychometrika, 16, 297−334.
Dedobbeleer, N., & Beland, F. (1991). A safety climate measure for
construction sites. Journal of Safety Research, 22, 97−103.
DeJoy, D. M., Schaffer, B. S., Wilson, M. G., Vandenberg, R. J., & Butts, M. M.
(2004). Creating safer workplaces: Assessing the determinants and role of
safety climate. Journal of Safety Research, 35, 81−90.
Diaz, R. I., & Cabrera, D. (1997). Safety climate and attitude as evaluation
measures of organizational safety. Accident Analysis and Prevention, 29,
643−650.
Donald, I., & Canter, D. (1994). Employee attitudes and safety in the
Chemical Industry. Journal of Loss Prevention in the Process
Industries, 7, 203−208.
Donald, I., & Young, S. (1996). Managing safety: an Attitudinal-based
approach to improving safety in organizations. Leadership & Organi-
zational Journal, 17, 13−20.
Eiff, G. (1999). Organizational Safety Culture. Proceedings of the Tenth
International Symposium on Aviation Psychology, Columbus.
Farrington-Darby, T., Pickup, L., & Wilson, J. R. (2005). Safety culture in
railway maintenance. Safety Science, 43, 39−60.
Fuller, C. (2000). Modeling continuous improvement and benchmarking
processes through the use of benefit curves. Benchmarking: An
International Journal, 7, 35−51.
Geller, E. S. (1994, September). Ten principles for achieving a Total Safety
Culture. Professional Safety, 18−24.
Glendon, A. I., & Litherland, D. K. (2001). Safety climate factors, group
differences and safety behavior in road construction. Safety Science, 39,
157−188.
Glendon, A. I., & Stanton, N. A. (2000). Perspectives on Safety Culture.
Safety Science, 34, 193−214.
Glendon, A. I., Stanton, N. A., & Harrison, D. (1994). Factor analyzing a
performance shaping concepts questionnaire. In S. A. Robertson (Ed.),
Contemporary Ergonomics 1994: Ergonomics for All London: Taylor &
Francis.
Grote, G., & Künzler, C. (2000). Diagnosis of safety culture in safety
management audits. Safety Science, 34, 131−150.
Guldenmund, F. W. (2000). The nature of safety culture: A review of theory
and research. Safety Science, 34, 215−257.
Hair, J., Anderson, R., Tatham, R., & Black, W. (1998). Multivariate Data
Analysis. Englewood Cliffs, NJ: Prentice-Hall.
Hale, A. R., Heming, B. H. J., Carthey, J., & Kirwan, B. (1997). Modeling of
Safety Management Systems. Safety Science, 26, 121−140.
Havold, J. I. (2005). Safety-culture in a Norwegian shipping company.
Journal of Safety Research, 36, 441−458.
Hayes, B. E., Perander, J., Smecko, T., & Trask, J. (1998). Measuring
perceptions of workplace safety: Development and validation of a Work
Safety Scale. Journal of Safety Research, 29, 145−161.
Hofmann, D. A., Jacobs, R., & Landy, F. (1995). High reliability process
industries: Individual micro and macroorganizational influences on
safety performance. Journal of Safety Research, 26, 131−149.
Hofmann, D. A., & Stetzer, A. (1996). A cross-level investigation of factors
influencing unsafe behaviors and accidents personnel. Psychology, 49,
307−339.
HSC. (1993). ACSNI Study Group on Human Factors. 3rd Report:
Organising for Safety London: Health and Safety Commission,
HMSO.
HSE. (1997). Successful Health and Safety Management, HSG65. London:
HSE Books.
 B. Fernández-Muñiz et al. / Journal of Safety Research 38 (2007) 627–641
Huang, Y. H., Ho, M., Smith, G. S., & Chen, P. Y. (2006). Safety climate and
self-reported injury: Assessing the mediating role of employee safety
control. Accident Analysis and Prevention, 38, 425−433.
