Sunteți pe pagina 1din 13

See

discussions, stats, and author profiles for this publication at:


https://www.researchgate.net/publication/43488437

Perceptions of safety at work: A


framework for linking safety climate
to safety performance, knowledge,
and motivation

Article in Journal of Occupational Health Psychology · August 2000


DOI: 10.1037//1076-8998.5.3.347 · Source: OAI

CITATIONS READS

565 7,052

2 authors:

Mark Griffin Andrew Neal


University of Western Australia University of Queensland
103 PUBLICATIONS 5,026 CITATIONS 93 PUBLICATIONS 3,688 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Centre for Transformative Work Design View project

Work and Personality development View project

All content following this page was uploaded by Andrew Neal on 23 June 2017.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original docum
and are linked to publications on ResearchGate, letting you access and read them immediately.
Journal of Occupational Health Psychology Copyright 2000 by the Educational Publishing Foundation
2000, Vol. 5, No. 3, 347-358 1076-8998/00/$5.00 DOI: 10.1037fl1076-8998.5.3.347

Perceptions of Safety at Work: A Framework for Linking Safety


Climate to Safety Performance, Knowledge, and Motivation
Mark A. Griffin Andrew Neal
Queensland University of Technology University of Queensland

Research in the areas of organizational climate and work performance was used to develop a
framework for measuring perceptions of safety at work. The framework distinguished perceptions
of the work environment from perceptions of performance related to safety. Two studies
supported application of the framework to employee perceptions of safety in the workplace.
Safety compliance and safety participation were distinguished as separate components of safety-
related performance. Perceptions of knowledge about safety and motivation to perform safely
influenced individual reports of safety performance and also mediated the link between safety
climate and safety performance. Specific dimensions of safety climate were identified and
constituted a higher order safety climate factor. The results support conceptualizing safety climate
as an antecedent to safety performance in organizations.

The present study combines theories of individual safety, together with factors such as work pressure
performance (Borman & Motowidlo, 1993; Camp- and communication practices (Hofmann & Stetzer,
bell, McCloy, Oppler, & Sager, 1993) with theories 1996). A key assumption of much of this literature is
of organizational climate (L. A. James & James, that the relationship between safety climate and sys-
1989; L. R. James & Mclntyre, 1996) to develop a tem safety is at least partially mediated by individual
framework for investigating perceptions of safety in safety behavior. However, few studies have elabo-
organizations. This framework provides a link be- rated the mechanisms through which organizational
tween perceptions of the work environment and in- factors influence individual safety behavior at work.
dividual behavior within the work environment. We Because employee perceptions are central to the
differentiate perceptions of the work environment, measurement of climate, this article develops a
perceptions of individual behavior in the workplace, framework for measuring employee perceptions of
and mediational processes thought to underlie indi- safety-related factors in the work environment. We
vidual work performance. These distinctions are im- first discuss safety climate in terms of a higher order
portant because they allow systematic assessment of factor similar to other types of climate in organiza-
conceptually distinct perceptions that may have dif- tions. We next discuss employee safety performance
ferent causes and consequences within organizations. as an aspect of work performance. We then propose
Recent reviews of the safety literature emphasize a model that links safety climate perceptions to indi-
the influence of organizational factors on measures of vidual safety performance and describe two studies
system safety such as accidents and near misses of employee perceptions that test aspects of the
(Hofmann, Jacobs, & Landy, 1995; Hofmann & model.
Stetzer, 1999). Safety climate is an organizational
factor commonly cited as an antecedent of system
Safety C l i m a t e

Mark A. Griffin, School of Management~ Queensland Safety climate is one type of climate that can be
University of Technology, Brisbane, Queensland, Australia; experienced by individuals in organizations. Other
Andrew Neal, School of Psychology, University of Queens- climate types that have been identified include cli-
land, Brisbane, Queensland, Australia. mate for customer service (Burke, Borucki, & Hur-
This research was supported by the Australian Research
Council (Grant No. A79801483). We are grateful for the ley, 1992; Schneider, Wheeler, & Cox, 1992) and
assistance provided by Heidi Bushell and Adrian Lewis in climate for innovation (N. R. Anderson & West,
collecting the data for the study. 1998; Klein & Sporra, 1996). All types of climate are
Correspondence concerning this article should be ad- based on individuals' perceptions of the practices,
dressed to Mark A. Griffin, School of Management,
Queensland University of Technology, Brisbane 4001, procedures, and rewards in the organization (Schnei-
Queensland, Australia. Electronic mail may be sent to der, 1990).
m.griffin @qut.edu.au. Perceptions of climate also appear to be hierarchi-

