Professional Documents
Culture Documents
www.elsevier.com/locate/ssci
Abstract
Traditional approaches on the prevention of accidents/injuries in mines reached its limit of eVec-
tiveness in improving safety performance and a fresh approach is utmost required. Behavioral
safety analysis has been identiWed as an eVective alternative in many industries. This paper is there-
fore sought to examine the role of behavioral factors on the occurrence of mine accidents and inju-
ries through a case study. Data were collected from two neighboring underground coalmines
operating under a large public sector organization of India. High–low plots and t-test were done to
explore the diVerences between behavioral characteristics of accident involved (case) and non-
involved (control) workers. How these diVerences could cause accidents/injuries in mines was esti-
mated through structural equation modeling. The case study results show that accident group of
workers (cases) are more job dissatisWed, negatively aVected, and highly risk taking compared to the
non-accident group of workers (controls). The accident model path analysis shows that negative
aVectivity, job dissatisfaction, and risk taking behaviors predict an increased number of injuries in
mines. Apart from direct inXuences to work injuries, negative aVectivity and job dissatisfaction
make workers to take more risks and behave unsafely. These Wndings contribute to the design of
safety programs including safety training, which should be behaviorally motivated. Mine safety
management of the case study mines should outskirt their age old belief that accidents/injuries are
due to hazardous nature of mining and only engineering control and regulatory monitoring are
suYcient for improving safety of the mines. The multivariate analysis also shows that experience
bears no relationships with work injury indicating that a less experienced worker is equally likely to
be injured as an experienced worker. It implies that experience though helps workers in understand-
ing the physical hazards, however, avoiding the imminent danger is much more behavioral. The
*
Corresponding author. Tel.: +91 3222283750; fax: +91 3222 282272.
E-mail addresses: partha_sp99@yahoo.com (P.S. Paul), jmaiti@iitkgp.ac.in, jmaiti@iem.iitkgp.ernet.in
(J. Maiti).
0925-7535/$ - see front matter © 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ssci.2006.07.006
450 P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471
variables negative aVectivity, job dissatisfaction, and risk taking behaviors are therefore crucial in
avoiding accident/injuries in mines.
© 2006 Elsevier Ltd. All rights reserved.
Keywords: Mine safety management; Occupational injury; Behavioral factors; Structural equation modeling
1. Introduction
is often characterized as an enduring aspect of the organization (Cox and Cheyne, 2000).
Organizational safety culture as reported by Cox and Cheyne (2000) composed of (i) an
organizational attribute which manifests in safety policy, systems and processes, structures
and reports, (ii) perceptions of the organizations which manifest in employee, contractors,
and external perceptions, and (iii) individual perceptions which manifest in employees
commitment, attitudes, responsibility and behaviors.
The organizational approaches to safety focus on organizational factors aVecting safety,
both macro and micro levels, which stem through improved process and procedure, man-
agement policies, supervision and training, management-workers relationships, manage-
ment commitment and others. Several recent publications on injury investigation point to
the increasing importance of the role of organizational factors, as antecedents to the
sequence of an injury. Amongst these studies are those of Turner (1978), Wright (1986),
Wagenaar and Groeneweg (1987), Shrivastava’s Bhopal (1987), Dawson et al. (1991),
Hurst et al. (1991), Embrey (1992), Simard and Marchand (1995), Hofmann and Stetzer
(1996), Kamp and Krause (1997), Brown et al. (2000), and Cox and Cheyne (2000). Wright
(1986), for example, had investigated the relevance of organizational factors as the genesis
of the sequence of the injury situation, and identiWed three areas namely dangerous pro-
cesses, production/task related pressure and communication as the major antecedents to
the sequence of an injury. Wagenaar and Groeneweg (1987) in a study in which they review
one hundred injury situations at sea to determine the types of human error, also detected
organizational factors such as information processes, and social pressure aVect human
behaviors. Turner (1978) in his studies of worksite disasters pointed out that communica-
tion problems contribute to injury causation. He argued, however, that the events that
cause disasters accumulate because they are ignored or misinterpreted, as the result of hab-
its or routine, false beliefs, inadequate communication, thoughtless norms or instructions
and unjustiWed optimism. He also emphasized that safety practices require the intervention
of management and all the work force. Managers must create a positive safety culture and
an open atmosphere of learning in which errors and incidents can be openly discussed
without blame or recrimination. At a more organizational level, Shrivastava’s Bhopal
(1987) states that injuries are the consequence of the following three factors (i) human
beings (ii) organizations and (iii) technology used. Other authors have also voiced equally
to the above factors. Embrey (1992), for example, identiWed key factors of organization
that may have an inXuence on the development of safety: the balance between safety and
production, time pressure, communication and co-ordination systems and the safety cul-
ture. Dawson et al. (1991) similarly concluded that the channels of communication and the
safety culture are two key factors that may inXuence the understanding of technical and
social systems of organization by the worker. In their study, Simard and Marchand (1995)
investigated the propensity of workgroups to take safety initiatives and the inXuence of
various organizational factors that may impact on workers’ safety behaviors. They found
that the micro organizational factors such as work-process and hazards, workgroup cohe-
siveness and co-operation, and supervisor’s experience and approaches to safety manage-
ment are the primary determinants of the propensity of workgroups to take safety
initiatives. Similar Wndings were reported by Reason (1995, 1997) for rail accidents, and
Caird and Kline (2004) for highway accidents.
Studies on the perspective of workplace safety therefore unfold two major categories of
factors namely, individual and organizational. The authors representing the view under the
individual-as-cause heading concluded that employee’s attitudes and behaviors are the
452 P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471
most important antecedents to unsafe acts, accidents, and injuries. The general implica-
tions here is that employee centered Wxes can solve safety problems (Brown et al., 2000).
Although, the researchers are most certainly aware of the role of organizational system,
however, they base their persons centered solution on the assumptions that Wrst line
employee play the primary role in accidents.