Hurst, N. (1997). From research to practical tools – developing assessment
tools for safety management and safety culture. Journal of Loss
Prevention in the Process Industries, 10, 63−66.
Hurst, N. W., Young, S., Donald, I., Gibson, H., & Muyselaar, A. (1996).
Measures of safety management performance and attitudes to safety at
major hazard sites. Journal of Loss Prevention in the Process Industries,
9, 161−172.
IAEA. (1986). Summary Report on the Post Accident Review Meeting on the
Chernobyl Accident.Vienna: Author IAEA 75-INSAG-1.
ILO. (2001). Guidelines on Occupational Safety and Health Management
Systems. Ginebra: Author ILO-OSH 2001.
Kirwan, B. (1998). Safety Management Assessment and Task Analysis: a
Missing Link. In A. Hale, & M. Baram (Eds.), Safety Management: The
Challenge of Change (pp. 67−92). Oxford: Elsevier.
Kletz, T. A. (1985). An Engineer's View of Human Error. Warwickshire,
England: Institution of Chemical Engineers.
Laurence, D. (2005). Safety rules and regulations on mine sites — The
problem and a solution. Journal of Safety Research, 36, 39−50.
Lee, T. R. (1996, October). Perceptions, attitudes and behavior: the vital
elements of a safety culture. Health and Safety, 1−15.
Lee, T., & Harrison, K. (2000). Assessing safety culture in nuclear power
stations. Safety Science, 34, 61−97.
Lu, C., & Shang, K. (2005). An empirical investigation of safety climate in
container terminal operators. Journal of Safety Research, 36, 297−308.
McDonald, N., Corrigan, S., Daly, C., & Cromie, S. (2000). Safety manage-
ment systems and safety culture in aircraft maintenance organisations.
Safety Science, 34, 151−176.
Mearns, K., Whitaker, S. M., & Flin, R. (2003). Safety climate, safety
management practice and safety performance in offshore environments.
Safety Science, 41, 641−680.
Melià, J. L. (2004). The psychosocial causal model of occupational
accidents of Valencia University: Prospects and new developments. 3rd
International Conference on Occupational Risk Prevention, Santiago de
Compostela, Spain.
Melià, J. L., & Sesé, A. (1999). La medida del clima de seguridad y salud
laboral. Anales de Psicología, 15, 269−289.
Michael, J. H., Evans, D. D., Jansen, K. J., & Haight, J. M. (2005).
Management commitment to safety as organizational support: Relation-
ships with non-safety outcomes in wood manufacturing employees.
Journal of Safety Research, 36, 171−179.
Mitchison, N., & Papadakis, G. A. (1999). Safety Management System
under Seveso II: Implementation and Assessment. Journal of Loss
Prevention in the Process Industries, 12, 43−51.
National Safety Council. (2001). Accident prevention manual for business
and industry: Administration and programs, 12th ed. Itasca, IL: Author.
Niskanen, T. (1994). Safety Climate in the Road Administration. Safety
Science, 17, 237−255.
Nunnally, J. C. (1978). Psychometric Theory. New York: McGraw-Hill.
O'Toole, M. (2002). The Relationship between Employees' Perceptions of
Safety and Organizational Culture. Journal of Safety Research, 33,
231−243.
Olive, C., O'Connor, M., & Mannan, M. S. (2006). Relationship of safety
culture and process safety. Journal of Hazardous Materials, 130,
133−140.
Ordiz, M., & Fernández, E. (2003). High-involvement practices in human
resource management: concept and factors that motivate their adoption.
International Journal of Human Resource Management, 14, 511−529.
Ordoñez, P. (2002). Knowledge management and organizational learning:
typologies of knowledge strategies in the Spanish manufacturing industry
from 1995 to 1999. Journal of Knowledge Management, 6, 52−62.
Ostrom, L., Wilhelmsen, C., & Daplan, B. (1993). Assessing safety culture.
Nuclear Safety, 34, 163−172.