347
348 GRIFFIN AND NEAL

cally structured. Individuals evaluate specific fea- reduced version of Zohar's (1980) measure with a
tures of their environment in terms of their personal U.S. sample, identified three factors: management
values and the significance of these features for their concern, management action, and physical risk. De-
overall well-being. L. A. James and James (1989), for dobbeleer and Beland (1991) identified only two di-
example, demonstrated that a number of climate di- mensions of safety climate: management commit-
mensions loaded onto a common higher order factor. ment to safety and worker involvement in safety
This higher order factor reflected "the extent to which activities. Other studies have obtained a range of
the environment is personally beneficial versus per- factor solutions, incorporating constructs such as in-
sonally detrimental (damaging or painful) to one's dividual attitudes to safety and the physical work
sense of well-being" (L. R. James, James, & Ashe, environment (Cox & Tait, 1991; Hayes, Perander,
1990, p. 53). Smecko, & Trask, 1998; Niskanen, 1994). The re-
Conceptualizations of both general organizational sults of these studies using different measurement
climate and specific types of climate (e.g., customer instruments suggest that it is premature to propose a
service) provide a basis for defining the key features definitive structure for the first-order dimensions of
of safety climate. Specifically, safety climate should safety climate.
be conceptualized as a higher order factor comprised As in other areas of organizational assessment, the
of more specific first-order factors. The first-order purpose of the assessment should determine whether
factors of safety climate should reflect perceptions of specific first-order factors or a global higher order
safety-related policies, procedures, and rewards. The factor is most appropriate (Hogan & Roberts, 1996).
higher order factor of safety climate should reflect the For some purposes, such as determining the overall
extent to which employees believe that safety is impact of safety climate on safety outcomes, a higher
valued within the organization. order factor of safety climate will be most appropri-
The preceding arguments suggest some conditions ate. For other purposes, such as determining the
for assessing perceptions of safety climate. Most impact of distinct organizational practices on task
importantly, only those perceptions that involve in- performance, using specific first-order climate factors
dividuals' assessment of workplace attributes con- will provide more detailed diagnostic information. In
cerned with safety are considered to be perceptions of the present article, we developed a model that distin-
safety climate. For example, employee views about guished perceptions of safety climate from other per-
management values for safety and personnel policies ceptions related to safety. Therefore, the article fo-
about safety are clearly perceptions about values and cuses on the role of a higher order safety climate
procedures in the wider work environment. In con- factor rather than specific first-order factors.
trast, ratings of risk in the environment, affective
reactions to safety issues, normative beliefs about Safety Performance
safety, and self-reports of safety behavior are not
considered to be perceptions of climate. Perceptions of safety climate should be distin-
There is currently no clear agreement about the guished from perceptions of individual knowledge,
specific first-order factors that constitute the higher motivation, and behavior that influence safety in the
order safety climate factor. One factor that has often workplace. Work behaviors relevant to safety can be
been included in safety climate assessment is man- conceptualized in the same way as other work behav-
agement values for safety. Management values have iors that constitute work performance. In this way,
been measured in terms of management concern for models of performance can be applied to safety per-
employee well-being (R. L. Brown & Holmes, 1986), formance in the workplace. The model of perfor-
management attitudes toward safety (Dedobbeleer & mance used in the present study makes a distinction
Beland, 1991; Niskanen, 1994; Zohar, 1980), percep- among the components of performance, the determi-
tions that safety is important to management (De Joy, nants of performance, and the antecedents of perfor-
1994), and production and safety trade-offs (Hof- mance (Campbell, Gasser, & Oswald, 1996).
mann & Stetzer, 1996). However, there is variation in The components of performance describe the ac-
other factors that have been assessed as part of safety tual behaviors that individuals perform at work. Bor-
climate. Zohar (1980), for example, identified eight man and Motowidlo (1993) proposed two major
dimensions of safety climate, including factors such components of performance: task performance and
as management attitudes, effects of safe conduct on contextual performance. These two components of
promotion, work pace, and status of safety officers. performance can be used to differentiate safety be-
R. L. Brown and Holmes (1986), however, using a haviors in the workplace. First, based on definitions
PERCEPTIONS OF SAFETY 349

of task performance, we use the term safety compli- ganization can be an important antecedent of perfor-
ance to describe the core safety activities that need to mance. At least part of this relationship appears to be
be carried out by individuals to maintain workplace mediated by motivation. Individuals attribute mean-
safety. These behaviors include adhering to tagout ing and value to features of their work environment
and lockout procedures and wearing personal protec- (L. A. James & James, 1989). This process is thought
tive equipment. Second, based on definitions of con- to influence motivation and, subsequently, perfor-
textual performance, we use the term safety partici- mance. Supporting this proposition, S. P. Brown and
pation to describe behaviors such as participating in Leigh (1996) found that motivation mediated the link
voluntary safety activities or attending safety meet- between organizational climate and task performance
ings. These behaviors may not directly contribute to for sales people. It has also been proposed that the
workplace safety, but they do help to develop an relationship between climate and performance can be
environment that supports safety. mediated by knowledge. Organizational climate can
The determinants of performance represent the influence knowledge by increasing participation in
proximal causes of variability in performance. These activities such as training (Morrison, Upton, & Cord-
are the factors that are directly responsible for indi- ery, 1997). These findings suggest that safety climate
vidual differences in task and contextual perfor- should be classified as an antecedent of safety per-
mance. Campbell et al. (1993) argued that there are formance and that the relationship between safety
only three determinants of individual performance: climate and safety performance may be mediated by
knowledge, skill, and motivation. Safety perfor- determinants of safety performance, such as safety
mance, therefore, must be determined by knowledge motivation and safety knowledge.
and skills necessary for particular behaviors and by
the motivation of individuals to perform the behaviors. A Model of Safety Climate and Performance
The antecedents of performance represent the dis-
tal causes of variability in performance. These factors Figure 1 provides a schematic representation of the
influence performance through their effects on proposed link between safety climate and safety per-
knowledge, skill, and motivation. For work perfor- formance based on previous research within the work
mance in general, antecedents have been identified at performance literature. The model shows that safety
both the individual and the organizational level. At climate is a higher order factor comprised of more
the individual level, research suggests that factors specific first-order factors and that the influence of
such as ability and experience are important anteced- safety climate on safety performance is mediated by
ents of task performance, whereas personality con- knowledge, skill, and motivation.
structs, such as conscientiousness, are important an- Existing models of safety climate do not make the
tecedents of contextual performance (McHenry, distinctions contained in the proposed framework and
Hough, Toquam, Hanson, & Ashworth, 1990; Moto- sometimes include determinants or components of
widlo & Van Scotter, 1994; Wise, McHenry, & safety performance as elements of safety climate. For
Campbell, 1990). example, worker involvement in safety is commonly
At the organizational level, the climate of an or- cited as an important element of safety climate (De-

Antecedents of Determinants of Components


Safety Performance Safety Performance Safety Performance

no w, Ooe O 'as e.o=n e


~ y Contextual P e r f o ~ , )