Studies in the area of mine safety Weld are primarily descriptive. Many studies focused
on the relationship between accidents and factors that are, from the standpoint of manage-
ment, diYcult to alter, for example, age, experience, and some personality characteristics of
the work force (PXeider and Krug, 1973; Theodore Barry and Associates, 1972; Bennett,
1982; Bennett and Passmore, 1986). Other studies identiWed factors important to the func-
tioning of an eVective safety program. These studies included the compiling of suggestions
of mine safety oYcials and supervisory personnel (Pfeifer et al., 1976; Demichiei et al.,
1982), the description of safety activities of award wining companies (Rhoton, 1980), and
the comparison of safety eVorts of high and low accident mines (Demichiei et al., 1982;
Gaertner et al., 1987). Although these studies are of important initial steps, the conclusions
that can be drawn from them are limited. Recent studies suggest that there should be more
in-depth studies and analysis of human error (behavior) need to device ways and means to
further reduce the accidents/injuries in mines (Souder, 1988; Shaw et al., 1989; Simson,
1994; Ray and Bhattacharjee, 1997; Maiti and Bhattacharjee, 1999; Maiti and Bhattachar-
jee, 2000; Maiti et al., 2004). For example, Maiti et al. (2004) investigated the determinants
of work injury through structural equation modeling and found that emotionally unstable
and dissatisWed individuals are major safety problems for the mine studied. They recom-
mended behavioral safety approaches might be eVective in accident/injury reduction.
In this paper, the individual as cause axiom was investigated. SpeciWc emphasis was
given on risk taking behaviors, job dissatisfaction and negative aVectivity of mine workers
leading to accidents/injuries in mines. The pattern and strength of relationships of these
factors with work injuries were assessed through structural equation modeling.
2. Theoretical model
The basic causes for high injury experience rates are unsafe conditions, unsafe acts, or
both. Unsafe conditions may arise through insuYcient mine design, unanticipated geologi-
cal conditions, inadequately maintained equipment, inadequate supervision, or a combina-
tion of these factors (Bhattacharjee, 1991). Unsafe acts mainly arise through behavioral
related causes. Unsafe behaviors are said to both directly and indirectly contribute to 90%
of all workplace accidents and incidents (Holnagel, 1993). Given the importance of behav-
ioral patterns in the accident process, it is not surprising that safety improvements focused
on individual behaviors have acquired popularity in the development of safety perfor-
mance (Cooper et al., 1994; Cox and Cox, 1996; Krause et al., 1990; Krause, 1995; Krause
et al., 1999). However, studies on behavioral indicators of mine accident are rather limited.
This is particularly due to several reasons. First, mining is a hazardous occupation. It was
initially believed that elimination of hazardous conditions alone would result in better
safety performance. Second, mining industry is highly regulated. Stringent rules and regu-
lation are in practice and the regulating bodies are perhaps complacent with their function-
ing for overall safety improvement in mines. Recent studies (Maiti, 1999; Maiti and
Bhattacharjee, 2001; Maiti et al., 2001) showed that individual miner’s obligation plays an
important role in accident/injury causation in mines. Therefore, based on the analysis of
P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471 453
mining and non-mining industries, a theoretical model was developed including the follow-
ing variables: miner’s age, experience, negative aVectivity, job dissatisfaction, risk taking,
safety performance rating and work injury. The theoretical model is shown in Fig. 1. Brief
description of the Wndings of literature review for each variable used in this study and their
hypothesized relationships with work injuries are described below.
2.1. Variables
2.1.1. Age
The relationship between age and safety performance was an issue since the beginning
of the twentieth century when the concept of accident proneness was Wrst proposed by
Greenwood and Woods (1919). It is believed that as aged person tend to have higher expe-
rience they are likely to be more familiar with the safe behavior of the work which led them
to be less injurious. Other school of thought is that aged workers tend to be fatigued and
their reduced reXex action placed them into accident even if they perceive the situation
fully. Subsequently, many studies were conducted to observe the eVect of age on work inju-
ries. The National Academy of Science’s study looked at the relationship between the age
and various accident rates for 15 of the largest underground coal producing companies in
USA (1982) and they found strong negative correlation between age and work injuries in
coal mining. Shafai-Sahrai (1973), Oi (1974), and Root (1981) showed that the injury rates
decrease with age whereas Durry (1965) and Bennett and Passmore (1986) observed the
opposite trend. Bennett (1982), Kenny (1993), and Maiti and Bhattacharjee (1999) found
no relationship between age and work injuries.
2.1.2. Experience
Experience has received major attention in prior research on work injuries in mines,
although conceptual disagreement exists regarding the potential direction of this relation.
Experience represents the amount of time an employee engaged in his work. This may be
?
Age
Safety
? Performance
Experience -
? ?
Work
? + Injury
+
Negative +
Affectivity ? -
+
Risk
Taking
Job +
Dissatisfaction
Fig. 1. Hypothesized accident model path diagram [‘+’ indicates positive relationships, ‘¡’ indicates negative rela-
tionships and ‘?’ indicates no deWnite relationships].
454 P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471
his total mining experience or the job experience. Experience has been argued to have a
positive relation with injury (Hansen, 1989). Employees with higher experience generally
are assigned to jobs with greater skill requirements, responsibility and accident risk (Han-
sen, 1989; Iverson and Erwin, 1997). The concept of familiarity and perception of hazards
dictates that experience should have negative relation with work injury. Theodore Barry
and Associates (1972) examined the eVect of job experience and total mining experience on
accident rate. Though job experience had a strong relationship with accident rate, the total
mining experience was not signiWcant regarding accident occurrence. Root and Hoefer
(1979) indicated that 40% of injuries occurred during the Wrst year of employment. Bennett
(1982) showed that more experienced miners were slightly less probable of receiving inju-
ries than less experienced miners. However, Bennett and Passmore (1986), Phiri (1989), and
Maiti and Bhattacharjee (1999) found that injury severity was not related to miner’s total
mining experience.
behaviors are assessed to involve a high degree of risk in comparison with other equivalent
behaviors, and also involve a high degree of actual risk as measured by the probability of
death, injury, Wnancial loss and so on. Employees diVer in their willingness to take chances.
