You are on page 1of 13

International Journal of Occupational Safety and

Ergonomics

ISSN: 1080-3548 (Print) 2376-9130 (Online) Journal homepage: https://www.tandfonline.com/loi/tose20

The relationships between OHS prevention costs,


safety performance, employee satisfaction and
accident costs

Metin Bayram, Mustafa C. Ünğan & Kadir Ardıç

To cite this article: Metin Bayram, Mustafa C. Ünğan & Kadir Ardıç (2017) The relationships
between OHS prevention costs, safety performance, employee satisfaction and accident
costs, International Journal of Occupational Safety and Ergonomics, 23:2, 285-296, DOI:
10.1080/10803548.2016.1226607

To link to this article: https://doi.org/10.1080/10803548.2016.1226607

Published online: 26 Sep 2016.

Submit your article to this journal

Article views: 1206

View related articles

View Crossmark data

Citing articles: 14 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=tose20
International Journal of Occupational Safety and Ergonomics (JOSE), 2017
Vol. 23, No. 2, 285–296, http://dx.doi.org/10.1080/10803548.2016.1226607

The relationships between OHS prevention costs, safety performance, employee satisfaction and
accident costs
Metin Bayram, Mustafa C. Ünğan ∗ and Kadir Ardıç
Sakarya University, Turkey

Little is known about the costs of safety. A literature review conducted for this study indicates there is a lack of survey-
based research dealing with the effects of occupational health and safety (OHS) prevention costs. To close this gap in the
literature, this study investigates the interwoven relationships between OHS prevention costs, employee satisfaction, OHS
performance and accident costs. Data were collected from 159 OHS management system 18001-certified firms operating
in Turkey and analyzed through structural equation modeling. The findings indicate that OHS prevention costs have a
significant positive effect on safety performance, employee satisfaction and accident costs savings; employee satisfaction
has a significant positive effect on accident costs savings; and occupational safety performance has a significant positive
effect on employee satisfaction and accident costs savings. Also, the results indicate that safety performance and employee
satisfaction leverage the relationship between prevention costs and accident costs.
Keywords: OHS costs; OHSAS 18001; employee satisfaction; safety performance

1. Introduction is an important cost element for a company.[4] How-


In today’s modern society, occupational˛e accidents cost ever, prevention costs have a potential to prevent occu-
much.[1] There are many more occupational accidents pational accident injuries, decrease employer’s costs and
involving death in developing countries in comparison increase employer’s profits.[5] A number of authors and
with developed countries. This is because developed coun- organizations state that accident costs resulting from not
tries attach much more importance to occupational health taking OHS measures are much higher than OHS preven-
and safety management (OHSM) practices based on a pre- tion costs, and economic reasons justify spending on the
ventive (proactive) approach. Comparing the occupational prevention of occupational accidents.[6–8]
accident data of Turkey with those of developed coun- Since they have a positive effect on the performance
tries, Ceylan [2] reports that Turkey ranks first in terms of of companies, the implementation of OHS MSs is not an
deaths. economic burden, but an opportunity for companies. As
Standard No. OHSAS 18001:2007 [3] is the most com- a matter of fact, the countries with the highest occupa-
mon standard across the world for providing a structural tional safety are the countries with the highest competitive
approach to occupational health and safety management power.[9] It is realized more and more that providing
systems (OHS MSs), and organizations which wish to safer and healthier workplaces increases labor produc-
have an OHS MS that conforms to this standard can offi- tivity and job satisfaction and yields stronger balance
cially be certified by certification bodies. Standard No. sheets.[10–12] Like the total quality management (TQM)
OHSAS 18001:2007, which is an occupational health and system, OHS MSs are regarded as an element assur-
safety (OHS) standard, is widely accepted throughout the ing employee satisfaction.[13] One of the reasons for
world. Companies from a wide range of sectors and having the increased importance of employee satisfaction for
different sizes implement this standard. establishments is the close association between this con-
The total cost of occupational accidents consists of cept and employees’ health and productivity in work.[14]
the expenses made to prevent occupational accidents, all Occupational safety may remove a lot of adverse effects
kinds of expenses made after the occurrence of occupa- (e.g., rising economic costs or unproductive working
tional accidents and the losses these accidents cause. The caused by increased work stoppage, absenteeism and
loss of human resources that occurs when employees do turnover rate due to employees who are worn out in
not take part in production processes permanently or tem- terms of health and social relations).[15] When non-wage
porarily because of accidents taking place in a workplace factors are considered, it is seen that job dissatisfaction

*Corresponding author. Email: ungan@sakarya.edu.tr

© 2016 Central Institute for Labour Protection – National Research Institute (CIOP-PIB)
286 M. Bayram et al.