Figure 1. Summary of relationship among antecedents, determinants,and componentsof


safety performance.
350 GRIFFIN AND NEAL

dobbeleer & Beiand, 1991; Rundmo, 1994). How- sample of 1,264 employees for the analyses. The data were
ever, worker involvement refers to the behavior of archival records from an organization conducting quality
assurance audits for the mining and manufacturing industry
individuals and should be classified as a component
in Australia. Because the data were obtained from archival
of safety performance. Similarly, the value that indi- records, response rates for the survey were not available.
viduals personally place on safety (e.g., Cox & Tait, Measures. A questionnaire consisting of 81 items was
1991) is an individual motivational construct rather used to assess a wide range of attitudes, behaviors, and
than an aspect of safety climate. safety procedures. Employees responded on a 5-point scale
to all items ranging from 1 (strongly agree) to 5 (strongly
disagree). Items from the survey were matched to constructs
The Present Research in the conceptual model by three raters. Items were retained
if they were consistently classified by all raters as percep-
tions of the workplace (12 items), self-reports of safety
This article reports two separate studies of safety
knowledge (7 items), or self-reports of individual behavior
climate conducted across a range of manufacturing at work (6 items).
and mining organizations. The overall goal of the Safety climate was assessed by four subscales. Manager
research was to assess the applicability of the pro- Values were assessed by three items that asked about the
posed model to employee perceptions of safety in degree to which managers valued safety in the workplace
(ct = .72). An example item was "I think management is
their workplace. Support for the model of anteced- sincere in its efforts to ensure employee safety." Safety
ents, determinants, and components of performance Inspections were assessed by three items that asked about
would provide a basis for systematic evaluation of the effectiveness of safety inspections in the organization
safety perceptions at work and for linking safety (a = .69). An example item was "Inspections in this com-
perceptions to safety outcomes in organizations. pany are effective at identifying unsafe conditions." Person-
nel Training was assessed by two items that asked about the
Study 1 was based on an organizational survey degree to which staff were trained in safety procedures (a =
conducted in seven large manufacturing and mining .73). An example item was "Trained people are available to
organizations in Australia. The survey assessed per- handle emergencies." Safety Communication was assessed
ceptions of a range of safety conditions and practices by four items that asked about the way in which safety
in each organization. However, the first study was issues were communicated (a = .74). An example item was
"I receive regular updates on technical aspects of safety."
conducted using archival survey data. Safety knowl- Safety Knowledge was assessed by seven items that
edge was the only determinant of performance that asked employees to rate their knowledge about safety prac-
was included in this archival data set. Study 1, there- tices and procedures (a = .73). An example item was "I
fore, tested a model in which knowledge was the only understand the health and safety regulations relating to
my work."
mediator between safety climate and safety perfor-
Two dimensions of performance in the workplace were
mance. Study 2 was conducted in three manufactur- measured. Safety Compliance was assessed by three items
ing organizations and was based on a revised version that asked about individual performance of safety compli-
of the instrument used in the first study. The revision ance (a = .56). An example item was "I use the correct
allowed measurement of a greater number of safety personal protective equipment for the task I am doing."
Safety Participation was assessed by three items that asked
climate subdimensions and included knowledge and
about participation that supported safety in the organization
motivation as mediators of the link between safety but did not necessarily involve performance related to safety
climate and safety perception. A specific measure of (ct = .66). An example item was "I often take part in
skill was not developed in either of the studies. development of the safety requirements for my job."
The first specific goal of the research was to test Correlations among all the scales are reported below the
diagonal in Table 1.
whether employee perceptions of safety at work Analysis. Analyses were conducted using confirmatory
could be differentiated in terms of the components, factor analysis (CFA) and structural equation modeling
determinants, and antecedents of safety performance. (SEM). The CFA included a higher order factor analysis in
The second goal of the studies was to test the medi- which the first-order climate factors were indicators of a
ational role of knowledge and motivation described higher order safety climate factor. The SEM analysis as-
sessed the proposed mediational model based on the higher
in the proposed model. order factor model. Goodness of fit of the various models
was judged in terms of Joreskog and Sorbom's (1989)
goodness-if-fit index (GFI), the nonnormed fit index (NNFI;
Study 1 Bentler & Bonneu, 1980), the comparative fit index (CFI;
Bentler, 1990), and root mean square error of approximation
Me~od (RMSEA; Hu & Bentler, 1999). The CFI is recommended
as the best approximation of the population value for a
Sample. The sample for Study 1 consisted of 1,403 single model (Medsker, Williams, & Holahan, 1994). The
employees in seven Australian manufacturing and mining NNFI is recommended as an index that takes account of the
organizations. Listwise deletion of cases resulted in a final parsimony of the estimated parameters (Medsker et al.,
PERCEPTIONS OF SAFETY 351

Table 1
Means, Standard Deviations, Correlations, and Reliabilities for All Scales in Study 1
Scale M SD 1 2 3 4 5 6 7
1. Manager Values 2.67 0.88 -- .78 .47 .76 .52 .60 .61
2. Safety Inspections 2.87 0.85 .57 -- .46 .72 .45 .55 .60
3. Personnel Training 2.46 0.88 .36 .33 -- .49 .39 .46 .36
4. Safety Communication 3.00 0.81 .56 .54 .37 -- .59 .68 .83
5. Safety Knowledge 2.26 0.55 .44 .37 .36 .50 -- .60 .53
6. Safety Compliance 2.62 0.60 .31 .30 .26 .36 .38 -- .71
7. Safety Participation 2.90 0.77 .43 .43 .26 .57 .45 .38 --
Note. Values above the diagonal represent correlations among the first-order factors; values below the diagonal represent
correlations among the scales.