The propensity to assure or avoid risk has signiWcant impact on safety. Risk taking may
have bad consequences such as work injuries but the workers occasionally repeat this
behavior to complete the work fast or for other deliberate intention sometimes knowing
the outcomes of such behaviors. Cooper and Sutherland (1987) and Sutherland and Coo-
per (1991) stated that type A workers were more likely to experience work injuries than
type B workers. Maiti et al. (2004) found that risk taking behaviors signiWcantly led work-
ers to be accident prone.
2.2. Hypotheses
Epidemiological studies revealed that there are signiWcant diVerences between behav-
ioral characteristics of accident involved and not-involved employees at work. Epidemio-
logical accident causation postulates that these diVerences may lead to accident/injury
occurrences in mines. The present study was directed to explore the diVerences between the
accident and non-accident group of workers based on their biographic and behavioral
characteristics. The following hypotheses were tested in this study:
H1: Personal level variables such as age, experience, negative aVectivity and job dissatis-
faction aVect risk taking behavior. Negative aVectivity and job dissatisfaction are
sought to have positive relationships with risk taking behaviors. No deWnite relation-
ships are assumed between age, and experience with risk taking behaviors as litera-
ture showed conXicting results. Safety performance also aVects risk taking behavior
negatively.
H2: The personal variables age, experience, and negative aVectivity predict safety perfor-
mance. Negative aVectivity is hypothesized to have negative relationships with safety
456 P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471
performance. No deWnite relationships are assumed between age, and experience with
safety performance.
H3: The four personal level variables are assumed to predict work injury in the mine.
Negative aVectivity, and risk taking behaviors are hypothesized to have positive rela-
tionships with work injury. Age and experience again bear no deWnite relationships
with work injury as conXicting results were indicated by literature review. Risk taking
behavior triggers work injury whereas good safety performance is assumed to moder-
ate these relationships negatively.
3.1. Sample
The setting for this study was two neighboring underground coalmines within a large
public sector organization in the eastern part of India. The mines were selected because
work injuries of these mines are high. One mine was declared as accident prone mine. The
mines comprised of a total 1000 underground workers those are directly involved in pro-
duction. The sample consists of 300 underground mine workers who participated in this
study (response rate 80%). Of the participants 50% were recorded to be injured, 48% were
loaders who works in the face for loading coals into tubs, 12% were engaged in coal prepa-
ration, 14% in transport work, 6% in supporting, 7% in engineering department, and the
rest were involved in others works. On an average, the participants were 37.34 (SD D 9.01)
years old, and held their current jobs for 14.58 (SD D 9.25) years. Of the injured 65 were
from the Mine 1 and 85 from the Mine 2.
3.2. Instruments
Measures used in this research were miners’ age, experience, negative aVectivity, risk
taking, job dissatisfaction, safety performance and work injury. Miners’ age and experience
were collected from the mines’ personnel log book. The constructs negative aVectivity, risk
taking, and job dissatisfaction were measured through questionnaire survey. The scale
items, means, standard deviations of each of the three constructs are shown in Table 1. The
safety performance of each respondent was obtained from supervisors’ rating. Three
immediate supervisors (mine sirdar, overman, and section incharge) were asked to rate the
Table 1
Construct scale items, means, and standard deviations
Construct Scale items Summed item Construct Sourcesb
means (SD) reliabilitya
Risk taking Do you like to take risk 2.20 (0.94) 0.82 MOSHAB (1998),
Do you think life is dull if there is no danger in life 1.99 (0.91) Demichiei et al. (1982),
Are you ready to take risk in order to increase your income 2.04 (0.96) and Dhar et al. (1997)
Do you engaged in illegal mining in order to earn more money 2.00 (0.91)
Do you take risk to reach your goal 2.20 (0.91)
Do you take risk to achieve other objectives 1.95 (0.96)
Do you have tendency to work quickly 2.31 (0.94)
Do you like to walk quickly inside the mine 2.06 (0.91)
Do you ready to work in a daring situation 2.33 (0.83)
Do you venture knowingly into unsupported place 2.29 (0.59)
Have you been motivated to take risk while seeing others are taking risk 1.84 (0.88)
Negative aVectivity Do you become disappointed easily when your thinking does 2.29 (0.91) 0.80 Demichiei et al. (1982)
not match with other and and Dhar et al. (1997)
Do you feel problem to express your thought 1.78 (0.94)
Do you become restless from little matter 1.55 (0.79)
Can you decide what will have to do in your work even at the end of the day 1.85 (0.84)
Can you talk freely with your top bosses 1.50 (0.82)
If anybody blames you without any cause then do you feel guilty yourself 1.83 (0.93)
If your oYcer wants to meet with you, whether it appears 1.92 (0.88)
to you that you have done some mistake
Without any reason do you feel sometimes angry 1.92 (0.94)
P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471
Do you feel sometimes happy or frustration without any reason 1.69 (0.85)
Do your behavior change without any reason 1.76 (0.79)
Do you take your task lightly 1.73 (0.91)
Do you feel frustration and sorrow oftenly 1.61 (0.82)
Do you think that you speak large 1.61 (0.84)
Do some undesirable thought come to your mind and harass you 2.09 (0.89)
Do you fall in tension when you think about the future of your life 2.30 (0.84)
(continued on next page)
457
458
Table 1 (continued)
Construct Scale items Summed item Construct Sourcesb
means (SD) reliabilitya
Job dissatisfaction Do you think you have got right job for yourself 1.78 (0.91) 0.83 International
Do you think your life has become burden to you because of this job 1.59 (0.81) Survey Research
Do you want to change your company if you get same opportunity there 2.01 (0.86) Corporation (1992)
You are doing this job because you are getting suYcient salary 2.04 (1.10)
inspite of your disliking to this job
Do you feel the workers are not motivated due to same manner of work 2.09 (0.85)
Are you satisWed with the condition of your workplace 2.12 (0.95)
Does your oYcer appreciate your eligibility and quality 1.82 (0.74)
Had you been an oYcer would you behave in the same manner as your oYcer does 2.43 (0.80)
Do you think that people get promotion here through oiling only 2.28 (0.88)
Do you think this work is a source respect or a man like you 1.82 (0.87)
Do other people respect to your work 1.79 (0.73)
Is there any prospect in your work 1.74 (0.85)
Is there any opportunity to get promotion for workers 1.78 (0.90)
a
Computed taking the signiWcant items.