results in labor turnover, absenteeism and occupational 2.1.1. OHS prevention costs
accidents.[16] Failure costs decrease as OHS prevention costs increase
Although a limited number of case studies and anec- because more expenses are made to eliminate risks.[19]
dotal evidence suggest that prevention costs do have a For Van den Broek et al.,[20] investing in OHS has pos-
positive impact on safety performance, employee satisfac- itive effects. Similarly, the American Society of Safety
tion and accidents costs, Fernández-Muñiz et al. [17] in a Engineers (ASSE) recommends enterprises to invest in
study conducted in Spain observe that employers consider OHS as part of their business strategies to reduce injuries
prevention costs unnecessary and think that they do not and occupational diseases, increase profitability and cre-
contribute to the production objectives of companies and ate new lines of business.[7] Research and statistics show
thus have negative effects on profitability and competitive that investment made in OHS and prevention costs yields a
power. Thus, the aim of the present study is to empiri- profit of approximately 1:1–1:10 in return.[7,9] Research
cally prove whether prevention cost is an extra burden or on the construction sector in England reports that the
not on companies by analyzing the interwoven relation- total benefit gained by preventing occupational accidents
ships between OHS prevention costs, employee satisfac- is three times as much as the expenses incurred by
tion, OHS performance and accident costs. The research accidents.[6]
findings may encourage those managers who are reluctant Zou et al. [21] listed the benefits of investing in occu-
about investing in OHS. pational safety as cost saving as a result of a decrease in
The most important contributions of this study to the the number of accidents, fewer interruptions in produc-
literature are twofold. First, new scales were developed tion and increased productivity. According to the European
for accident costs and prevention costs. Second, it is the Agency for Safety and Health at Work,[22] the benefits to
first survey-based study dealing with the effects of pre- be gained through investment in health and safety fall into
vention costs on safety performance, employee satisfaction two groups: direct and indirect. Among direct benefits are
and accident costs. reduced insurance premiums, reduced sue charges, reduced
The article proceeds as follows. In the literature sick pay costs, improved production and productivity rates,
review section, we present the previous studies high- reduced product and material damages, and reduced acci-
lighting prevention cost, accident cost, safety perfor- dent costs and production lags. Some indirect benefits, on
mance and employee satisfaction. This is followed by the the other hand, are reduced absenteeism, reduced person-
research model and hypothesis development section, where nel, improved corporate image, convenience in winning
we develop related arguments concerning the conceptual tender, and improved job satisfaction and morale. This
model comprising the interrelationships between preven- finding clearly shows that OHS prevention costs do have
tion cost, accident cost, safety performance and employee a positive influence on safety performance, accident cost
satisfaction. The research methodology is then introduced savings and employee satisfaction.
with a detailed explanation of the measures and sample. Van den Broek et al. [12] mention that poor OHS per-
The results are then presented with a thorough descrip- formance is a factor motivating company management to
tion of the empirical analysis, including explanatory factor invest in OHS. Lin and Mills [23] indicate that company
analysis (EFA) and the measurement and structural mod- size and thus financial power are important factors influ-
els, using structural equation modeling (SEM). Finally, a ential on OHS performance, and that OHS (prevention)
discussion and implications are presented along with lim- costs play an important part in decreasing accidents and
itations; the article concludes with suggestions for future increasing productivity. Noweir et al. [24] found out that
research. small and medium size enterprises (SMEs) display poor
OHS performance because of inadequate financial support,
management and occupational safety knowledge and cul-
2. Literature review
ture. Similarly, Shalini [8] determined that the enterprises
2.1. OHS costs which have the highest accident rates are the small enter-
The standard mathematical model for OHS cost is as fol- prises which do not allocate enough financial resource to
lows: safety cost (SC) = prevention cost (PC) + accident OHS. López-Alonso et al. [25] carried out research on 40
cost (AC).[18] Although some studies have been con- building sites in the construction sector and found that the
ducted to determine accident costs, Panopoulos [18] average number of accidents is inversely proportional to
observed that there are few studies which adopt a system- accident prevention costs and is directly proportional to the
atic approach to prevention costs. Panopoulos [18] incor- total number of employees, the average number of subcon-
porated costs related to administrative fees, criminal fines, tractors and the OHS budget. Many other authors report
lawsuit processes, loss of corporate image and administra- that investing in occupational safety has positive effects on
tive failure into the aforementioned formula. Behm et al. OHS performance and accident rates.[17,23,26–29] All of
[19] developed a cost model that measures and assesses these findings imply that occupational safety should not
trends in total occupational safety costs by time. be regarded as a financial burden, but rather it improves
International Journal of Occupational Safety and Ergonomics (JOSE) 287

the profitability of a business by reducing accident-related performance.[35] However, the occupational safety liter-
costs. ature contains few studies that deal with the effects of
OHSM practices associated with employee satisfaction
on occupational accidents and diseases.[36] Fernández-
2.1.2. Accident costs
Muñiz et al. [4] empirically showed that employee sat-
In general, accident costs are classified as economic vs. isfaction has a positive effect on a company’s competi-
non-economic, insured vs. non-insured, direct vs. indi- tiveness. Gupta and Upadhyay [37] indicated that taking
rect, fixed vs. variable, internal vs. external and as illness safety measures has a positive effect on employee satis-
costs.[18,30] Economic costs resulting from occupational faction. Omusulah [38] concluded that there is a strong
accidents may include not only direct costs such as finan- relationship between OHS practices and job satisfaction.
cial damage, lost time, financial losses arising from rising Nahavandi and Malekzadeh [39] report that lack of occu-
insurance premiums and health expenses, but also indi- pational safety and bad working conditions lead to stress
rect (hidden) costs such as overtime required by acci- which brings about certain organizational results such as
dents, retraining costs, loss of prestige, worsening indus- increased absenteeism, rise in labor turnover rate, fall in
trial relations, decrease in production quality and fall in performance, decrease in productivity and quality, and
profitability.[8] increased complaints and uneasiness among employees.
The research conducted by Austria’s Allgemeine According to Torp and Moen,[40] employees have
Unfallversicherungsanstalt (Social Insurance for Occupa- higher satisfaction and participate more in OHS activities
tional Risks) (AUVA) in 2010 reveals that the occupational when improvements are introduced. According to Graw-
accidents taking place in the construction sector bring a itch et al.,[41] healthy workplace practices are effective
direct and indirect cost of EUR 1.6 billion to the Austrian in predicting emotional commitment, emotional exhaus-
economy.[9] It is estimated that indirect costs are several tion and employee welfare, which reflects employee sat-
times as much as direct costs (sometimes 20 times as much isfaction.
as direct costs) at enterprise level.[8,9] Grant et al. [42] lay emphasis on a series of important
UK Health and Safety Executive (HSE) statistics methods for increasing the welfare of employees and orga-
[31,32] reveal the human and financial costs resulting nizations. Öğer [14] determined that employee satisfaction
from unsuccessful OHS practices in the United King- level is influential on employee performance, absenteeism,
dom. According to Miller,[5] US employers have to spend behaviors harming the enterprise and employee health.
approximately USD 200 billion for occupational accident Kanoğlu [16] empirically detected that occupational safety,
injuries every year – that means over USD 1700 per which is a non-wage factor, has a higher influence on job
employee; and injury costs are mostly preventable. The satisfaction in comparison with wage.
cost of injuries and occupational diseases mostly resulting Wachter and Patrick [27] empirically showed that there
from the existing working conditions to the English com- is an evident negative relationship between accident rates
munity in the 2012/13 period is estimated to be GBP 14.2 and the emotional and cognitive participation levels of
billion.[33] It is estimated that total cost of occupational occupational safety-oriented employees.
accidents and diseases in Turkey, which is a developing
country, was nearly USD 28 billion in 2010.[34] Shalini [8]
measured the economic cost of the occupational accidents 2.3. Occupational safety performance
taking place in Mauritius, which is a small island country, Previous studies have not reached any consensus on
using quantitative and qualitative approaches, and found it the components of occupational safety performance.
to be MUR 168 million. Researchers use different criteria in measuring OHS perfor-
mance. Among the most widely used criteria are accident
rates and fatal accident rates.[28] Performance indica-
2.2. Employee satisfaction tors of this sort are called lagging indicators.[43] It is
Besides investing in an occupational safety management stated that traditional indicators such as accident statis-
system, organizations should also win the minds and hearts tics indices do not precisely represent companies’ safety
of their employees to prevent and reduce occupational acci- performance.[44] Factors based on accident severity rates
dents and improve occupational safety performance.[27] are effective indicators for the management of accidents
Increased employee satisfaction means improved produc- resulting in employee injury, but they are ineffective for
tion quality, productivity, profitability, competition, qual- bringing major hazard risks under control.[45] Another
ified staff and quality of life.[15] When employee satis- group of indicators consists of positive or leading perfor-
faction is low, health problems become widespread, which mance indicators. Being an important feedback tool, these
brings about a serious cost burden for establishments and indicators help to identify the existing defects and deficien-
governments.[14] cies before they cause any event, accident or injury and
Employee satisfaction has widely been focused allow organizations to correct such deficiencies in their
on because of its important effect on organizational OHSM processes.[46] For Fernández-Muñiz et al.,[17]
288 M. Bayram et al.