1994). The RMSEA has a known distribution and compen- nants, and components of safety performance. The
sates for model complexity (Hu & Bentler, 1999). results also support a distinction between safety com-
pliance and safety participation.
Results and Discussion Finally, a higher order factor analysis was con-
ducted with the first-order safety climate factors aci-
The first analysis tested the proposed factor struc- ing as indicators of a higher order safety climate
ture comprising safety knowledge, safety compli- factor. This model showed an acceptable fit to the
ance, safety participation, and a higher order safety data (Hu & Bentler, 1999), although there was a
climate factor with four subdimensions. A series of
significant decrease in the fit of this model compared
five CFA models were compared, and the results of
with the previous model in which the seven first-
each model are presented in Table 2. First, a null
order factors were free to correlate, A x 2 ( l l , N =
model that specified no covariation among items was
1,264) = 92.56, p < .001. Correlations among the
estimated. This null model provided a baseline for
first-order factors are shown above the diagonal in
comparison with subsequent models. Second, a one-
Table 1. Because the higher order model displayed an
factor model was estimated to test the possibility that
acceptable fit, subsequent tests of study hypotheses
all items loaded on a general factor regarding safety
were conducted using the higher order model.
perceptions. The one-factor model was a significantly
The factor loadings of the items are presented in
better fit than the null model, AX2(24, N = 1,264) =
Table 3. Most of the loadings were at acceptable
6,771.04, p < .001. Third, a three-factor model was
estimated that specified the components, determi- levels. However, the loadings for all of the safety
nants, and antecedents of performance depicted in compliance items were lower than expected (less
Figure 2 as three factors. This model allowed for the than .45), and one of the factor loadings for the
possibility that different dimensions were perceived knowledge items was also below .40. This item asked
by employees but that distinctions among specific about knowledge for using equipment safely,
subdimensions were not meaningful. The model whereas other items for this factor asked about un-
showed a significant improvement in fit, AX2(3, N = derstanding of safety requirements.
1,264) = 800.13, p < .001. Fourth, a four-factor The next analysis tested structural paths among
model that differentiated the two components of the constructs using the higher order safety climate
safety performance was estimated. This model also structure. The hypothetical model, in which knowl-
showed further significant improvement in fit, AX2(3, edge mediated the link between safety climate and
N = 1,264) = 61.30, p < .001. Fifth, a seven-factor performance, was compared with a saturated struc-
model that differentiated among all the proposed tural model in which safety climate predicted
subdimensions depicted in Figure 1 was estimated. safety knowledge, safety compliance, and safety
This model also showed a further significant im- participation. The saturated structural model has
provement in fit, AX2(15, N = 1,264) = 6,771.04, equivalent fit properties to the higher order mea-
p < .001. The results support the proposition that surement model and provided a point of compari-
individual employee perceptions of safety can be son for the hypothesized model. The lower section
differentiated in terms of the antecedents, determi- of Table 2 shows that the hypothesized model was
352 GRIFFIN AND NEAL

Table 2
Assessment of Measurement Model for Individual Responses to Survey Items in Study 1
Model df X2 GFI RMSEA CFI NNFI
Measurement models
Null covariance model 276 9,393.36 .40 .25
One-factor model 252 2,622.32 .85 .08 .74 .72
Three-factor model
One climate dimension 249 1,822.19 .89 .07 .83 .81
One knowledge dimension
One performance dimension
Four-factor model
One climate dimension 246 1,760.89 .90 .07 .83 .81
One knowledge dimension
Two performance dimensions
Seven-factor model
Four climate dimensions 231 1,017.95 .94 .05 .91 .90
One knowledge dimension
Two performance dimensions
Higher order factor model 242 1,110.51 .93 .05 .90 .89
Structural models
Hypothesized model 244 1,404.64 .92 .06 .87 .86
Final model 242 1,110.51 .93 .05 .90 .89
Note. GFI = goodness-of-fit index; RMSEA = root mean square error of approximation; CFI = comparative fit index;
NNFI = nonnormed fit index.

a significantly poorer fit than the saturated struc- The results of Study 1 supported the differenti-
tural model, AX2(2, N = 1,264) = 294.13, p < ation of the constructs in the proposed model. The
.001. Inspection of the path values in the saturated safety climate dimensions were distinguished from
structural model indicated that both the path from the other measures in the survey. However, knowl-
safety climate to safety compliance and the path edge about safety only partially mediated the link
from safety climate to safety participation were between safety climate and safety performance di-
necessary for the model. The path coefficients for mensions. Safety climate also displayed a direct
the final structural model are presented in Figure 2. positive relationship with safety compliance and
The figure also shows the loadings of the first- safety participation. The path from safety knowl-
order safety climate factors on the higher order edge to safety participation was also not statisti-
factor. cally significant.

57
v Compliance
.85 I .25
.81 Safety .62
.13

Figure 2. Path estimates from final model for Study t. ns = not statistically significant.
PERCEPTIONS OF SAFETY 353

Table 3
Factor Loadings for Items in Study 1
Factor loading
Factor 1 2 3 4 5
Manager Values .73 .66 .65
Safety Inspections .77 .66 .58
Personnel Training .82 .69
Safety Communication .69 .68 .62 .58
Safety Knowledge .73 .68 .67 .49 .36
Safety Compliance .44 .44 .44 .37
Safety Participation .71 .61 .56

