Available online at www.sciencedirect.

com

ScienceDirect
Procedia Manufacturing 2 (2015) 286 290

2nd International Materials, Industrial, and Manufacturing Engineering Conference, MIMEC2015,

4-6 February 2015, Bali Indonesia

Occupational Accident Direct Cost Model Validation Using

Confirmatory Factor Analysis
Jafri Mohd Rohania*, Mohamed Fitri Joharia, Wan Harun Wan Hamida, Hood Atana,
Ademola James Adeyemia, Amirmudin Udinb
a
Faculty of Mechanical Engineering, Universiti Teknologi Malaysia, 81300 Skudai, Johor, Malaysia
b
Faculty of Education, Universiti Teknologi Malaysia, 81300 Skudai, Johor

Abstract

Accident cost estimation become more important since it may cause financial implication to the company. However, there are no
consensus between researchers in classifying direct cost component. The objective of this study is to evaluate direct cost model
of occupational accident cost by using confirmatory factor analysis (CFA) method. There are three cost components proposed
for direct cost estimation which are medical cost, rehabilitation cost and compensation cost. Local specific approach (by
reviewing previous accident record in the company) is adopted in data collection process and only accidents occurred between
2008 and 2013 were considered in this study. There are 344 cases in total which came from six different sectors in
manufacturing industry. For each occupational accident, the site safety officer in charge, human resource and account department
officer were interviewed in order to estimate the direct cost. Then, the relationship between three observed variables for direct
cost is evaluated by using CFA. The CFA result shows that the occupational accident direct cost model has acceptable goodness
of fit that proved relationship between three observed variables (medical, rehabilitation and compensation cost) and its latent
construct. In conclusion, the model justify that direct cost consists of three components, which are medical cost,
rehabilitation cost and compensation cost.

Authors.Published
2015 The Authors. Publishedby
byElsevier
ElsevierB.V.
B.V.This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Selection and Peer-review under responsibility of the Scientific Committee of MIMEC2015.
Selection and Peer-review under responsibility of the Scientific Committee of MIMEC2015
Keywords: Confirmatory factor analysis; occupational accident cost; direct cost; manufacturing industry; local specific

* Corresponding author. Tel.: +6075534568

E-mail address: jafri@fkm.utm.my

2351-9789 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Selection and Peer-review under responsibility of the Scientific Committee of MIMEC2015
doi:10.1016/j.promfg.2015.07.050
 Jafri Mohd Rohani et al. / Procedia Manufacturing 2 (2015) 286 290 287

1. Introduction

Occupational accident at the workplace has become huge issue surrounding employers especially in
manufacturing industry. Malaysia economy growth in past decade has created more manufacturing based companies
and providing jobs to the citizen. Eventually, manufacturing sector had been recognized as one of the catalyst for
national economy growth. However, the significant growth in this industry has sparked a new concern especially
related to occupational accident at the workplace.

Jallon et.al (2011) stated that occupational accident definitely will give negative impact to the employers,
employees and even to the nation as a whole. For employers, they shall expect lower productivity and profit due to
force closure or stop work order that being issued by enforcement agency [10]. Disruption of production process
will cause inability for the company to fulfill customer orders. As for national, the government need to spend money
for compensation payment to the victim and also covers medical expenses associated with that accident. In contrast,
occupational accident not only cause suffering to the victim alone but also affecting their household income and
quality of life. In the worst case scenario, family may not just lost their love ones but also lost someone who take
care for their family.

Based on statistics provided by Social Security Organization (SOCSO), there are 16,684 reported case for
manufacturing sector in 2012 alone. This figure represent 25% of total 66,152 cases across industries which can
be considered as alarming situation since the manufacturing sector is one of the major contributor to the
Malaysian economy, such high occupational accidents obviously will have an impact on the productivity of
the workers [12]. Lower productivity by the workers will in turn cause an adversarial effect to the organization
especially on financial viewpoint and could also affect national economy in the long run. Therefore, reliable method
on accident cost assessment has become crucial in order to show the true monetary cost to the managers with
hope that this assessment can change their mind set and to encourage them to take greater measure to curb this
cost.

Most researchers suggested that, accident cost estimation can be separated into two broad categories which are
direct cost and indirect cost (Gosselin, 2004) [3]. However there are slight differences regarding to the
terminologies used for those cost categories. LaBelle (2000) consider direct cost as visible cost which can be
identified easily and already aware by the company [6] while Sun et.al (2006) stated direct cost as a cost that being
compensated by insurance entity after the event of accident occurred [12]. Basically, the employer would pay
certain amount of premium annually to the insurance company. In return the insurance company will cover medical
and hospitalization expenses of the victim (Jallon et.al, 2011) [10]. This kind of practice normally happened in most
developed country. Different definition proposed by various researchers has caused different way in classifying the
direct cost components. Generally, direct cost shall include: (a) cost of medical treatment and hospitalization for the
victim, (b) cost of rehabilitation for worker recovery before returning to work and (c) sum of compensation
claimable by the victim from insurance company and social security agency.