the near-miss rate is an important numerical indicator of H 2 : The higher the OHS prevention costs, the higher the
occupational safety consciousness in an industry. OHS performance.
Agwu [29] concluded that when employees have a H 3 : The higher the OHS prevention costs, the higher the
better occupational safety performance, accidents, liabil- employee satisfaction.
ities, financial damages, statutory payments and medical H 4 : The higher the OHS performance, the higher the
costs decrease; and when a company has a better occupa- savings in accident costs.
tional safety performance, productivity, profitability, effi- H 5 : The higher the employee satisfaction, the higher the
ciency, quality, corporate image and innovative capacity savings in accident costs.
are improved. Fitzgerald [47] introduced an occupational H 6 : The higher the OHS performance, the higher the
safety climate evaluation approach for allowing an organi- employee satisfaction.
zation to make cost-effective evaluation. The relationship
between employee satisfaction and safety performance
can be taken two ways. For example, Fernández-Muñiz 4. Method
et al. [4] found out that safety performance has a positive
4.1. Data collection and sampling
effect on employee satisfaction. Some other researchers
– Gyekye and Salminen,[48] Gyekye [49] and Rundmo There are many steps in the methodology, as presented
[50] – determined that employee satisfaction has a pos- in Table 1. A questionnaire approach was used to vali-
itive effect on occupational accidents. In our study, the date the research model. The questionnaire was created
impact of safety performance on employee satisfaction was following an intense literature review. Opinions were then
investigated. There are two reasons for this. We were more taken from the experts of the Ministry of Labor and
interested in the effect of safety performance on employee Social Security, the industry and three certification bodies
satisfaction since we would like to draw the attention of granting OHSMS 18001 certification.[3] With their rec-
employers to a safer workplace for satisfied employees. ommendations, a draft survey form was generated. This
The second reason is that in the literature the effect of draft was completed by 12 authorized managers. Both
satisfaction on safety performance is studied more com- experts and managers were chosen because they are the
pared with vice versa. We wanted to contribute to the most knowledgeable about safety performance, cost issues
less studied and more recent area of the research. Apart and employee satisfaction. Based on their feedback, the
from those already mentioned, there are various studies on questionnaire was finalized. After developing the ques-
OHS performance in the literature, including: Arezes and tionnaire, we adopted the parallel-translation method in
Miguel [44] who summarize the occupational safety per- developing the questionnaire in Turkish. Items were first
formance measurement process and analyze the potential translated from English into Turkish by one person and
role of safety culture in this process; Idoro [28] who com- then back-translated into English by a second person. The
pares the OHS performances and administrative efforts of two translators then jointly reconciled all differences.
Nigerian building contractors; and Cagno et al. [51] who The target group for data collection was companies
suggest a systematic and intervention-oriented model of certified with Standard No. OHSAS 18001:2007.[3] The
safety performance designed for SMEs. reason for including OHSAS 18001:2007-certified firms
was that only OHSAS 18001:2007 was granted to the com-
panies requesting OHS MS certification in Turkey. The
3. Research model and hypothesis survey form was sent to 570 OHS MS-certified firms. A
total of 162 survey forms were filled in and sent back by
Based on the literature review, a research model (Figure 1)
companies, of which 159 survey forms were subjected to
and the following hypotheses were developed:
analysis because three of the survey forms were unusable.
The response rate was 28% (159/570) which is superior
H 1 : The higher the OHS prevention costs, the higher the to those usually obtained in survey-based studies. We fol-
savings in accident costs. lowed Armstrong and Overton [52] to test non-response
bias. The adopted method tests for significant differences
between early respondents (i.e., those responses received
before the first follow-up letter) and late respondents (i.e.,
those received after the first follow-up letter). Concur-
rently, the first 20 surveys received were compared with the
last 20 received and the t test using randomly selected mea-
sures was conducted, resulting in no significant difference
between the two sets of responses.
All types of manufacturing and service companies were
included in the survey. The questionnaire was mailed to the
Figure 1. Research model. general manager or managing director of each organization
International Journal of Occupational Safety and Ergonomics (JOSE) 289

Table 1. Methodological steps.

Step Activity

1 Questionnaire was developed to test the hypothesized relationships among the constructs.
2 Pilot study was conducted with practitioners to make sure that the content is valid.
3 Target companies for the data collection were determined.
4 Data were collected.
5 Factorability of data and sampling adequacy for factor analysis were assessed.
6 Explanatory factor analysis (EFA) conducted to assess whether the items for each construct shared one underlying factor.
7 Confirmatory factor analysis (CFA) was conducted to assess the convergent validity of the measurement model. Convergent
validity was used to show that measures which should be related are in reality related.
8 CFA was conducted to assess the discriminant validity of the measurement model. Discriminant validity was used to show that
measures which should not be related are in reality not related.
9 Quality of the structural model was assessed to determine the predictive relevance of the model.
10 Explanatory power of the structural model was assessed.
12 Goodness of fit of the structural model was assessed to see whether there is a significant difference between observed and expected
correlations among constructs.
13 Multicollinearity between independent variables was assessed to make sure that it does not affect the hypothesis testing results.
14 Hypothesis testing was conducted to see whether the proposed relations between the constructs are supported.
15 Mediating effects and indirect effects between certain constructs were assessed