Study 2 subscales. Manager Values were assessed by four items that


asked about the degree to which managers were perceived
T h e results o f S t u d y 1 partially supported the h y - to value safety in the workplace (a = .90). An example item
was "Management values safety." Safety Communication
p o t h e s i z e d m e d i a t i o n a l m o d e l o f safety perceptions.
was assessed by four items that asked about the way in
B e c a u s e m o t i v a t i o n w a s n o t i n c l u d e d in the first which safety issues were communicated (a = .80). An
study, it w a s n o t surprising that the full m e d i a t i o n a l example item was "Feedback is given on safety suggestions
m o d e l w a s n o t supported. T h e s e c o n d s t u d y included and complaints." Safety Practices were assessed by three
m o r e specific m e a s u r e s o f individual m o t i v a t i o n for items that asked about the thoroughness of safety practices,
b o t h safety c o m p l i a n c e a n d safety participation. T h e such as inspections and investigations (c~ = .84). An exam-
ple item was "Accident investigations are always conducted
m e a s u r e o f safety m o t i v a t i o n w a s i n c l u d e d as part o f
thoroughly." Personnel Training was assessed by four items
a m a j o r revision o f the organizational s u r v e y u s e d in
that asked about the effectiveness of training for safety in
Study 1. the organization (a = .74). An example item was "This
T h e revision p r o c e s s resulted in a n u m b e r o f organization has a useful and well-developed safety training
c h a n g e s to the c o n t e n t o f the survey. In addition to program." Safety Equipment was assessed by three items
the m e a s u r e o f motivation, the r e v i s e d s u r v e y also that asked about the adequacy of safety equipment (o~ =
.75). An example item was "The correct tools and equip-
i n c l u d e d a m o r e c o m p r e h e n s i v e a s s e s s m e n t o f safety
ment are readily available."
climate. T h e scale c o n s t r u c t t e r m e d Safety I n s p e c -
Safety Knowledge was assessed by four items that asked
tions in Study 1 w a s e x p a n d e d to a c c o m m o d a t e other about each employee's knowledge of safety practices and
m a n a g e m e n t practices relating to safety, s u c h as in- procedures (a = .84). An example item was "I have the
cident investigations. A n e w scale w a s a d d e d to the knowledge I need to safely operate equipment used in
s u r v e y a s s e s s i n g perceptions a b o u t the a d e q u a c y o f my job."
Two types of motivational measure were used to match
safety e q u i p m e n t . M a n y o f the i t e m s in the n e w
the distinction between task and contextual performance
s u r v e y were c h a n g e d so that only o n e i t e m in e a c h o f
used with measures of safety behavior. Compliance Moti-
the c o m m o n d i m e n s i o n s r e m a i n e d the s a m e across vation was assessed by three items that asked about moti-
the two surveys. vation to perform safety-related tasks (a = .80). An exam-
ple item was "It is important to consistently use the correct
personal protective equipment." Participation Motivation
Me~od
was assessed by three items that assessed motivation to
participate in activities supporting safety in the organization
Sample. The sample for Study 2 consisted of 381
(or = .81). An example item was "I believe it is important to
employees in three Australian manufacturing organizations.
promote our safety program."
Listwise deletion of cases resulted in a final sample of 326
Two dimensions of safety performance were assessed.
employees. The survey was administered by the same or-
Safety Compliance was assessed by two items that asked
ganization as for Study 1 The overall response rate for the
survey was 57%. about individual proficiency in safety compliance (a = .56).
Measures. The revised survey assessed five dimen- An example item was "I use safe procedures for handling
sions of safety climate: one knowledge dimension, two hazardous materials." Safety Participation was assessed by
motivation dimensions, and two performance dimensions. three items that assessed safety behavior that was not di-
The Method section for Study 1 reports the items that were rectly related to safety compliance (o~ = .73). An example
the same in both surveys. Employees responded on a 5-point item was "I volunteer for safety-related tasks."
scale to all items ranging from 1 (strongly agree) to 5 Means, standard deviations, reliability coefficients, and
(strongly disagree). Safety climate was assessed by five correlations among all scales are presented in Table 4.
354 GRIFFIN AND NEAL

Table 4
Means, Standard Deviations, Correlations. and Reliabilities for All Scales in Study 2
Scale M SD 1 2 3 4 5 6 7 8 9 10
1. Manager Values 2.24 0.75 -- .54 .46 .49 .39 .43 .22 .25 .16 .24
2. Safety Communication 2.89 0.79 .48 -- .65 .64 .58 .56 .19 .24 .39 .44
3. Safety Practices 2.53 0.79 .45 .52 -- .55 .48 .43 .19 .17 .31 .29
4. Personnel Training 2.40 0.64 .45 .52 .42 -- .65 .58 .27 .26 .45 .38
5. Safety Equipment 2.50 0.77 .37 .46 .40 .53 -- .55 .18 .09 .31 .21
6. Safety Knowledge 2.52 0.73 .36 .47 .35 .50 .44 -- .26 .33 .50 .70
7. Compliance Motivation 1.44 0.53 .19 .14 .16 .26 .13 .20 -- .80 .32 .26
8. Participation Motivation 1.84 0.64 .22 .22 .16 .26 .12 .24 .65 -- .29 .53
9. Safety Compliance 1.65 0.69 .16 .26 .27 .34 .20 .34 .27 .18 -- .50
10. Safety Participation 2.83 0.89 .18 .33 .24 .32 .19 .47 .19 .42 .30 --
Note. Values above the diagonal represent correlations among the first-order factors; values below the diagonal represent
correlations among the scales.

Results and Discussion model, each sequential nested model s h o w e d a sig-


A similar analytic procedure for Study 1 was nificant i m p r o v e m e n t in fit. The series o f compari-
adopted in Study 2. First, a series o f C F A m o d e l s was sons proceeded f r o m the null covariance model to a
estimated including a higher order factor model. The 1-factor model, AX2(32, N = 326) = 2,204.19, p <
top section o f Table 5 s h o w s the fit o f each altema- .001; a 4-factor model, AX2(6, N = 326) = 1,096.02,
tive factor model. Beginning with the null covariance p < .001; a 5-factor model, AX2(4, N = 326) =