P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471
b
Guided in framing questionnaires; bold-faced questions are reverse scored.
P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471 459
participants based on their overall safety performance. These three supervisors are respon-
sible to control the works as well as workers in each shift of work. The work injury was
measured through a dichotomous variable (1 D injury, 0 D no injury) obtained from each
respondent’s personnel records. The following section describes the details of the three
behavioral constructs measured through multi-item questionnaire survey.
A multi-item survey measure was administered to the participants during their working
hours. A three point Likert-type scale format (Kothari, 2001) was used to measure the par-
ticipants’ responses on each item of the three constructs. Initially, a factor analysis was
done to explore the factor loadings of each of the items to the construct (factor) of interest.
A loading of 0.3 or more was considered signiWcant (Rahman, 2000), and items with factor
460 P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471
loading less than 0.3 were discarded. Then item correlations of the remaining items were
examined, and items with low (<0.3) correlations were deleted. Finally, the items which met
the factor loading and item correlation criteria were considered as signiWcant indicators of
the construct. The internal consistency reliability of each of the constructs was measured
based on Cronbach’s alpha (SPSS, 1999). For example, negative aVectivity was measured
through 15 items. A factor analysis revealed that two items showed a factor loading of <0.3
and were discarded. The item correlations of the remaining 13 items suggested for deletion
of another 2 items. Finally, 11 items were considered as the signiWcant indicators of nega-
tive aVectivity. Cronbach’s alpha was 0.83. Similarly, the risk taking and job dissatisfaction
were treated through factor loadings and item correlations. Two items for risk taking and
one item for job dissatisfaction were discarded. Finally, risk taking and job dissatisfaction
were measured by 9 and 12 questions, respectively. The Cronbach’s alphas were 0.82 and
0.83, respectively.
3.4. Analysis
The analysis of this study was conducted into three phases. First, a high–low plots of the
behavioral variables were done using SPSS 10 (SPSS, 1999) to explore the variations in
responses to the diVerent questionnaires by the accident and non-accident group of work-
ers. These plots depict the major weak links of the two groups of workers for their safety in
mines. Second, a t-test was conducted to see whether the variations in their responses are
random or there are signiWcant diVerences between the two groups. Finally, how these
diVerences can cause accident/injury in mines was evaluated through structural equation
modeling. The unique features of the structural equation modeling is that it can portray the
complex structural framework of the behavioral safety system, analyze them simulta-
neously, and evaluate the pattern and strength of relationships of the factors with work
injuries.
4. Result
High–low plots show how several values of one variable relate to one value of another
variable. Typically, each variable value on the horizontal axis has several corresponding
values on the vertical axis. High–low plots of risk taking, negative aVectivity, job dissatis-
faction, and safety performance have been shown in Fig. 2. X-axis represents the raw score
of all the respondents for speciWc variable and Y-axis represents the number of respondents
from accident group (AG) and non-accident group (NAG). For example, the minimum
and maximum score of the variable risk taking are 9 and 27 and it is seen from the plot that
AG workers are mostly concentrated at high scoring zone and NAG workers are mostly
concentrated at low scoring zone. Similarly, for the variables negative aVectivity and job
dissatisfaction, it is also seen from the plots that AG and NAG workers are mostly concen-
trated at high and low scoring zones, respectively. It implies that most of the accident
group (AG) of workers are negatively aVected, dissatisWed, and more risk taking than non-
accident group (NAG) of workers. The high–low plot of safety performance rating shows
that NAG workers are exhibiting better safety performance than the AG workers. The
means and mean diVerences of the variables for the AG and NAG workers have been
P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471 461
14
12
12
10
Number of Respondents
Number of Respondents
10
8
8
6
6
4
4
2 2
0 0
9 11 13 15 17 19 21 23 25 27 12. 14 16 18 20 22 24 26 28 30 32 34 36
16 14
14 12
12
Number of Respondents
10
Number of Workers
10
8
8
6
6
4
4
2
2
0
0
10 27 35 41 47 52 56 60 64 70 85
11 13 15 17 19 21 23 25 27 29 31
located on their respective high–low plots. Whether the mean diVerences of the respective
variables are statistically signiWcant or not were done through t-test.
4.2. T-test
The Statistical Package for Social Science (SPSS, 1999) was used for performing t-test of
raw scores for each factor. The t-values, standard deviations, and 2-tailed signiWcances to
test the equality of factor means of diVerent variables for the NAG and AG workers are
shown in Table 2. It is seen from Table 2 that all behavioral variables namely risk taking,
negative aVectivity, and job dissatisfaction emerged as signiWcant factors. The null hypoth-
esis for the equality of factors means is rejected in each case, which indicates that mean
scores of the NAG and AG workers are signiWcantly diVerent for the four behavioral fac-
tors. This suggests that human behavior plays a signiWcant role in accident involvement.