to ensure respondents’ thorough knowledge of the firms’ had 250–500 employees, and 38% of the firms had >500
safety-related costs, safety performance and employee sat- employees.
isfaction. The covering letter outlined the objectives and
importance of the study. It was requested from the partici-
pating companies that the survey forms would be filled by 4.2. Scale development
the most knowledgeable people about the survey questions. Five scales were developed for this study, two of which are
The majority of respondents filled positions with titles such new in OHS literature. The new scales are OHS preven-
as OHS management representatives, OHS managers and tion costs and accident costs. Participants were requested
OHS experts. The respondents’ titles were concerned with to express their perceptions of change on each scale in the
OHS-related costs, safety practices and employee satisfac- last three years. All scales except for OHS prevention costs
tion; thus, they were likely to pay careful attention to the were measured with a 5-point Likert-type scale: 1 = I
measures used in this study. We concluded that the survey’s strongly disagree, 2 = I disagree, 3 = I neither agree nor
respondents did possess the knowledge required to answer disagree, 4 = I agree and 5 = I strongly agree. For OHS
the questions appropriately. prevention costs, a different 5-point Likert scale was used:
Although the organizations that make up the sam- 1 = no expense, 2 = as little expense as possible, 3 = as
ple represent various manufacturing industries and service much expense as required for fulfilling minimum legal
industries (e.g., transportation, wholesale, financial and requirements, 4 = as much expense as required for com-
health care), the sample is skewed toward manufacturing pletely fulfilling legal requirements and OHS MS require-
firms. Twenty-six percent of the sample is comprised of ments and 5 = as much expense as required for achieving
service firms. The most well-represented industries in the zero accident target besides fulfilling legal requirements
manufacturing sector are fabricated metal (15.1%); rubber and OHS MS requirements.
and miscellaneous products (10%); chemical and pharma- The safety literature contains a couple of studies on
ceutical (8%); energy (8%); and glass and ceramic (9.8%). OHS prevention costs.[18,21] However, there were no
Our cross-industry sample is appropriate; as the service available scales in the survey-based research. The 8-item
industry is included in the category of less hazardous work- scale for OHS prevention cost (PRV) was therefore devel-
places and companies in this industry are less likely to oped based on the aforementioned studies.
have OHS MS certification, there were more companies Another newly developed scale for this study is acci-
participating in the study from the manufacturing sector. dent cost. The 6-item scale for accident cost (ACC) was
Similarly, companies from hazardous and very haz- developed based on the studies of the European Agency
ardous categories that need to attach more importance to for Safety and Health at Work,[53] HSE,[31] Aaltonen
OHS participated more. Approximately 38%, 47% and et al.,[54] Abiuso and De La Figuera,[55] Rikhardsson and
15% of the participants belong to the very dangerous, Impgaard,[1] Lahiri et al. [56] and Battaglia et al.[30]
dangerous and less dangerous hazard class, respectively. The 4-item employee satisfaction (SAT) scale was
Approximately 13% of the firms had <50 employ- developed based on the work of the European Agency for
ees, 15% of the firms employed 50–99 workers, 19% of Safety and Health at Work,[53] Van den Broek et al.,[20]
the firms had 100–249 on the payroll, 15% of the firms Fernández-Muñiz et al. [4,17] and Öğer.[14]
290 M. Bayram et al.

Table 2. Questionnaire items. the significance of the scale factor loadings in the mea-
surement model and that of the path coefficients in the
Prevention costs
structural model.[60]
PRV1: Costs for OHS personnel
PRV2: Costs for the OHS training of employees
PRV3: Technological investment costs
PRV4: Costs for the collective protection of employees 5.1. Exploratory phase
PRV5: Costs for personal protective equipment (PPE) and In the first step, we performed an EFA for each construct in
other occupational safety equipment order to assess whether the items for each latent construct
PRV6: Costs for occupational safety signs and boards
shared one underlying factor. Based on the procedure by
PRV7; Outsourcing costs
PRV8: Costs for fire and emergency management Lumpkin and Dess,[61] we performed an EFA using a vari-
Accident costs max rotation. Before EFA, we performed Bartlett’s test of
ACC1: Treatment costs reduced sphericity [62] to investigate the factorability of the data,
ACC2: Payments to employees reduced and the Kaiser–Meyer–Olkin (KMO) test [63] to measure
ACC3: Administrative costs reduced the sampling adequacy. Results indicated a significant test
ACC4: Costs resulting from labor loss reduced statistic for Bartlett’s test of sphericity, p < 0.000, and a
ACC5: Losses of production reduced
ACC6: Costs resulting from failure in the timely delivery of
KMO value of 0.880, meaning that the data were suitable
orders reduced for structure detection. Items with a factor loading below
Employee satisfaction 0.4 and items strongly loading on more than one factor
SAT1: Employee performance improved (>0.400) were excluded. This procedure finally resulted in
SAT2: Employee absenteeism reduced 4 factors with an eigenvalue >1 which explained 72.08%
SAT3: Employee behaviors harming the enterprise reduced of the total variance.
SAT4: Employees’ physical and mental health statuses The first factor called PRV, with an eigenvalue of 9.09
improved
SAT5:Employee turnover rate improved and accounting for 37.76% of the total variance, included
8 items. The second factor called PER, with an eigenvalue
Safety performance
PER1: Accident frequency rate improved of 4.446 and accounting for 18.61% of the total variance,
PER2: Accident severity rate improved included 5 items. The third factor called SAT, with an
PER3: Accidents involving death and/or loss of limb reduced eigenvalue of 2.084 and accounting for 8.683% of the
PER4: Tangible losses reduced total variance, included 5 items. Finally, the fourth factor
PER5: Near-miss rate improved called ACC, with an eigenvalue of 1.652 and accounting
Note: OHS = occupational health and safety. for 6.883% of the total variance, included 5 items. To test
the common method bias, Harman’s one-factor test was
used.[64] If the measures were to be affected by common
All five of the OHS performance (PER) scale items method bias, then they would have tended to load on a sin-
were taken from occupational accident statistics such as gle factor.[65] The factor analysis resulted instead in four
the accident frequency rate, accident severity rate, number factors, which confirmed that common method bias was
of accidents involving death and/or loss of limb, tangi- not a problem in this research.
ble damages and near misses. The questionnaire items are
presented in Table 2.
5.2. Confirmatory phase
After the EFA the remaining items were entered into
SmartPLS version 3 for assessment of the measurement
5. Results and structural models.
We followed Anderson and Gerbing [57] in examining the
scales with regard to unidimensionality, internal consis-
tency, composite reliability and convergent and discrim- 5.3. Measurement model
inant validity. Results were categorized into two main Confirmatory factor analysis (CFA) was used to assess the
phases: exploratory and confirmatory. For the exploratory convergent and discriminant validity of the measurement
phase, SPSS version 18 was used. For the confirmatory model. Convergent validity of the scales is contingent on
phase, partial least-squares structural equation modeling the fulfillment of three criteria [66,67]: (a) all indicator
(PLS-SEM) was then undertaken using SmartPLS ver- loadings should exceed 0.703; (b) composite reliabilities
sion 3.[58] PLS-SEM was selected because it is able to should exceed 0.8 or alternatively Cronbach’s α should
identify the key driving constructs and deal with a non- exceed 0.65 (α can theoretically take values from 0 to 1)
normal data set, and it has minimum demand for sample [68]; and (c) the average variance extracted (AVE) for each
size.[59] A bootstrapping estimation procedure, in which construct should exceed 0.5. Psychometric properties of
500 random observation samples with replacements were constructs and items are presented in Table 3. As Table 3
generated from the original dataset, was used to examine shows, all of the indicator loadings except for SAT5 are
International Journal of Occupational Safety and Ergonomics (JOSE) 291

Table 3. Psychometric properties of constructs and items.