Table 5
Assessment of Measurement Model for Individual Responses to Survey Items in Study 2
Model df X2 GFI RMSEA CFI NNFI
Measurement models
Null covariance model 496 5,019.39 .30 .26
One-factor model 464 2,815.20 .58 .12 .48 .44
Four-factor model 458 1,719.18 .73 .08 .72 .70
One climate dimension
One knowledge dimension
One motivation dimension
One performance dimension
Five-factor model 454 1,663.25 .74 .07 .73 .71
One climate dimension
One knowledge dimension
One motivation dimension
Two performance dimensions
Six-factor model 449 1,569.59 .75 .07 .75 .73
One climate dimension
One knowledge dimension
Two motivation dimensions
Two performance dimensions
Ten-factor model 419 660.26 .89 .04 .95 .94
Five climate dimensions
One knowledge dimension
Two motivation dimensions
Two performance dimensions
Higher order climate model 444 704.01 .88 .04 .94 .94
Structural models
Hypothetical model 448 725.12 .88 .04 .94 .93
Final model 447 707.11 .88 .04 .94 .94
Note. til~l = goodness-ot-ht index; RMSEA = root mean square error of approximation; CFI = comparative fit index;
NNFI = nonnormed fit index.
PERCEPTIONS OF SAFETY 355

55.93, p < .001; a 6-factor model, AX2(5, N = General Discussion


326) = 93.66, p < .001; and the full 10-factor model,
AX2(30, N = 326) = 909.33, p < .001. The study's framework for assessing employee
This result supported a distinction among the pro- perceptions of safety is valuable because it allows for
posed first-order constructs. The higher order factor systematic assessment of different types of employee
model provided an acceptable fit, although it was a perceptions. These perceptions can then be linked to
significantly poorer fit compared with the 10-factor safety outcomes in organizations. Little research has
addressed the process through which safety climate
model, AXZ(25, N = 326) = 43.75, p < .05. The
translates into safety outcomes for organizations. In-
higher order model was retained to test the media-
vestigation of this process is hampered because there
tional links in the proposed model. Correlations
is no coherent framework that links employee per-
among the first-order factors from the measurement
ceptions of the work environment to specific safety
model are shown above the diagonal in Table 4. The
behaviors at work.
factor loadings for the higher order factor model are The proposed framework was based on an integra-
shown in Table 6. The loadings for all of the items tion of recent theories of work performance with
were acceptable and generally higher than the load- theories of organizational climate. Four key attributes
ings obtained for Study 1. of the framework were (a) the use of a higher order
Next, we estimated structural paths among the construct of safety climate comprising specific first-
constructs to assess the hypothesized mediational order factors, (b) the distinction of perceptions of
model. Results of these comparisons are presented in climate from other safety perceptions, (c) the distinc-
the lower part of Table 4. The hypothesized model tion between task and contextual performance related
was a significantly poorer fit than a saturated struc- to safety, and (d) the inclusion of mediational role for
tural model, AXz(4, N = 326) = 21.11, p < .001. knowledge and motivation. Implications of each of
Inspection of significant paths in the saturated struc- these four issues are addressed below.
tural model indicated a significant negative path from First, the framework incorporated safety climate as
safety compliance motivation to safety participation. a higher order factor comprising specific first-order
Inclusion of this path resulted in a final model that factors (L. R. James & Mclntyre, 1996). Perceptions
was not significantly different from the saturated of workplace systems related to safety were the de-
model, AX2(3, N = 326) = 3.10, p > .05. Figure 3 fining feature of the higher order safety climate fac-
shows the final path estimates from the SEM analy- t o r . This conceptualization is important because it
allows the role of the higher order factor to be inves-
sis. The results supported the proposed mediational
tigated in relation to other safety-related perceptions
structure because there was no direct path from the
and in relation to organizational outcomes. The pro-
higher order safety climate factor to the two safety
posed framework extends previous discussions of
performance factors.
safety climate that have focused on the dimensional-
ity of safety climate (Dedobbeleer & Beland, 1991;
Zohar, 1980). The framework does not resolve dis-
agreement about the number of subdimensions of
safety climate or provide a definitive conclusion
Table 6 about the content of these dimensions. However, the
Factor Loadings for Items in Study 2 framework allows for consistent development of fur-
ther specific dimensions of safety climate within the
Factor loadings construct of a higher order safety climate factor.
Factor 1 2 3 4 Second, the framework provides a basis for dis-
Manager Values .91 .89 .85 tinguishing other perceptions related to safety. In
Safety Communication .75 .73 .70 .62 particular, safety climate was distinguished from
Safety Practices .79 .78 .74 .71 self-reports of individual safety performance and
Personnel Training .76 .74 .63 from the motivational and knowledge determinants
Safety Equipment .86 .81 .52
Safety Knowledge .70 .60 .58 .54 of this behavior. Differentiating safety climate
Compliance Motivation .93 .79 .59 from other perceptions of safety in organizations
Participation Motivation .81 .79 .71 allows hypotheses regarding the mechanisms
Safety Compliance .71 .64 through which safety climate influences safety per-
Safety Participation .76 .70 .68 formance to be developed and tested. Previous
356 GRIFFIN AND NEAL

.61 .19

~ .14
_/
~--Safety~~ /

Figure 3. Path estimates from final model for Study 2. ns = not statisticallysignificant.