462 P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471
Table 2
T-test for equality of factor means
Variables Mean Mean SD t-Value 2-Tail
diVerence signiWcance
NAG AG NAG AG
Age 35.39 39.28 ¡3.89 8.04 9.52 ¡3.82 0.000
Experience 12.75 16.41 ¡3.67 8.43 9.68 ¡3.50 0.001
Job dissatisfaction 22.04 25.05 ¡3.01 6.06 5.90 ¡4.35 0.000
Negative aVectivity 19.11 22.65 ¡3.54 5.71 5.67 ¡3.54 0.000
Risk taking 17.08 20.09 ¡3.01 5.08 5.68 ¡5.00 0.000
Safety performance rating 58.23 37.45 20.78 15.45 18.06 10.71 0.000
Age and experience are also signiWcantly diVerent between NAG and AG workers. The
factor safety performance of the workers as administered by their respective supervisors
reveals that safety performance of the AG workers is signiWcantly poorer than that of
NAG workers. T-test although implies that there are signiWcant diVerences in the mean
scores of all the factors between AG and NAG workers, however, it is not possible to say
how these diVerences would cause an accident or injury in mines in a multivariate situa-
tion. The multivariate inXuences of the behavioral factors on accident/injury causation
were explored through structural equation modeling.
The Linear Structural Relations computer program (LISREL 8.30, 1998) developed by
Joreskog and Sorbom was used to estimate the structural relations as shown in Fig. 1. The
input data for the structural model run was the correlation matrix of seven variables col-
lected from case study mines and are presented in Table 3. Hypotheses H1, H2, and H3
were investigated using structural equation modeling developed by Joreskog and Sorbom
(LISREL 8.30, 1998). As suggested by Mulaik and James (1995), a series of nested models
were estimated before testing the hypothesized model. As expected, the results of all the
nested model analyses indicated that the models did not Wt well or the signiWcant path rela-
tionship between the latent constructs had not achieved. The hypothesized structural
model was tested next by Wxing the paths as depicted in Fig. 1.
Table 3
Descriptive statistics and correlations amongst constructs
Construct Mean SD Correlations
1 2 3 4 5 6 7
Age 37.34 9.01 1.00
Experience 14.58 9.25 0.92¤¤ 1.00
Safety performance 49.55 20.05 ¡0.10 ¡0.10 1.00
Risk taking 18.58 5.41 0.01 0.03 ¡0.14¤¤ 1.00
Negative aVectivity 20.94 5.91 ¡0.11¤ ¡0.08 ¡0.22¤¤ 0.59¤¤ 1.00
Job dissatisfaction 23.54 6.16 0.01 0.03 ¡0.11¤ 0.52¤¤ 0.48¤¤ 1.00
Work injury 0.50 0.50 0.27¤¤ 0.25¤¤ ¡0.76¤¤ 0.35¤¤ 0.37¤¤ 0.31¤¤ 1.00
¤
P < 0.05.
¤¤
P < 0.01.
P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471 463
0.93*
-0.11
Age
Safety
-0.02 Performance
-0.24*
Experience 0.28* -0.71*
-0.07
Work
0.10* Injury 0.27*
Negative 0.12*
Affectivity 0.13*
-0.01
0.03
0.03
Job 0.45*
Risk
Dissatisfaction 0.30*
Taking
0.58*
¤
Fig. 3. Final accident path model ( indicates the signiWcant paths at 0.05 probability level).
The proposed structural model yielded very good Wt to the data and signiWcant path
relationships between the latent constructs were achieved. The structural analysis of the
proposed accident path model with their signiWcant paths is shown in Fig. 3. The following
sections provide the overall Wt and results of the Wnal model run.
Table 4
Goodness of Wt indices for the accident path model
Chi-square with 1 degree of freedom 0.03 (P D 0.86)
Goodness of Wt index (GFI) 0.99
Adjusted goodness of Wt index (AGFI) 0.99
Root mean square residual (RMR) 0.002
Normed Wt index (NFI) 0.99
Relative Wt index (RFI) 0.99
Expected cross validation index (ECVI) 0.17
Square multiple correlations for
Safety performance 0.07
Work injury 0.73
Risk taking 0.42
Table 5
Total eVect of the signiWcant variables on work injury
Variables Direct Indirect Total Rank order
Age 0.28a 0.08 0.36a 2
Negative aVectivity 0.10a 0.23a 0.33a 3
Job dissatisfaction 0.12a 0.04a 0.16a 4
Risk taking 0.13a — 0.13a 5
Safety performance ¡0.71a ¡0.01 ¡0.72a 1
a
Indicates 0.01 probability level of signiWcance.
P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471 465
indirect eVect of negative aVectivity on work injury can be interpreted as follows. Negative
aVectivity makes a person more risk taking and risk taking behaviors are always injurious.
The linkage of negative aVectivity, safety performance and work injury reveals that nega-
tively aVected workers fail to exhibit proper safety performance. As a result, they are facing
more injuries. Similarly, the linkage of job dissatisfaction, risk taking and work injury
reveals that job dissatisWed persons exhibit more risk taking behaviors, which makes the
workers more injurious. It can be concluded from the result that all the three behavioral
variables are strong predictors of work injuries in mines.
5. Discussion
indicates that aged persons are more liable to be involved in accidents. Similar Wndings were
reported by Iverson and Erwin (1997), Bennett and Passmore (1986), and Durry (1965). As
older employees tend to have higher tenure, they are assigned greater responsibility at work
which places them at greater risk of injury (Iverson and Erwin, 1997). Further as suggested
by Maiti (1999), the old workers tend to be fatigue and sluggish due to their age, and their
reduced reXex action caused themselves to be more injury prone. The design of training pro-
grams should consider the age factor in targeting speciWc training. Interestingly, experience
does bear no relationships with work injury, risk taking and safety performance. Similar
Wndings were reported by Bennett and Passmore (1986), Phiri (1989), and Maiti (1999).
Experience makes workers to be well aware of the physical hazards of the mining environ-
ment and thereby experience is crucial to a full reorganization of the extent and range of
hazards (Theodore Barry and Associates, 1972; Kenny, 1993; Maiti, 1999). However, multi-
variate analysis showed that a less experienced worker is equally likely to be injured as an
experienced worker. It implies that experience through helps workers in understanding the
physical hazards, however, avoiding the imminent danger is much more behavioral.