Variable M SD Factor loading t Cronbach’s α CR AVE

ACC 0.95 0.96 0.80


ACC1 3.79 0.95 0.90 42.61
ACC2 3.73 0.99 0.91 42.88
ACC3 3.65 0.96 0.90 35.78
ACC4 3.72 0.93 0.92 55.73
ACC5 3.75 0.89 0.90 37.83
ACC6 3.56 0.95 0.86 30.99
SAT 0.86 0.91 0.70
SAT1 3.72 0.86 0.81 23.08
SAT3 3.41 1.07 0.80 19.38
SAT4 3.86 0.89 0.84 29.09
SAT2 3.69 0.93 0.84 29.06
SAT5** 3.22 1.07 0.67
PRV 0.93 0.94 0.65
PRV1 4.23 0.81 0.71 11.80
PRV2 4.15 0.85 0.83 26.47
PRV3 4.01 0.89 0.85 32.25
PRV4 4.04 0.93 0.80 19.45
PRV5 4.43 0.81 0.83 23.33
PRV6 4.40 0.75 0.82 22.66
PRV7 4.33 0.79 0.80 19.62
PRV8 4.25 0.82 0.82 24.92
PER 0.93 0.95 0.77
PER1 2.81 1.30 0.93 79.65
PER2 2.87 1.28 0.92 45.07
PER3 3.10 1.64 0.83 23.22
PER4 3.00 1.37 0.91 65.16
PER5 2.64 1.36 0.80 18.91

**Item deleted due to low factor loading.


Note: See Table 2 for the variables. AVE = average variance extracted; CR = composite reliability.

above the recommended threshold. Therefore, SAT5 was addition, the CFA shows that the loading of each indica-
removed from further analysis. After dropping SAT5, CR tor is greater than all of its cross-loadings, which indicates
ranges from 0.91 to 0.96, and AVE ranges from 0.65 to discriminant validity on the indicator level.[69]
0.80. All three conditions for convergent validity thus hold.
To evaluate discriminant validity, Fornell and Larker [66] 5.4. Structural model
suggest that the square root of AVE of a latent variable
Prior to the hypotheses testing, the quality of path model
should be greater that the correlations between the rest of
can also be evaluated by calculating the Q2 statistic.
the latent variables.
The Q2 statistic measures the predictive relevance of
Table 4 presents the correlation of latent variables
the model by reproducing the observed values by the
and the square root of AVE. A comparison of all of the
model itself. Q2 > 0 means the model has predictive rel-
correlations and square roots of AVE on the diagonal indi-
evance; whereas Q2 < 0 means that the modal lacks pre-
cates adequate discriminant validity for all constructs. In
dictive relevance.[70] In PLS two kinds of Q2 statis-
tics are estimated. The cross-validated communality (CV-
Table 4. Correlation of latent variables and square root of communality) measures the capacity of the model to pre-
AVE. dict the manifest variables (MVs) directly from their latent
variables (LVs) by cross-validation. It uses only the mea-
ACC SAT PER PRV surement model. The prediction of an MV of an endoge-
ACC 0.900 nous block is carried out using the MVs of the same
SAT 0.561** 0.795 block. The cross-validated redundancy (CV-redundancy)
PER 0.564** 0.421** 0.881 measures the capacity of the path model to predict the
PRV 0.287** 0.280** 0.178* 0.808 endogenous MVs indirectly from a prediction of their
*p < 0.050, **p < 0.010. own LV using the related structural relation, by cross-
Note: See Table 2 for the variables. AVE = average validation,[71] and serves as a sign of the quality of the
variance extracted. structural model. For this study, both the measurement and
292 M. Bayram et al.

Table 5. Quality of measurement and structural model. of the structural model analysis, showing the path coeffi-
cients along with their significance levels and the R2 values
CV-communality CV-redundancy
of endogenous constructs. Based on the scores of R2 , it can
PRV 0.724 0.346 be interpreted that the model explains 44% of the vari-
ACC 0.497 0.137 ance (R2 ) of the ACC. The R2 value is moderate and if
SAT 0.661 0.024 certain inner path model structures explain an endogenous
PER 0.545 – latent variable by only a few exogenous latent variables,
Note: See Table 2 for the variables. CV = cross-validated. ‘moderate’ R2 may be acceptable.[69]
Henseler et al. [72] introduce the standardized root
mean square residual (SRMR) as a goodness-of-fit mea-
structural models demonstrate a good quality since both sure for Smart PLS. The SRMR is defined as the difference
CV-communality and CV-redundancy values are positive between the observed correlation and the predicted correla-
(Table 5). tion. Thus, it allows assessing the average magnitude of the
Besides the examination of cross-validation CV- discrepancies between observed and expected correlations
communality and CV-redundancy indices, the explana- as an absolute measure of (model) fit criterion. A value less
tory power of the proposed model can also be assessed than 0.10 or of 0.08 – in a more conservative version view;
by observing the coefficient of determination (R2 ) of the see Hu and Bentler [73] – are considered a good fit. For this
endogenous latent variables. Figure 2 displays the results study, the SMRM value of 0.056 indicates that goodness of
the model fit is above the threshold criteria.
No serious problems of multicollinearity exist between
the independent variables because the variance inflation
factors of all items ranged between 1.000 and 1.260,
which are far below the common cut-off threshold range
of 5–10.[74]
The results of the statistical model offer support for all
hypotheses (Figure 2). Detailed results of the hypothesis
testing are presented in Table 6. Although all hypotheses
are supported, we see stronger support for H 4 , H 5 and H 6 .
In addition to the direct effect of prevention costs on
accident costs, mediating effects of safety performance and
employee satisfaction were assessed. We first verified the
Figure 2. Structural model.
*p < 0.050, **p < 0.001. correlations between mediating variables and both PRV

Table 6. Structural properties of the constructs.

Hypothesis Link Path coefficient t p Supported?

H1 PRV-ACC 0.127 2.147 0.032 Yes


H2 PRV-PER 0.178 2.195 0.029 Yes
H3 PRV-SAT 0.189 2.560 0.011 Yes
H4 PER-ACC 0.400 5.835 0.000 Yes
H5 SAT-ACC 0.337 4.005 0.000 Yes
H6 PER-SAT 0.380 5.743 0.000 Yes

Note: See Table 2 for the variables

Table 7. Results for mediating effects.


Root mean Root mean
Independent Dependent Path without square Link with square
variable variable Mediator mediator Coefficient residual mediator Coefficient residual Result

PRV ACC PER PRV-ACC 0.289** 0.061 PRV-ACC 0.126* 0.056 Partial
SAT PRV-PER 0.178*
PRV-SAT 0.259**
PER-ACC 0.401**
SAT-ACC 0.336**

*p < 0.050, **p < 0.001.