Petersen, D. (2000, January). Safety Management 2000. Our Strengths &
Weaknesses. Professional Safety, 16−19.
639
Pidgeon, N. F. (1991). Safety Culture and Risk Management in Organiza-
tions. Journal of Cross-Cultural Psychology, 22, 129−141.
Prieto, I., & Revilla, E. (2004). Information technologies and human behaviors
as interacting knowledge management enablers of the organizational
learning capacity. In proceedings of the Conference on Organizing
Knowledge, Learning and Capabilities, Innsbruck, Austria, 2–3, April.
Reason, J. (1997). Managing the Risk of Organizational Accidents.
Aldershot, Hants: Ashagate Publishing Ltd.
Reason, J. (1998). Achieving a safe culture: Theory and practice. Work and
Stress, 12, 293−306.
Richter, A., & Koch, A. (2004). Integration, differentiation and ambiguity in
safety cultures. Safety Science, 42, 703−722.
Rundmo, T. (1996). Associations between risk perception and safety. Safety
Science, 24, 197−209.
Rundmo, T., & Hale, A. (2003). Managers' attitudes towards safety and
accident prevention. Safety Science, 41, 557−574.
Safety Research Unit. (1993). The Contribution of Attitudinal and
Management Factor to Risk in the Chemical Industry (Final Report to
the Health and Safety Executive). Guilford: Psychology Department
University of Surrey.
Santos-Reyes, J., & Beard, A. L. (2002). Assessing safety management
systems. Journal of Loss Prevention in the Process Industries, 15, 77−95.
Seo, D. C. (2005). An explicative model of unsafe work behavior. Journal of
Safety Research, 43, 187−211.
Seo, D. -C., Torabi, M. R., Blair, E. H., & Ellis, N. T. (2004). A cross-
validation of safety climate scale using confirmatory factor analytic
approach. Journal of Safety Research, 35, 427−445.
Shafai-Sahrai, Y. (1971). An Inquiry into factors that might explain
differences in occupational accident experience of similar size firms in
the same industry. East Lansing, Michigan: Division of Research,
Graduate School of Business Administration, Michigan State University.
Shannon, H. S., Mayr, J., & Haines, T. (1997). Overview of the relationship
between organizational and workplace factors and injury rates. Safety
Science, 26, 201−217.
Silva, S., Lima, M. L., & Baptista, C. (2004). OSCI: an Organizational and
Safety Climate Inventory. Safety Science, 42, 205−220.
Simonds, R. H., & Shafari-Sahrai, Y. (1977). Factors Apparently Affecting
Injury Frequency in Eleven Matched Pairs of Companies. Journal of
Safety Research, 9, 120−127.
Siu, O. L., Phillips, D., & Leung, T. W. (2004). Safety climate and safety
performance among construction workers in Hong Kong: The role of
psychological strains as mediators. Accident Analysis and Prevention,
36, 359−366.
Smith, M. J., Cohen, A., Cohen, H. H., & Cleveland, R. S. (1978).
Characteristics of Successful Safety Programs. Journal of Safety
Research, 10, 5−15.
Sorensen, J. N. (2002). Safety culture: a survey of the state-of-the-art.
Reliability Engineering & System Safety, 76, 189−204.
Steenkamp, J. B., & Van Trijp, H. C. M. (1991). The use of LISREL in
validating marketing constructs. International Journal of Research in
Marketing, 8, 283−289.
Tam, C. M., Zeng, S. X., & Deng, Z. M. (2004). Identifying elements of poor
construction safety management in China. Safety Science, 42, 569−586.
Vecchio-Sadus, A. M., & Griffiths, S. (2004). Marketing strategies for
enhancing safety culture. Safety Science, 42, 601−619.
Vredenburgh, A. (2002). Organizational safety: Which management
practices are most effective in reducing employee injury rates? Journal
of Safety Research, 33, 259−276.