studies have confounded these issues (Dedobbeleer Finally, the mediation of safety climate through
& Beland, 1991). Furthermore, the exclusion of motivation and knowledge within the framework pro-
factors such as affective reactions to safety and vides an individual process that links safety climate
perceptions of risk from the construct of safety to specific performance outcomes. The results sup-
climate allows a clearer focus on the key element port the proposal that knowledge and motivation me-
of climate, namely, the perception of values, pol- diate the impact of safety climate on individual safety
icies, and procedures within the organization. behavior. This distinction is important because it
The third feature of the framework was the di- identifies mechanisms through which safety climate
vision of safety performance into behaviors di- is likely to influence safety performance (Campbell et
rectly related to safe work practices (safety com- al., 1993). In addition, the framework allows the
pliance) and behaviors that support the overall inclusion of other antecedents that may influence
safety of the organization (safety participation). knowledge and motivation for safety. For example,
The results support a distinction between task and
the risk level of a particular industry, organizational
contextual performance within the safety domain
characteristics such as size and structure, and char-
(Borman & Motowidlo, 1993). It is meaningful to
acteristics of the workforce can also be included in
distinguish between safety activities that are pre-
the framework as antecedents to individual safety
scribed as part of the job and safety activities that
performance. These antecedents may be related to
support the broader organizational context. Previ-
safety climate but are also conceptually distinct.
ous research has focused on safety behaviors such
as wearing the correct protective clothing. These Although the framework received general sup-
behaviors are directly related to the safety of the port, two aspects were not fully supported and
individual performing the behavior. The current require further investigation. First, the negative
framework extends safety performance to include link between safety compliance motivation and
activities such as participating in safety meetings safety participation was not expected. This nega-
and encouraging the safety of others. These con- tive relationship is consistent with resource allo-
textual behaviors can enhance the safety of the cation models of performance that propose goal-
team, the work environment, or the organization as oriented task motivation can reduce participation
a whole. It has long been recognized that contex- in contextual behaviors (Kanfer & Ackerman,
tual behaviors play an important role in maintain- 1989; Wright, George, Farnsworth, & McMahan,
ing organizational effectiveness (Katz & Kahn, 1993). Second, safety knowledge was not related
1966). Contextual safety behaviors are likely to to safety participation in Study 1. However, safety
play an important role in the maintenance of over- knowledge was positively related to both safety
all system safety. compliance and safety participation in Study 2.
PERCEPTIONS OF SAFETY 357

Limitations and F u r t h e r D e v e l o p m e n t elaboration of safety performance using the current


framework may include subdimensions of task and
The aim of this study was to develop a framework contextual performance to provide more detailed di-
for measuring perceptions of safety in the workplace. agnosis of safety perceptions in the organization.
The use of self-report measures within the framework
is therefore appropriate. However, the study is lim-
Conclusion
ited by not having access to measures of safety be-
yond individual perceptions. The common method of
The theories of performance developed by Camp-
measurement may have increased the strength of the
bell, Motowidlo, and their colleagues, together with
relationship among constructs. However, the pro-
theories of organizational climate, provide a useful
posed distinctions among constructs were supported
framework for measuring perceptions of safety at
by the confirmatory analyses, and previous studies
work. The results of the study provide a systematic
have demonstrated a link between safety behaviors framework for further investigation of the impact of
and accidents in organizations (Hofmann & Stetzer,
employee perceptions, employee behavior, and orga-
1996). Measurement of the relationship between em- nizational safety outcomes. The framework will pro-
ployee perceptions and indicators of both the work vide a basis for ongoing articulation of the core
context and individual behavior is necessary to fur- dimensions of safety climate and the mediating role
ther validate the constructs and to assess the impact of employee knowledge, skill, and motivation. This
of safety perceptions on workplace outcomes. A fur- framework will also aid the design and evaluation of
ther limitation of the study was cross-sectional mea- organizational interventions designed to enhance
surement. It was not possible to test the causal rela- safety outcomes. The results of the study, therefore,
tionships proposed in the theoretical framework, and provide a new step in developing a coherent model
longitudinal assessment would provide further vali- that links safety climate to safety in the workplace.
dation of specific relationships. Finally, neither study
included assessment of skill. Further development of
the model should incorporate a distinct measure of References
skill. The defining quality of skilled performance is
that it is accurate and effortless (J. R. Anderson, Anderson, J. R. (1983). The architecture of cognition. Cam-
bridge, MA: Harvard University Press.
1983). This quality makes it difficult to assess skill Anderson, N. R., & West, M. A. (1998). Measuring climate
independently of motivation and performance when for work group innovation: Development and validation
using self-report data. One way of assessing skill of the team climate inventory. Journal of Organizational
independently of other dimensions is for individuals Behavior, 19, 235-258.
Bentler, P. M. (1990). Comparative fit indexes in structural
to assess their prior performance in relation to the models. Psychological Bulletin, 107, 238-246.
amount of effort that they put in. Bentler, P. M., & Bonnet, D. G. (1980). Significance tests
In addition to addressing the above limitations, and goodness of fit in the analysis of covariance struc-
future research can be guided by the proposed frame- tures. Psychological Bulletin, 88, 588-606.
Borman, W. C., & Motowidlo, S. J. (1993). Expanding the
work in two ways. First, additional dimensions of
criterion domain to include elements of contextual per-
climate can be incorporated. The framework clearly formance. In N. Schmitt, W. C. Borman, & Associates
identifies the way that these climate dimensions (Eds.), Personnel selection in organizations (pp. 71-98).
should be conceptualized and assessed. Specific cli- San Francisco: Jossey-Bass.
mate dimensions should focus on perceptions of rel- Brown, R. L., & Holmes, H. (1986). The use of a factor-
analytic procedure for assessing the validity of an em-
evant features of the work context that are likely to be ployee safety climate model. Accident Analysis and Pre-
shared across members of the work group. For ex- vention, 18, 455-470.
ample, the perceived clarity of goals for safety could Brown, S. P., & Leigh, T. W. (1996). A new look at
be assessed to capture the degree to which individu- psychological climate and its relationship to job involve-
ment, effort, and performance. Journal of Applied Psy-
als perceive the goals of the organization to be linked
chology, 81, 358-368.
to safety in the workplace. Burke, M. J., Borucki, C. C., & Hurley, A. E. (1992).
Second, more specific measures of safety perfor- Reconceptualizing psychological climate in a retail ser-
mance can be developed to extend the way that vice environment: A multiple stakeholder perspective.
performance is reported by individuals in organiza- Journal of Applied Psychology, 77, 717-729.
Campbell, J. P., Gasser, M. B., & Oswald, F. L. (1996). The
tions. Performance requirements vary across organi- substantive nature of performance variability. In K. R.
zations, and it is likely that successful safety perfor- Murphy (Ed.), Individual differences and behavior in
mance also depends on the work context. Further organizations (pp. 258-299). San Francisco: Jossey-Bass
358 GRIFFIN AND NEAL