The result also indicates that negatively aVected persons are more liable to be involved
in accidents. Considering negative aVectivity, research suggests that high negatively
aVected individuals suVer from attention lapse or states of distractibility on the job which
make them susceptible to accident (Hansen, 1989). Furthermore, studies have shown high
negatively aVected individuals are less inclined to seek to control their work environment
(Judge, 1993). Instead they display a preference for less direct coping strategies, such as
emotion-focused coping (Parkes, 1986, 1990) increasing their accident potential. Regarding
job dissatisfaction and work injuries, a signiWcant and positive relationship was observed
in this study. This Wnding is consistent with several other studies (Cooper and Sutherland,
1987; Holcom et al., 1993; Zwerling et al., 1996). It can be concluded that the negatively
aVected and dissatisWed persons have less control to their task and as a result they are
unable to cope up with proper safety practices. The positive relationships between risk tak-
ing and work injury imply that more risk taking employees are facing more injuries. This
relation is consistent with prior research (Cooper and Sutherland, 1987; Sutherland and
Cooper, 1991; Maiti et al., 2004). Further, risk taking behaviors of employees are positively
caused by negative aVectivity and job dissatisfaction. This implies that emotionally unsta-
ble dissatisWed people are more risk taking.
The variable safety performance shows strong negative relationship with work injury
and it has a negative direct relationship with negative aVectivity. However, safety perfor-
mance was poorly predicted by the variables considered in this study. This indicates the
inXuence of external factors. Evidence from organizational contribution of accidents sug-
gests that safety behavior does not necessarily fall within the control of the individual. For
example, Reason (1995, 1997) traced latent and direct contributions of organizations and
found a variety of organizational processes and individual behaviors, when combined with
other factors contributed towards drivers’ (rail) accident. Organizational processes, man-
agement decisions, workplace conditions, and individual’s momentary and long-term
capabilities can either add to or subtract from defenses that prevent an accident (Reason,
1995, 1997). Importance of supervision, training, and management workers’ relations on
safe work behavior is also noteworthy. EVective training, good supervision and healthy
management workers’ relations encourage employees to avoid at-risk behaviors and
thereby help in reducing workplace accidents. Therefore, given the importance of both
organizational and individual factors in inXuencing work injuries, it is suggested that while
P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471 467
cross-sectional correlation data does not allow for causal inferences. The second limitation
was that all of the reported accidents were treated into the same accident category ignoring
the type of accidents. The accident types, such as minor, reportable and serious accidents,
may be related to certain safety behavior and personality characteristics. Third, the data
collected in this study, which were based on the self-report of the participants, were in some
respect indirect and hence, they are partially subject to bias and measurement errors.
Finally, the interaction between behavioral and organizational/situational factors should
be considered for further studies.
Acknowledgements
The authors would like to express their thanks to the management of the case study
mines for their kind co-operation during the course of the study. The Wnancial help of Uni-
versity Grant Commission, New Delhi through a major research project is highly appreci-
ated. The valuable comments of the reviewers in enriching the paper are gratefully
acknowledged.
References
Bagozzi, R.P., 1988. On the use of structural equation models in experimental designs. Journal of Marketing
Research 26, 271–284.
Bennett, J.D., 1982. Relationship Between Workplace and Worker Characteristics and Severity of Injuries in U.S.
Underground Bituminous Coal Mines, 1975–81. Ph.D. Thesis, The Pennsylvania State University, University
Park, PA, 128 pp.
Bennett, J.D., Passmore, D.L., 1986. Multinomial logit analysis of injury severity in U.S. underground bituminous
coal mines. Accident Analysis and Prevention 17 (5), 399–408.
Bhattacharjee, A., 1991. Mine safety management: an application of risk analysis, forecasting techniques, and
Markov process to injury experience data. Ph.D. Thesis, The Pennsylvania State University, University Park,
PA, 279 pp.
Brown, K.A., Willis, P.G., Prussia, G.E., 2000. Predicting safe employee behavior in the steel industry: develop-
ment and test of a socio-technical model. Journal of Operations Management 18 (4), 445–465.
Caird, J., Kline, T., 2004. The relationships between organizational and individual variables to on-the-job driver
accidents and accident free kilometers. Ergonomics 47 (15), 598–613.
Cooper, C.L., Sutherland, V.J., 1987. Job stress, mental health, and accidents among oVshore workers in the oil
and gas extraction industries. Journal of Occupational of Medicine 29, 119–125.
Cooper, M., Philips, R., Sutherland, V., Makin, P., 1994. Reducing accidents using goal setting and feedback: a
Weld study. Journal of Occupational and Organisational Psychology 67, 219–240.
Cox, S., Cox, T., 1996. Safety, Systems and People. Butterworth, London.
Cox, S., Cheyne, A., 2000. Assessing Safety Culture in OVshore Environments. Safety Science 34, 111–129.
Cox, S., Jones, B., Rycraft, H., 2004. Behavioral approaches to safety management within UK reactor plants.
Safety Science 42, 825–839.
Dawson, S., William, P., Clinton, A., Bamford, M., 1991. Safety at Work: The Limits of Self-Regulations. Cam-
bridge University Press, Cambridge, England.
Demichiei, J., Langton, J., Bullock, K., Wiles, T., 1982. Factors associated with disabling injuries in underground
coal mines. MSHA, 72 pp.
Dhar, B.B., Singh, A.P., Ratan, S., 1997. Socio-psychological correlation of coal mine accidents. In: Proceedings of
the 27th International Conference of Safety in Mines Research Institutes, New Delhi, February 20–22, vol. 2,
pp. 1239–1252.
Digman, J.M., 1990. Personality Strinctme: ///emergence of the Wve-factor model. In: Rosenzwing, M.R., Porter,
L.W. (Eds.), Annual Review of Psychology, vol. 41, pp. 417–440.