Wallace, A., & Neal, A. (2000). A Report on Safety in the Queen Island
Meat Industry. Report prepared for the Meat Industry Advisory Group
and The Australian Meat Industry Employees Union.
Wiegmann, D., Zhang, H., Von Thaden, T., Sharma, G., & Mitchell, A. (2002).
A Synthesis of Safety Culture and Safety Climate Research. Prepared for:
Federal Aviation Administration Atlantic City International Airport, NJ.
Williamson, A. M., Feyer, A. M., Cairns, D., & Biancotti, D. (1997). The
development of a measure of safety climate: The role of safety
perceptions and attitudes. Safety Science, 25, 15−27.
640 B. Fernández-Muñiz et al. / Journal of Safety Research 38 (2007) 627–641

Wu, T. -C., & Lee, J. -C. (2003). Developing a safety climate scale in Planning3 Prevention plans formulated setting measures to take on
laboratories in universities and colleges. Journal of Occupational Safety basis of information provided by evaluation of risks in all job
and Health, 11, 19−34. positions.
Zohar, D. (1980). Safety climate in industrial organizations: Theoretical and Planning4 Prevention plans clearly specify person responsible for carrying
applied implications. Journal of Applied Psychology, 65, 95−102. out action.
Zohar, D. (2000). A group-level model of safety climate: Testing the effect Planning5 Standards of action or work procedures elaborated on basis of
of group climate on microaccidents in manufacturing jobs. Journal of risk evaluation.
Applied Psychology, 85, 587−596. Planning6 Prevention plans circulated among all workers.
Planning7 Prevention plans periodically reviewed and updated when job
conditions modified or workers' health damaged.
Appendix A. Measurement Scales of Study Variables EMERGENCY PLANNING
Planning8 Firm has elaborated emergency plan for serious risks or catastrophes.
SAFETY POLICY Planning9 Firm has implemented its emergency plan.
Pol1 Firm coordinates its health and safety policies with other HR Planning10 All workers informed about emergency plan.
policies to ensure commitment and well-being of workers. Planning11 Periodic simulations carried out to check efficacy of emergency
Pol2 Written declaration is available to all workers reflecting plan.
management's concern for safety, principles of action and INTERNAL CONTROL
objectives to achieve. Contro1 Periodic checks conducted on execution of prevention plans
Pol3 Management has established in writing the functions of and compliance level of regulations.
commitment and participation and the responsibilities in safety Contro2 Standards or pre-determined plans and actions are compared,
questions for all organization members. evaluating implementation and efficacy in order to identify
Pol4 Safety policy contains commitment to continuous improvement, corrective action.
attempting to improve objectives already achieved. Contro3 Procedures in place (reports, periodic statistics) to check
EMPLOYEES' INCENTIVES achievement of objectives allocated to managers.
Incent1 Incentives frequently offered to workers to put in practice Contro4 Systematic inspections conducted periodically to ensure
principles and procedures of action (e.g., correct use of protective effective functioning of whole system.
equipment). Contro5 Accidents and incidents reported, investigated, analyzed and
Incent2 Modifications of production processes or changes in jobs recorded.
consulted directly with workers affected or their representatives. BENCHMARKING TECHNIQUES
Incent3 Resolutions frequently adopted that originated from consultations Contro6 Firm's accident rates regularly compared with those of other
with or suggestions from workers. organizations from same sector using similar production
Incent4 Meetings periodically held between managers and workers to processes.
take decisions affecting organization of work. Contro7 Firm's techniques and management practices regularly compared
Incent5 Frequent use of teams made up of workers from different parts with those of other organizations from all sectors, to obtain new
of organization to resolve specific problems relating to working ideas about management of similar problems.
conditions. MANAGERS' ATTITUDES
TRAINING IN OCCUPATIONAL HAZARDS Commit1 Managers consider that employees' participation, commitment
Train1 Worker given sufficient training period when entering firm, and involvement is fundamental to health and safety activities
changing jobs or using new technique. in order to reduce the work accident rate.