This paper proposes to validate direct cost model for occupational accident cost by using confirmatory factor
analysis (CFA) method. CFA will assess the relationship between three observed variable and their underlying
latent (unobserved) variable. In this case, latent construct is represented by direct cost (DirectC) while observed
variable proposed are medical cost, rehabilitation cost and compensation cost. In addition, the relationship also can
be regarded as hypothesis model for this study.

According to Diana (2014), CFA is quite popular in social science field for example psychology and social
studies [2]. Based on literature the application of CFA method also not common in research studies on engineering
and occupational accident cost estimation. Pratarelli (2002) adopt CFA technique to investigate the relationship
internet addiction with sexual factor, internet use factor and addict factor [9]. Diana (2014) validated airport delay
construct which consists of demand, operation and delay against six observed variables in civil aviation industry [2].
288 Jafri Mohd Rohani et al. / Procedia Manufacturing 2 (2015) 286 290

Lietdka (2002) used confirmatory approach to evaluate the effect of information content towards nonfinancial
performance measure in airline industry [8].

2. Methodology

The data has been collected through local specific approach from six different companies which represents six
different sectors in manufacturing industry. The sectors included as follows: (a) sawmills and planing (b)
manufacture of furniture and fixture (c) industries related to printing and publishing (d) manufacture of plastic
products (e) manufacture of electrical and electronic products and (f) iron and steel basic industry.

Occupational accident case was obtained by reviewing companys previous accident record. Case recorded
between 2008 and 2013 has been selected and severity of accident limited only to temporary disability. Besides,
first aid cases (which has no medical leave, MC = 0) were excluded from this study. Finally, there are 344 cases
available in total after data verification process is completed.

For each occupational accident, the site safety officer in charge, human resource and account department officer,
supervisor, victim and related co-workers were interviewed in order to estimate the actual cost for each cost
component. Cost components proposed are: medical cost (MedC), rehabilitation cost (RehabC) and compensation
cost (ComC). Then, the data is stored in SPSS Version 20 software. After that, model validity is examined by using
CFA method.

The objective of CFA is to examine whether there is any relationship between latent construct (unobserved) and
observed variables. Pratarelli (2002) added that, CFA can assess the ability of hypothesized model to fit with
collected data. The conceptual of direct cost model is shown as in Figure 1. Based on that figure, the cost
components also known as endogenous variables while latent construct (DirectC) and its residual terms (e1, e2 and
e3) known as exogenous variables. The observed variables are assumed to be known and usually embedded in
theories (Lei and Wu, 2007; Diana, 2014) [7, 2]. The value inside each observed variable can be determined by
collecting appropriate cost data inside the company.

Fig. 1. Conceptual of direct cost model

3. Result and discussions

Table 1 shows the factor loadings estimates for each observed variable and its squared multiple correlation
(SMC). Three proposed endogenous variables can be considered statistically significant since the p-value for those
variables are less than 0.05. In other words, the calculation of this three variables will give contributions to the
direct cost. For example, When direct cost goes up by 1 standard deviation, medical cost also goes up by 0.947
standard deviations and the same rule also applied to rehabilitation and compensation cost. As for magnitude of
factor loading, medical cost has the greatest effect to total direct cost and followed by rehabilitation and
compensation cost. Hence, the result reflects as in practice and in line with direct accident cost theory suggested by
previous study. Steady increase of medical service fees in recent decades proved the finding of this research
especially on medical cost.
 Jafri Mohd Rohani et al. / Procedia Manufacturing 2 (2015) 286 290 289

Table 1. Factor loading estimates and squared multiple correlation

Observed Variables Factor loading estimates SMC

MedC 0.947 0.897
RehabC 0.791 0.626
ComC 0.292 0.085

In other hand, this model can be regarded as saturated model since it has zero degree of freedom. It happened due
to number of unknown parameters exactly equals the number of known parameters. A saturated model indicates
ability of the model to reproduce almost the same result as in theory with slight difference. Besides, there is no
better model can be done with this data since the model is already saturated.

There are several fit indices that can be used to evaluate the hypothesis model. This evaluation is important in
order to make decision, either accept or reject the hypothesis proposed. There are three broad categories of fit
indices available which are absolute fit index, incremental fit index and parsimony fit index. However, this paper
only concentrate on absolute fit index specifically goodness-of-fit index (GFI). This kind approach also has been
adopted by Pratarelli et. al. (2002) in assessing causal factor of internet use and addiction.