Durry, D.M., 1965. A study on the literature on accidents in coal mines of the united states with comparisons of
the records in other coal-producing countries. Unpublished Doctoral Dissertation, University of Indiana.
P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471 469
Embrey, D.E., 1992. Incorporating management and organizational factors into probabilistic safety assessment.
Reliability Engineering and System Safety 38, 199–208.
Frone, R.M., 1998. Predictors of work injuries among employed adolescents. Journal of Applied Psychology 83
(4), 565–576.
Gaertner, G.H., Newman, P.D., Perry, M.S., Fisher, G.P., Whitehead, K., 1987. Determining the EVects of Man-
agement Practices on Coal Mines Safety. BuMines OFR 39-88, 348 pp., NTIS PB 88-221445.
Greenwood, M., Woods, H.M., 1919. A report on the incidence of industrial accidents upon individuals with spe-
cial reference to multiple accidents. In: Haddon, W., Suchman, E.A., Klein, D. (Eds.), Accident Proneness.
Harper and Row, Newyork.
Hair, J.F., Anderson, R.E., Tatham, R.L., Black, W.C., 1995. Multivariate Data Analysis with Readings. Prentice-
Hall, Englewood CliVs, NJ. 745 pp.
Hansen, C.P., 1989. A causal model of the relationship among accidents, biodata, personality and cognitive fac-
tors. Journal of Applied Psychology 74, 81–90.
Hofmann, D.A., Stetzer, A., 1996. A cross-level investigation of factors inXuencing unsafe behaviors and acci-
dents. Personnel Psychology 49 (2).
Holcom, M.L., Lehman, W.E.K., Simpson, D.D., 1993. Employee accidents: inXuence of personal characteristics,
job characteristics, and substance use in jobs diVering in accident potential. Journal of Safety Research 24,
205–221.
Holnagel, E., 1993. Human Reliability Analysis: Context and Control. Harcourt Brace, London.
Hurst, N.W., Bellamy, L.J., Geyer, T.A.W., Astley, J.A., 1991. A classiWcation scheme for pipe work failures to
include human and socio-technical errors and their contribution to pipe work failure frequencies. Journal of
Hazardous Materials 26, 159–186.
International Survey Research Corporation. Responses on Dimensions of Job Satisfaction, Employee Satisfac-
tion Surveys, Chicago, 1992, pp. 102.
Iverson, R.D., Erwin, P.J., 1997. Predicting occupational injury: the role of aVectivity. Journal of Occupational
and Organisational Psychology 70, 113–128.
Joreskog, K.G., Sorbom, D., 1989. LISREL 7: Users reference guide. ScientiWc Software, Inc., 1369 Neitzel Road,
Mooresvile, IN 16158, USA.
Joreskog, K.G., Sorbom, D., 1998. LISREL 8.30: Users reference guide. ScientiWc Software, Inc., 1369 Neitzel
Road, Mooresvile, IN 16158, USA.
Judge, T.A., 1993. Does aVective disposition moderate the relationship between job satisfaction and voluntary
turnover? Journal of Applied Psychology 78, 395–401.
Kamp, J., Krause, T.R., 1997. Selecting safe employees: a behavioral science perspective. Professional Safety 42
(4), 24–28.
Kenny, J.M., 1993. Haulage related accident in metal and non-metal surface mines. USBM IC 9365, 6 pp.
Komaki, J., Barwick, K.D., Scott, L.R., 1978. A behavioral approach to occupational safety: pinpointing and rein-
forcing safe performance in a food manufacturing plant. Journal of Applied Psychology 6, 434–445.
Kothari, C.R., 2001, Research Methodology – Methods and Techniques second ed. Wishwa Prakashan, Kolkata.
468 pp.
Krause, T.R., 1995. Employee-Driven Systems for Safe Behaviour. Van Nostrand Reinhold, New York.
Krause, T.R., Russell, L.R., 1994. The behavior-based approach to proactive accident investigation. Professional
Safety 39 (3).
Krause, T.R., Hidley, J.H., Hodson, S.J., 1990. The Behaviour-Based Safety Process: Managing Involvement for
an Injury-Free Culture. Van Nostrand Reinhold, New York.
Krause, T.R., Seymour, K.J., Sloat, K.C.M., 1999. Long term evaluation of a behaviour-based method for improv-
ing safety performance: a meta analysis of 73 interrupted time-series replications. Safety Science 32, 1–18.
Maiti, J., 1999. An investigation of multivariate statistical models to evaluate mine safety performance. unpub-
lished PhD Thesis, Department of Mining Engineering, Indian Institute of Technology, Kharagpur, 217 pp.
Maiti, J., Bhattacharjee, A., 1999. Evaluation of risk of occupational injuries among underground coal mine
workers through multinomial logit analysis. Journal of Safety Research 30 (2), 93–101. Elsevier Science.
Maiti, J., Bhattacharjee, A., 2000. A causal model for evaluation of mine safety. Transaction of the Institute of
Mining and Metallurgy (Section A – Mining Technology) 109, A55–A59.
Maiti, J., Bhattacharjee, A., 2001. Predicting accident susceptibility: a logistic regression analysis of underground
coal mine workers. Journal of the South African Institute and Metallurgy 101 (4), 203–208.
Maiti, J., Dasgupta, S., 2003. Mine safety management – recent trend in accident analysis and control. Accepted
for 19th World Mining Congress, 1–5 November, 2003, New Delhi.
470 P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471
Maiti, J., Bhattacharjee, A., Bangdiwala, S.I., 2001. Loglinear model for analysis of crosstabulated coal mine
injury data. Injury Control and Safety Promotion 8 (4), 229–236.
Maiti, J., Chatterjee, S., Bangdiwala, S.I., 2004. Determinant of work injury – an application of structural equation
modeling. Injury Control and Safety Promotion 11 (1), 29–37.