Train2 There is follow-up of training needs and of efficacy or repercussion Commit2 Managers consider training of employees is essential for
of training previously given. achieving a safe workplace.
Train3 Training actions continuous and periodic, integrated in formally Commit3 Managers consider internal communication is essential to
established training plan. understand and implement safety policy.
Train4 Training plans elaborated taking into account firm's Commit4 Managers consider that it is fundamental to monitor activities
particular characteristics. in order to maintain and improve safety activities.
Train5 Specific training plans elaborated according to section or job MANAGERS' BEHAVIOR
position. Commit5 Firm managers take responsibility for health and safety as well
Train6 Training plan decided jointly with workers or their representatives. as quality and productivity.
Train7 Training actions carried out during working day. Commit6 Managers actively and visibly lead in safety matters.
Train8 Firm helps workers to train in-house (leave, grants). Commit7 Safety is a work requirement and a condition of contracting.
Train9 Instruction manuals or work procedures elaborated to aid in Commit8 Managers regularly visit workplace to check work conditions
preventive action. or to communicate with employees.
COMMUNICATION IN PREVENTION MATTERS Commit9 Managers encourage meetings with employees and directors to
Commun1 There is a fluent communication embodied in periodic and discuss safety matters.
frequent meetings, campaigns or oral presentations to transmit EMPLOYEES' INVOLVEMENT
principles and rules of action. Involv1 Employees involved in creating guidelines for procedures and
Commun2 Information systems made available to affected workers prior instruction manuals.
to modifications and changes in production processes, job Involv2 Employees participate actively in devising, executing and
positions or expected investments. monitoring safety plans.
Commun3 When starting in new job position worker provided written Involv3 Employees comply with safety regulations.
information about procedures and correct way of doing tasks. Involv4 Employees provide written suggestions in event of any
Commun4 Written circulars elaborated and meetings organized to inform deficiencies in working conditions.
workers about risks associated with their work and how to SAFETY PERFORMANCE
prevent accidents. Safety1 Personal injuries
PREVENTIVE PLANNING Safety2 Material damage
Planning1 Firm has systems to identify risks in all job positions. Safety3 Employees' motivation
Planning2 Systems in place to evaluate risks detected in all job positions. Safety4 Absenteeism / Lost time
 B. Fernández-Muñiz et al. / Journal of Safety Research 38 (2007) 627–641
Beatriz Fernández-Muñiz is Assistant Professor of strategic management
at the University of Oviedo. She has been teaching in Technical Schools
and Faculties of Economics and Management since 1999. She received her
Ph.D. in business management in 2004 from the University of Oviedo. Her
research activity has centered on the study of human resource management,
corporate strategy, competitive advantage, organizational culture, safety
management and safety culture.
José M. Montes-Peón has been Associate Professor at the University of
Oviedo since 1999. He was awarded his degree in 1992. He obtained his
Ph.D. in Business Economics in 1996. He has been Vice Dean for
Academic Affairs at the Faculty of Economics, University of Oviedo. His
main current interests include human resource management, strategy and
organization, knowledge management, safety management and safety
culture. He has published papers on innovation and intangible assets in
refereed journals such as International Journal of Technology Management,
Technovation, International Journal of Biotechnology, and Journal of
Knowledge Management.
641
Camilo J. Vázquez-Ordás is Professor of Business Organization at the
University of Oviedo, where he has worked since 1987. His research
activity has centered on the study of the firm's operations management,
production strategy, safety management and safety culture. He was awarded
his degree in 1987. He obtained his Ph.D. in Business Economics in 1992.
He has been Sub-director of the University Institute of the Firm and
Director of Teaching Staff and Postgraduate Courses at the University of
Oviedo, and he is now Head of the Department of Business Administration
and Accountancy. He has published papers in refereed journals such as
International Journal of Production Research, Omega International
Journal of Management Science, European Journal of Innovation
Management, The Journal of Entrepreneurship, International Journal of
Technology Management and Technovation, among others.