Generally, GFI is an alternative of Chi-square index which calculate proportion of variance that is accounted for
by estimated population covariance (Hooper et.al, 2008) [4]. Bentler and Bonnet (1980) stated that GFI is
considerably better than Chi-square index because the latter is sensitive to sample size and may always reject the
properly specified model when the sample size become larger [1]. Besides, it may fail to differentiate either the
model has good or bad fitting (Kenny and McCouch, 2003) [5]. The range of this fit indices are between 0 and 1.
Higher value will reflect better fit.

The value of GFI computed by AMOS Version 21 software is 1.0 which indicates perfect model fit.
Traditionally, the cut-off point greater than 0.90 is recommended by various researchers. However, 0.95 is more
appropriate if the sample size is small. Perfect fit achieved by direct cost model shows significant relationship
between accident total direct cost and those three factor (medical cost, rehabilitation cost and compensation cost).
This finding also validates the theoretical assumption proposed for the direct cost.

4. Conclusion

This study propose theoretical assumptions of accident direct cost and validated through confirmatory factor
analysis (CFA). The data is obtained from six difference sector in manufacturing industry by using local specific
approach. The data is gathered by referring to previous accident record within last six years (2008 to 2013).
Meanwhile the cost information is done by interviewing management and relevant staff whom involve in the
accident event. Therefore, there are 344 cases in total occurred within that time frame.

Factors proposed for direct cost are medical cost, rehabilitation cost and compensation cost. This theoretical
assumption is tested using AMOS Version 21 software and shows statistically significant result based on goodness-
of-fit index (GFI). Factors loading estimates also show the effect of each cost components towards total direct cost.
In this case, medical cost plays an important role in influencing the total direct cost compared to the other factor. In
other hand, the model is considered as just-identified model since it has zero degree of freedom no better model can
be produced.
290 Jafri Mohd Rohani et al. / Procedia Manufacturing 2 (2015) 286 290

Future research should be done in order to assess the effectiveness of this model to different type of industry for
example construction, agriculture, utility and others. Currently, it is unclear whether the factor loading value and
model fit indices would remain consistent or not. Furthermore, there is no research on CFA has been done in other
industry. It is not easy to come with a single universal direct cost model that can be used for various types of
industry in calculating the accident direct cost. However, the theoretical framework provided in this study would
provide a new research perspective for researchers.

References

[1] Bentler, P.M. and Bonnet, D.C. (1980), "Significance Tests and Goodness of Fit in the Analysis of Covariance Structures," Psychological
Bulletin, 88 (3), 588-606.
[2] Diana, T. Validating delay constructs: An application of confirmatory factor analysis, Journal of Air Transport Management, Volume 35,
March 2014, Pages 87-91, ISSN 0969-6997,
[3] Gosselin, M. (2004).Analyse des avantages et des cots de la sant et de la scurit au travail en entreprise: Dveloppement de l'outil
d'analyse. Rapport de recherche R-375. Montral, QC: IRSST
[4] Hooper, D., Coughlan, J., Mullen, M.: Structural Equation Modelling: Guidelines for Determining Model Fit. Electronic Journal of Business
Research Methods, 6(1), 53-60
[5] Kenny, D.A. and McCoach, D.B. (2003), "Effect of the Number of Variables on Measures of Fit in Structural Equation Modeling," Structural
Equation Modeling, 10 (3), 333-51.
[6] LaBelle, J. E. (2000). What do accidents truly cost? Determining Total Incident Costs. Professional Safety, 45(4), 3842.
[7] Pui-Wa Lei and Qiong Wu (2007). Introduction to Structural Equation Modeling: Issues and Practical Considerations, The Instructional
Topics in Educational Measurement Series (ITEMS), 26(3)
[8] Liedtka, S. L. (2002). The information content of nonfinancial performance measures in the airline industry. Journal of Business Finance &
Accounting, 29(78), 1105-1121.
[9] Pratarelli, M. E., & Browne, B. L. (2002). Confirmatory factor analysis of internet use and addiction. Cyberpsychology & Behavior : The
Impact of the Internet, Multimedia and Virtual Reality on Behavior and Society, 5(1), 53-64.
[10] Romain, J., Imbeau, D., Marcellis-Warin, N. Development of an indirect-cost calculation model suitable for workplace use, Journal of Safety
Research, Volume 42, Issue 3, June 2011, Pages 149-164
[11] Sun, L., Paez, O., Lee, D., Salem, S., & Daraiseh, N. (2006). Estimating the uninsured costs of work-related accidents, part I: a systematic
review. Theoretical Issues in Ergonomics Science, 7(3), 227245