Maruyama, G., McGarvey, B., 1980. Evaluating gausal models: an application of maximum likelihood analysis of
structural equations. Psychological Bulletin 87 (3), 502–512.
Mines Occupational Safety and Health Advisory Board (MOSHAB) 1998. Risk Taking Behaviour in the Western
Australian Underground Mining Sector, Risk Taking Behaviour Working Party Report and Recommenda-
tions, 75pp.
Mulaik, A.M., James, L.R., 1995. Objectivity and reasoning in science and structural equation modeling. In:
Hoyle, R.H. (Ed.), Structural Equation Modeling.
National Academy of Sciences – National Research Council, 1982. Towards Safer Underground Coal Mines.
NAS, 190 pp.
Oi, W.Y., 1974. Economic and Empirical Aspects of Underground Safety. Rochester, New York: Graduate
School of Management, University of Rochester, February, 1974, NTIS PB 254 736.
Parkes, K.R., 1986. Coping in stressful episodes: the role of individual diVerences, environmental factors and situ-
ational characteristics. Journal of Personality and Social Psychology 51, 1277–1292.
Parkes, K.R., 1990. Coping, negative aVectivity, and the work environment; additive and interactive predictors of
mental health. Journal of Applied Psychology 75, 399–409.
Pfeifer, C., Stefanski, J., Grether, C., 1976. Psychological, Behavioural and Organisational Factors AVecting Coal
Miner Safety and Health. DHEW Contract HSM 99-72-151, DHEW, NTIS PB 275 599, 319 pp.
PXeider, E.P., Krug, A.D., 1973. The Development of Health and Safety Indices for the Evaluation of Under-
ground Coal Mining Systems. Report on Contract H0122028 with U.S. Bureau of Mines, The University of
Minnesota, Department of Civil and Mineral Engineering, Minneapolis, Minnesota, NTIS, PB 235-126, 200 pp.
Phiri, J., 1989. Development of Statistical Indices for the Evaluation of Hazards in Longwall Face Operations.
Ph.D. Thesis, The Pennsylvania State University, University Park, PA, 167 pp.
Quinn, R.P., Staines, G.L., 1979. The 1977 Quality of Employment Survey. University of Michigan, Institute for
Social Research, Ann Arbor.
Rahman, S., 2000. Towards Developing an International Market Selection Decision Framework, ANZMAC
2000 Visionary Marketing for the 21st Century: Facing the Challenge.
Ray, S.C., Bhattacharjee, A.B., 1997. The eVect of human behaviour on mine safety. In: Proceeding of the 27th
International Conference of Safety in Mines Research Institute, 20–22 February 1997, New Delhi, pp. 133–
142.
Reason, J.T., 1995. A system approach to organizational error. Ergonomics 34, 1708–1721.
Reason, J.T., 1997.Managing the Risks of Organizational Accidents Ashgate, Aldershot.
Rhoton, W., 1980. A procedure to improve compliance with coal mines safety regulations. Journal of Organisa-
tion and Behavioural Management 2, 243–249.
Robin, S.P., 1997. Organisational Behaviour – Concepts, Controversies and Applications, seventh ed. Prentica
Hall of India Pvt. Ltd, New Delhi. 751pp.
Root, N., 1981. Injuries at work are fewer among older employees. Monthly labour Review 104 (3), 30–39.
Root, N., Hoefer, M., 1979. The Wrst work-injury data available from new BLS study. Monthly Labour Review
104 (3), 30–39.
Savery, L.K., Wooden, M., 1994. The relative inXuence of life events and hassles on work-related injuries: some
Australian evidence. Human Relations 47, 283–305.
Shafai-Sahrai, Y., 1973. Determinant of Occupational Injury Experience. Michigan State University, East Lan-
sing, MI.
Shaw, B.E., Sanders, M.S., Peay, J., 1989. Research to determine the contribution of system factors including
human error in the occurrence of underground injury accidents. In: Proceedings XVI Training Resources
Applied to Mining, The Pennsylvania State University, University Park, August 14–16, 9 pp.
Shrivastava’s Bhopal, 1987.Anatomy of Crisis Balliger, Cambridge.
Simard, M., Marchand, A., 1995. A multilevel analysis of organizational factors related to the taking of safety ini-
tiatives by work groups. Safety Science 21, 113–129.
Simson, G.C., 1994. Promoting safety improvements via potential human error audits. The Mining Engineer 154
(395), 38–42.
Souder, W.E., 1988. A Catastrophe-Theory Model for Simulating Behavioural Accidents. United States Bureau of
Mines, IC 9178, 19 pp.
P.S. Paul, J. Maiti / Safety Science 45 (2007) 449–471 471
SPSS (Statistical Package for Social Sciences) Inc. 1999. 444 N, Michigan Avenue, Chicago, Illinois, 60611.
Sutherland, V.L., Cooper, C.L., 1991. Personality, stress and accident involvement in the oVshore oil and gas
industry. Personality and Industrial DiVerences 12, 195–204.
Theodore Barry and Associates, 1972. Fatality Analysis Database Development. USBM OFR 22-73, NTIS PB
219 250.
Turner, B.A., 1978. Man Made Disasters. Macmillan, London.
Wagenaar, W.A., Groeneweg, J., 1987. Accidents at sea: multiple causes and impossible consequences. Interna-
tional Journal of Man–Machine Studies 27, 587–598.
Watson, D., Clarke, L.A., Tellegen, A., 1988. Development and validation of brief measures of positive and nega-
tive aVect: the Panas scales. Journal of Personality and Social Psychology 54, 1063–1070.
Wright, C., 1986. Routine deaths: fatal accidents in the oil industry. Sociological Review 4, 265–289.
Zwerling, C., Sprince, N.L., Wallace, R.B., Davis, C.S., Witten, P.S., Heeringa, S.G., 1996. Risk factors for occupa-
tional injuries among older workers: an analysis of the health and retirement study. American Journal of Pub-
lic Health 86, 1306–1309.