Note: See Table 2 for the variables.
International Journal of Occupational Safety and Ergonomics (JOSE) 293

Table 8. Significance of indirect effects. OHS prevention cost was found to have a direct effect
Independent Dependent Indirect
on safety performance. This finding is consistent with the
variable Mediator variable effect findings of López-Alonso et al.[25]
A positive relationship was found between preven-
PRV PER ACC 0.071* tion cost and employee satisfaction. This finding is sup-
PRV SAT ACC 0.081**
ported by the European Agency for Safety and Health at
*p < 0.050, **p < 0.001. Work.[22]
Note: See Table 2 for the variables. OHS performance was found to have a positive effect
on employee satisfaction in the present study. This finding
is consistent with the findings of some other researchers.[4]
and ACC. In this regard, as shown in Table 4, the corre- Employee satisfaction was also seen to have a pos-
lation analysis provides evidence that necessary conditions itive effect on accident costs. This finding is consistent
for the mediation effects apply: both prevention costs and with the findings of Fernández-Muñiz et al. [4] that show
accident costs are correlated with safety performance and the positive effect of employee satisfaction on company
employee satisfaction. competitiveness.
To test the mediation effect, we built two different SEM It was determined in the present study that improve-
models as suggested by Little et al.[75] In our first model, ment in OHS performance leads to decrease in accident
we removed the mediators of PER and SAT from the struc- costs. This finding is consistent with the findings of the
tural model and assessed the significance of the coefficients study carried out by Agwu.[29]
on direct paths from PRV to ACC. As reported in Table 7, Findings of this study prove that, contrary to the com-
both of the path coefficients were significant. Next, we mon belief of managers, there is a good return of invest-
added the mediators to the model and examined the signif- ment on OHS, and it is hoped that the findings of the
icance of several path coefficients to test mediation effects. present study will encourage the employers to invest in
While the magnitudes were decreased, the coefficient on OHS.
the direct path (PRV to ACC) remained significant. The Readers should treat these results with caution, because
path coefficients from PRV to PER, from PRV to SAT, the study was only conducted in Turkey and is the first
from PER to ACC and from SAT to ACC were significant. survey-based research of its kind. It is quite possible that
Taken together, we concluded that the effects of PRV on the findings could differ from one country to another
ACC are partially mediated through PER and SAT. depending on their OHS MS maturity level. In other words,
Additionally, we assessed the indirect effects of PRV spending on OHS MS may not have the same effect on
through SAT and PER on ACC. As reported in Table 8, the safety performance and employee satisfaction in different
indirect effect of PRV on ACC through PER and SAT pres- countries (or companies) with different OHS MS applica-
sure was significant (p < 0.050, p < 0.010 respectively). tion levels since once OHS spending reaches a limit it may
not increase or leads to a marginal increase in safety per-
formance and employee satisfaction. It should be noted that
besides monetary factors, factors such as the motivation of
6. Conclusion
management and their willingness to lead, set measurable
This study aimed to analyze the relationships between OHS targets, provide resources and establishing accountabil-
prevention costs, employee satisfaction, OHS performance ity at all levels are very important determinants of safety
and accident costs. performance and employee satisfaction.
The relationships between OHS prevention costs and Some interesting research avenues emerged as a result
accident costs; between OHS prevention costs and safety of this study. One area of research should test the relative
performance; between prevention costs and employee sat- importance of monetary and non-monetary factors in safety
isfaction; between safety performance and accident costs; performance and satisfaction. It is also worth question-
between safety performance and employee satisfaction; ing whether there is any difference between the attitudes
and between employee satisfaction and accident costs of employers of the economically developed and develop-
were empirically tested. The findings confirm all these ing countries in terms of prevention costs. Future studies
relationships. Another important finding of this study is may also focus on the specific sectors and SMEs with a
that there is an indirect relationship between prevention poor safety performance. In addition, comparisons may be
costs and accident costs. Safety performance and employee made between companies having OHS MS certification
satisfaction partially mediate the relationship between and those not having it.
those two.
A direct positive relationship was found between OHS
prevention costs and decrease in accident costs. This find-
ing is consistent with the findings of Ikpe et al. [6] and Disclosure statement
Falkner et al.[9] No potential conflict of interest was reported by the authors.
294 M. Bayram et al.