Campbell, J. P., McCloy, R. A., Oppler, S. H., & Sager, Klein, K. J., & Spotra, J. S. (1996). The challenge of innova-
C. E. (1993). A theory of performance. In N. Schmitt, tion implementation. Academy of Management Review, 21,
W. C. Borman, & Associates (Eds.), Personnel selection 1055-1080.
in organizations (pp. 35-70). San Francisco: Jossey-Bass. McCloy, R. A., Campbell, J. P., & Cudeck, R. (1994). A
Cox, S., & Tait, R. (1991). Reliability, safety and risk confirmatory test of a model of performance determi-
management: An integrated approach. Oxford, England: nants. Journal of Applied Psychology, 79, 493-503.
Butterworth-Heinemann. McHenry, J. J., Hough, L. M., Toquam, J. L., Hanson,
Dedobbeleer, N., & Beland, F. (1991). A safety climate M. A., & Ashworth, S. (1990). Project A validity results:
measure for construction sites. Journal of Safety Re- The relationship between predictor and criterion do-
search, 22, 97-103. mains. Personnel Psychology, 43, 335-354.
DeJoy, D. M. (1994). Managing safety in the workplace: An Medsker, G. J., Williams, L. J., & Holahan, P. J. (1994). A
attribution theory analysis and model. Journal of Safety review of current practices for evaluating causal models
Research, 25, 3-17. in organizational behavior and human resources manage-
Hayes, B. E., Perander, J., Smecko, T., & Trask, J. (1998). ment research. Journal of Management, 20, 439-464.
Measuring perceptions of workplace safety: Develop- Morrison, D. L., Upton, D. M., & Cordery, J. (1997, April).
ment and validation of the work safety scale. Journal of Organizational climate and skill utilization. Paper pre-
Safety Research, 29, 145-161. sented to the 12th Annual Conference of the Society for
Hofmann, D. A., Jacobs, R. R., & Landy, F. J. (1995). High Industrial and Organizational Psychology, St. Louis,
reliability process industries: Individual, micro, and MO.
macro organizational influences on safety performance. Motowidlo, S. J., & Van Scotter, J. R. (1994). Evidence that
Journal of Safety Research, 26, 131-149. task performance should be distinguished from contex-
Hofmann, D. A., & Stetzer, A. (1996). A cross-level inves- tual performance. Journal of Applied Psychology, 79,
tigation of factors influencing unsafe behaviors and ac- 475-480.
cidents. Personnel Psychology, 49, 307-339. Niskanen, T. (1994). Assessing the safety environment in
Hofmann, D. A., & Stetzer, A. (1999). The role of safety the work organization of road maintenance jobs. Acci-
climate and communication in accident interpretation: dent Analysis and Prevention, 26, 27-39.
Implications for learning from negative events. Academy Rundmo, T. (1994). Associations between safety and con-
of Management Journal, 41, 644-657. tingency measures and occupational accidents on off-
Hogan, J., & Roberts, B. W. (1996). Issues and non-issues shore petroleum platforms. Scandinavian Journal of
in the fidelity-bandwidth trade-off. Journal of Organi- Work, Environment and Health, 20, 128-131.
zational Behavior, 17, 627-637. Schneider, B. (1990). The climate for service: An applica-
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit tion of the climate construct. In B. Schneider (Ed.),
indexes in covariance structure analysis: Conventional Organizational climate and culture (pp. 383-412). San
criteria versus new alternatives. Structural Equation Mod- Francisco: Jossey-Bass.
eling, 6, 1-55. Schneider, B., Wheeler, J. K., & Cox, J. F. (1992). A
James, L. A., & James, L. R. (1989). Integrating work passion for service: Using content analysis to explicate
environment perceptions: Explorations into the measure- service climate themes. Journal of Applied Psychology,
ment of meaning. Journal of Applied Psychology, 74, 77, 705-716.
739-751. Wise, L. L., McHenry, J., & Campbell, J. P. (1990). Iden-
James, L. R., & Mclntyre, M. D. (1996). Perceptions of tifying optimal predictor composites and testing for gen-
organizational climate. In K. Murphy (Ed.), Individual eralizability across jobs and performance factors. Per-
differences and behavior in organizations (pp. 416-450). sonnel Psychology, 43, 355-366.
San Francisco: Jossey-Bass.
Wright, P. M., George, J. M., Farnsworth, S. R., & McMa-
James, L. R., James, L. A., & Ashe, D. (1990). The meaning han, G. C. (1993). Productivity and extra-role behavior:
of organizations: The role of condition and values. In B. The effects of goals and incentives on spontaneous help-
Schneider (Ed.), Organizational climate and culture (pp.
ing. Journal of Applied Psychology, 78, 374-381.
40-84). San Francisco: Jossey-Bass.
Zohar, D. (1980). Safety climate in industrial organizations:
Joreskog, K. G., & Sorbem, D. (1989). LISREL VII." User's
Theoretical and applied implications. Journal of Applied
reference and guide. Mooresville, IN: Scientific Software. Psychology, 12, 78-85.
Kanfer, R., & Ackerman, P. L. (1989). Motivation and
cognitive abilities: An integrative/aptitude-treatment in-
teraction approach to skill acquisition. Journal of Applied
Psychology, 74, 657-690. Received June 9, 1999
Katz, D., & Kahn, R. L. (1966). The social psychology of Revision received D e c e m b e r 6, 1999
organizations. New York: Wiley. A c c e p t e d January 19, 2000 •

View publication stats

S-ar putea să vă placă și