References İSTAÇ A.Ş.] [master’s thesis]. İstanbul: Marmara Univer-


[1] Rikhardsson PM, Impgaard M. Corporate cost of occu- sity; 2007. Turkish.
pational accidents: an activity-based analysis. Accid Anal [17] Fernández-Muñiz B, Montes-Peón JM, Vázquez-Ordás CJ.
Prev. 2004;36:173–182. doi:10.1016/S0001-4575(02)00 Relation between occupational safety management and firm
147-1. performance. Saf Sci. 2009;47:980–991. doi:10.1016/j.ssci.
[2] Ceylan H. The general outlook of workplace accidents in 2008.10.022.
Turkey and a comparison with developed countries. Int J [18] Panopoulos GD. Economic aspects of safety in Greek
Eng Res Dev. 2011;3(2):18–24. conctruction industry [doctoral dissertation]. Aston: Aston
[3] British Standards Institution (BSI). Occupational health and University; 2003.
safety management systems – requirements. London: BSI; [19] Behm M, Veltri A, Kleinsorge IK. The cost of safety: cost
2007. Standard No. BS OHSAS 18001:2007. analysis model helps build business case for safety. Prof Saf.
[4] Fernández-Muñiz B, Montes-Peón JM, Vázquez-Ordás CJ. 2004;(April):22–29.
Safety climate in OHSAS 18001-certified organisations: [20] Van den Broek K, De Greef M, Van Der Heyden S.
antecedents and consequences of safety behaviour. Accid Final report socio-economic costs of accidents at work and
Anal Prev. 2012;45:745–758. doi:10.1016/j.aap.2011.10. work-related ill health; European Commission, Directorate-
002. General for Employment, Social Affairs and Inclusion. Ben-
efits of Occupational Safety and Health (benOSH); 2011.
[5] Miller TR. Estimating the costs of injury to U.S. employers.
(VT-2008/066).
J Saf Res. 1997;28(1):1–13. doi:10.1016/S0022-4375(96)
[21] Zou PXW, Shi VY, Li Z. An econometric evaluation frame-
00029-1.
work for investment in construction safety. In: Egbu C, edi-
[6] Ikpe E, Hammond F, Proverbs D, et al. Improving con-
tor. Proceedings 26th Annual ARCOM Conference; 2010
struction health and safety: application of cost-benefit anal-
Sep 6–8; Leeds (UK): Association of Researchers in Con-
ysis (CBA) for accident prevention. Int J Constr Manag.
struction Management; 2010. p. 251–260.
2011;11(1):19–35.
[22] European Agency for Safety and Health at Work. Occu-
[7] Hurns D. Investment in workplace safety triggers positive pational safety and health and economic performance in
bottom line results for businesses, economy, safety official small and medium-sized enterprises: a review. Luxem-
notes. ASSE [Internet] 2004 Mar 23 [cited 2014 Sep 01]; bourg: Office for Official Publications of the European
Research: [about 1 screen]. Available from: https://www. Communities; 2009.
asse.org/newsroom/releases/press362.htm. [23] Lin J, Mills A. Measuring the occupational health and safety
[8] Shalini RT. Economic cost of occupational accidents: evi- performance of construction companies in Australia. Facil-
dence from a small island economy. Saf Sci. 2009;47:973– ities. 2001;19(3/4):131–139. doi:10.1108/0263277011038
979. doi:10.1016/j.ssci.2008.10.021. 1676.
[9] Falkner L, Schneider J, Arnold J. Health and safety, [24] Noweir MH, Alidrisi MM, Al-Darrab IA, et al. Occu-
prevention and accident costs in construction industry pational safety and health performance of the manu-
in international comparison/Arbeitsschutz, Prävention und facturing sector in Jeddah Industrial Estate, Saudi Ara-
Unfallfolgekosten im Bauwesen im internationalen Ver- bia: a 20-years follow-up study. Saf Sci. 2013;53:11–24.
gleich. Geomech Tunn. 2012;5:621–630. doi:10.1002/geot. doi:10.1016/j.ssci.2012.09.005.
201200049. [25] López-Alonso M, Ibarrondo-Dávila MP, Rubio-Gámez MC,
[10] Lamm F, Massey C, Perry M. Is there a link between et al. The impact of health and safety investment on
workplace health and safety and firm performance and construction company costs. Saf Sci. 2013;60:151–159.
productivity? NZ J Emp Relat. 2007;32(1):72–86. doi:10.1016/j.ssci.2013.06.013.
[11] Boles M, Pelletier B, Lynch W. The relationship between [26] Loebbaka JK. Factors defining the relationships between
health risks and work productivity. J Occup Environ Med. safety management strategies and safety performance [doc-
2004;46(7):737–745. doi:10.1097/01.jom.0000131830. toral dissertation]. San Diego (CA): Alliant International
45744.97. University; 2008.
[12] European Agency for Safety and Health at Work. [27] Wachter JK, Patrick LY. A system of safety manage-
Quality of the working environment and productivity: ment practices and worker engagement for reducing and
research findings and case studies. Luxembourg: Office preventing accidents: an empirical and theoretical inves-
for Official Publications of the European Communities; tigation. Accid Anal Prev. 2014;68:117–130. doi:10.1016/
2004. j.aap.2013.07.029.
[13] Podgórski D. Occupational health and safety manage- [28] Idoro GI. Comparing occupational health and safety (OHS)
ment in Polish enterprises implementing total quality man- management efforts and performance. J Constr in Dev
agement systems. Int J Occup Saf Ergon. 2000;Special Ctries. 2011;16(2):151–173.
Issue:85–101. doi:10.1080/10803548.2000.11105110. [29] Agwu MO. Impact of employees safety culture on organ-
[14] Öğer CB. İnsan kaynaklarında çalışan memnuniyeti ve bir isational performance in shell bonny terminal integrated
örnek olay [Employee satisfaction in human resource man- project (BTIP). Eur J Bus Soc Sci. 2012;1(5):70–82.
agement: a case study] [master’s thesis]. İstanbul: Yıldız [30] Battaglia M, Frey M, Passetti E. Accidents at work and costs
Technical University; 2009. Turkish. analysis: a field study in a large Italian company. Ind Health.
[15] Aytaç S. İşyeri kazalarını önlemede güvenlik kültürünün 2014;52:354–366. doi:10.2486/indhealth.2013-0168.
önemi [The importance of safety culture to prevent work- [31] Health and Safety Executive. Perceptions of the cost impli-
place accidents]. Türkmetal Dergisi. 2011;Oct–Nov:1–8. cations of health and safety failures. Nottingham: HSE
Turkish. Books; 2005. (No. RR403).
[16] Kanoğlu B. Çalışan memnuniyeti ve motivasyonuna etki [32] Health and Safety Executive. Assessing the cumulative eco-
eden unsurlar: İSTAÇ A.Ş. örnek uygulama [Factors affect- nomic impacts of health and safety regulations. Kent: HSE
ing employee satisfaction and motivation: case study Books; 2009. (No. 2009RR692).
International Journal of Occupational Safety and Ergonomics (JOSE) 295

[33] Health and Safety Executive. Costs to Britain of [51] Cagno E, Micheli GJL, Jacinto C, et al. An interpretive
workplace fatalities and self-reported injuries and ill model of occupational safety performance for small- and
health, 2012/13. London: Health and Safety Executive; medium-sized enterprises. Int J Ind Ergon. 2014;44:60–74.
2004. doi:10.1016/j.ergon.2013.08.005.
[34] Türkiye Büyük Millet Meclisi [Grand National Assembly of [52] Armstrong JS, Overton TS. Estimating non-response
Turkey]. Occupational health and safety law draft, Ankara; bias in mail surveys. J Mark Res. 1977;14:396–402.
2012. Turkish. doi:10.2307/3150783.
[35] Huang Y-T, Sharyn R-T. The moderating effect of cultural [53] European Agency for Safety and Health at Work. Inventory
congruence on the internal marketing practice and employee of socioeconomics costs of work accidents. Luxembourg:
satisfaction relationship: an empirical examination of Aus- Office for Official Publications of the European Communi-
tralian and Taiwanese born tourism employees. Tour ties; 2002.
Manag. 2014;42:196–206. doi:10.1016/j.tourman.2013. [54] Aaltonen M, Oinonen K, Kitinoja JP, et al. Costs of occupa-
12.005. tional accidents – effects of occupational safety on company
[36] Yorio PL, Wachter JK. The impact of human performance business. European Productivity Conference; 2006 Aug 30;
focused safety and health management practices on injury Espoo (Finland): Dipoli Congress Center; 2006.
and illness rates: do size and industry matter? Saf Sci. [55] Abiuso FL, Serra De La Figuera D. An analysis of the
2014;62:157–167. doi:10.1016/j.ssci.2013.08.014. costs of work-related accidents and illnesses in Catalonia:
[37] Gupta A, Upadhyay D. Impact of occupational health safety a methodological proposal and figures for the years 2006
on employee satisfaction. Int J Sci Res. 2012;1(7):118–120. and 2007 [Internet]. 1st ed. [ebook]. Catalonia: Government
[38] Omusulah ID. Percieved influence of occupational health of Catalonia, Ministry of Employment, Labour Relations
and safety practices on job satisfaction among employ- Department; 2007 [cited 2015 Aug 11]. Available from:
ees in Chemelil Sugar Company Limited [master’s thesis]. http://empresaiocupacio.gencat.cat.
Nairobi: The University Of Nairobi; 2013. [56] Lahiri S, Gold J, Levenstein C. Net-cost model for work-
[39] Nahavandi A, Malekzadeh AR. Organizational behavior. place interventions. J Safety Res. 2005;36(3):241–255.
Upper Saddle River (NJ): Prentice-Hall; 1998. doi:10.1016/j.jsr.2005.06.006.
[40] Torp S, Moen BE. The effects of occupational health and [57] Anderson JC, Gerbing DW. Structural equation modeling
safety management on work environment and health: a in practice: a review and recommended two-step approach.
prospective study. Appl Ergon. 2006;37:775–783. doi:10. Psychol Bull. 1988;103(3):411–23. doi:10.1037/0033-2909.
1016/j.apergo.2005.11.005. 103.3.411.
[41] Grawitch MJ, Trares S, Kohler JM. Healthy workplace [58] Ringle CM, Wende S, Will S. SmartPLS. Version 2.0 (M3)
practices and employee outcomes. Int J Stress Manag. beta. Hamburg: 2005 [cited 2012 Aug 21]. Available from:
2007;14(3):275–293. doi:10.1037/1072-5245.14.3.275. http://www.smartpls.de.
[42] Grant AM, Christianson MK, Price RH. Happiness, [59] Hair JF, Hult GTM, Ringle C, et al. A primer on par-
health, or relationships? Managerial practices and employee tial least squares structural equation modeling (PLS-SEM).
well-being tradeoffs. Acad Manag Perspect. 2007;51–63. Thousand Oaks (CA): Sage; 2014.
doi:10.5465/AMP.2007.26421238. [60] Henseler J, Ringle C, Sinkovics R. The use of partial least
[43] Lingard H, Wakefield R. The development and test- squares path modeling in international marketing. Adv Int
ing of a hierarchical measure of project OHS perfor- Mark. 2009;20:277–319.
mance. Eng, Const Architect Manag. 2011;18(1):30–49. [61] Lumpkin GT, Dess GG. Linking two dimensions of
doi:10.1108/09699981111098676. entrepreneurial orientation to firm performance: the mod-
[44] Arezes PM, Miguel SA. The role of safety culture in safety erating role of environment and industry life cycle. J Bus
performance measurement. Meas Bus Excell. 2003;7(4):20– Ventur. 2001;16:429–451. doi:10.1016/S0883-9026(00)000
28. doi:10.1108/13683040310509287. 48-3.
[45] Health and Safety Executive key process safety perfor- [62] Bartlett MS. A note on the multiplying factors for vari-
mance indicators, a short guide for directors and CEOs ous chi square approximations. J R Stat Soc Series B Stat
[Internet]. London: Health and Safety Executive [cited 2015 Methodol. 1954;16:296–298.
Feb 27]. Available from: http://www.hse.gov.uk/leadership/ [63] Kaiser HF. An index of factorial simplicity. Psychometrika.
keyindicators.pdf. 1974;39:31–36. doi:10.1007/BF02291575.
[46] Lingard H, Wakefield R, Nick B. “If you cannot measure [64] Podsakoff PM, MacKenzie SB, Lee JY, et al. Common
it, you cannot improve it”: measuring health and safety per- method biases in behavioral research: a critical review of
formance in the construction industry. In: McCarthy MJV, the literature and recommended remedies. J Appl Psychol.
Editor KS, editors. Proceedings of the 19th CIB World 2003; 88:879–903. doi:10.1037/0021-9010.88.5.879.
Building Congress; 2013 May 5–9; Brisbane, Queensland [65] Kathuria R. Competitive priorities and managerial per-
(Australia): Queensland University of Technology; 2013. formance: a taxonomy of small manufacturers. J Oper
[47] Fitzgerald MK. Safety performance improvement through Manag. 2000;18(6):627–641. doi:10.1016/S0272-6963(00)
culture change. Process Saf Environ Prot. 2005;83(4):324– 00042-5.
330. doi:10.1205/psep.04381. [66] Fornell C, Larker DF. Evaluating structural equation models
[48] Gyekye SA, Salminen S. Making sense of industrial acci- with unobservable variables and measurement error. J Mark
dents: the role of job satisfaction. J Soc Sci. 2006;2(4):127– Res. 1981;18(1):39–50. doi:10.2307/3151312.
134. [67] Hair JF Jr, Anderson RE, Tatham RL, et al. Multivariate data
[49] Gyekye SA. Workers’ perceptions of workplace safety and analysis. 5th ed. Upper Saddle River (NJ): Prentice Hall;
job satisfaction. Int J Occup Saf Ergon (JOSE). 2005; 1998.
11(3):291–302. doi:10.1080/10803548.2005.11076650. [68] Cronbach LJ. Coefficient alpha and the internal structure
[50] Rundmo T. Associations between safety and contingency of tests. Psychometrika. 1951;16(3):297–334. doi:10.1007/
measures and occupational accidents on offshore petroleum BF02310555.
platforms. Scand J Work Environ Health. 1994;20:128–131. [69] Chin WW. The partial least squares approach to structural
doi:10.5271/sjweh.1421. equation modeling. In: Marcoulides GA, editor. Modern
296 M. Bayram et al.

methods for business research. Mahwah (NJ): Erlbaum; [73] Hu L, Bentler PM. Fit indices in covariance struc-
1998. p. 295–336. ture modeling: sensitivity to underparameterized model
[70] Fornell C, Cha J. Partial least squares. In: Bagozzi RP, misspecification. Psychol Methods. 1998;3(4):424–453.
editor. Advanced methods in marketing research. Oxford: doi:10.1037/1082-989X.3.4.424.
Blackwell; 1994. p. 52–78. [74] Kleinbaum DG, Kupper LL, Muller KE. Applied regression
[71] Tenenhaus M, Esposito V, Chatelin Y, et al. PLS path analysis and other multivariate analysis methods. Boston
modelling. Comput Stat Data Anal. 2005;48:159–205. (MA): PWS-Kent; 1988.
doi:10.1016/j.csda.2004.03.005. [75] Little TD, Card NA, Bovaird JA, et al. Structural equation
[72] Henseler J, Dijkstra TK, Sarstedt M, et al. Common beliefs modeling of mediation and moderation with contextual fac-
and reality about partial least squares: comments on Rönkkö tors. In: Little TD, Bovaird JA, Card NA, editors. Modeling
& Evermann (2013). Organ Res Methods. 2014;17(2):182– contextual effects in longitudinal studies. Mahwah (NJ):
209. doi:10.1177/1094428114526928. Erlbaum; 2007. p. 207–230.

You might also like