Content uploaded by Michael Burke
Author content
All content in this area was uploaded by Michael Burke on Dec 11, 2017
Content may be subject to copyright.
PERSONNEL PSYCHOLOGY
2002,s
GENERAL SAFETY PERFORMANCE:
A
TEST
OF
A
GROUNDED THEORETICAL MODEL
MICHAEL
J.
BURKE
A. B. Freeman School
of
Business
Tulane University
SUE ANN SARPY
Department
of
Environmental Health Sciences
School
of
Public Health and Tropical Medicine
Tulane University
PAUL E. TESLUK
R. H. Smith School
of
Business
University
of
Maryland
Department
of
Psychology
Tulane University
KRISTIN SMITH-CROWE
In
this investigation, we report the results of
2
studies designed to
(a) conduct confirmatory factor analytic tests of a model of general
safety performance
with
performance ratings from
550
coworker ap-
praisals (Study
1),
and (b) examine hypothesized relationships between
indicators of breadth and depth of knowledge constructs and con-
firmed safety performance factors (from Study
1)
with training history
data and supervisory appraisals for
133
hazardous waste workers in
23
jobs
and
4
organizations (Study
2).
Confirmatory factor analytic re-
sults from Study
1
provided support for a 4-factor model of general
safety performance with performance factors labeled Using Personal
Protective Equipment, Engaging
in
Work Practices to Reduce Risk,
Communicating Health and Safety Information, and Exercising Em-
ployee Rights and Responsibilities. In general, the results from Study
2
supported the hypothesized dominance of depth of knowledge over
breadth
of
knowledge in the prediction of performance with respect
to more routine, consistent safety tasks. Issues concerning the general-
izability of these factors to other types of work and the human resource
management implications of these results are discussed.
Within the management and organizational behavior literatures, the
linkage between organizational practices and individual performance is
generally conceptualized and studied as follows: Societal and organiza-
tionally espoused values will engender particular types of general and
human resource management practices. These management practices
will in turn influence employee attitudes, perceptions
of
the
work en-
Correspondence and requests
for
reprints shouid be addressed
to
Michael
I.
Burke,
A.
B.
Freeman School
of
Business, Tulane University, New Orleans, LA
70118;
mburkel @tulane.edu.
COPYRIGHT
0
2002
PERSONNEL PSYCHOLOGY, INC.
429
430
PERSONNEL
PSYCHOLOGY
vironment, and knowledge and skill acquikition; ultimately, these latter
individual difference variables will directly impact individual perfor-
mance. In these literatures, individual performance is most often viewed
as synonymous with the actions that employees engage in or display.
Al-
though decades of theory and research have been devoted to conceptual-
izing and measuring antecedents of individual performance, individual
performance as a construct domain has received very little theoretical
attention or research (see Borman
&
Motowidlo,
1993;
Campbell,
1990;
Hunt,
1996).
Without a clearer understanding of the substantive dimen-
sionality of individual performance-arguably the most critical individ-
ual difference domain in organizational behavior and human resource
management-the study of predictors and casual models of individual
performance Will remain enigmatic (Austin
&
Villanova,
1992).
Although there have been numerous calls over the last
50
years to
spend as much time and energy theorizing and studying individual per-
formance criteria as has been spent on the predictor side (cf. Austin
&
Villanova,
1992),
we have only recently begun to understand how
broadly defined individual performance should be for the purpose of
studying predictor-criterion relationships. Evidence from recent decades
of validity
generalizatiodmeta-analysis
research and several large scale
selection test validation studies suggest that it is primarily at relatively
broad analytic levels (ie., with respect to broad groupings of jobs or job
families) that we should expect certain individual attributes (i.e., cogni-
tive abilities and personality variables) to differentially relate to individ-
ual performance (e.g., see McHenry, Hough, Toquam, Hanson,
&
Ash-
worth,
1990).
That is, the available empirical evidence indicates that the-
orizing and studying the dimensionality of job performance with respect
to specific jobs or positions (for the subsequent purpose of examining in-
dividual antecedents of job performance) may not always be necessary.
Aviable conclusion is that general or generic modeling of the dimension-
ality of performance within a performance domain may be appropriate
for both scientific and practice concerns.
Several studies (e.g., Campbell, McHenry,
&
Wise,
1990;
Hunt,
1996)
have provided empirical support for a common factor structure
of job performance within a performance domain. For instance, Hunt
explored the dimensionality of
job
performance across a wide range of
entry-level jobs in samples drawn from retail stores, home improvement
centers, supermarkets, drug stores, restaurants, and theatre chains. He
asserted that actions, such as maintaining personal hygiene and treating
coworkers civilly, are basic, core behaviors that employees are required
to display to some degree in almost all entry-level, hourly wage jobs, such
as kitchen workers, cashiers, stock handlers, and data processing clerks.
Hunt’s study provided strong factor-analytic evidence for the multi-
MICHAEL
J.
BURKE
ET
AL.
43
1
dimensional nature of general job performance across these diverse en-
try level jobs.
In a similar vein, employees are required to display core, basic safety
behaviors to some degree across jobs in certain industries such as manu-
facturing, mining, chemical processing, nuclear power plant operations,
municipal public services (fire, police, and emergency medical services),
and others. In these industries, there exist segments or groupings of jobs
that require high levels of procedural knowledge and skill in order to
perform safely, thus protecting the public, the environment, the organi-
zation, and the workers themselves (cf. Cantor, 1992; Hofmann, Jacobs,
&
Landy, 1995).
In the growing body of research on behavioral aspects
of
safety, a
number of researchers have measured general safety behaviors
in
dif-
ferent industries (e.g., Cheyne, Cox, Oliver,
&
Tomas, 1998; Chhokar,
1990; Griffin
&
Neal, 2000; Hofmann
&
Stetzer, 1996; Lingard
&
Rowl-
inson, 1997; McDonald, Corrigan, Daly,
&
Cromie, 2000; Rudmo, 2000).
For the most part, this literature has focused on employee safety com-
pliance, the extent to which employees adhere to safety procedures and
carry out work in a safe manner. With only a few exceptions (cf. Griffin
&
Neal, 2000; Neal, Griffin,
&
Hart,
2000),
researchers measure safety
performance with respect to a single, overall scale. It is significant that
an examination of the dimensionality of safety performance within an
industry has not appeared in the literature. Understanding the mean-
ing (dimensionality) of safety performance as well as the knowledge-
and skill-related antecedents of safety performance is critical for guiding
management efforts toward the enhancement of safe work behavior and
the reduction
of
negative individual, organizational, and societal con-
sequences of unsafe work behavior. This understanding is particularly
important within high risk industries, such as nuclear power, where safe
work behavior has broad consequences.
The goal of this research is to develop and empirically evaluate a
model of general safety performance that is potentially applicable to
safety performance in many work domains. The remainder of the intro-
duction unfolds as follows. First, we discuss basic assumptions concern-
ing the measurement of safety performance. Subsequently, we propose
an a priori 4-factor model
of
general safety performance. Finally, we
provide information on the context for this research (i.e., a test
of
the
4-factor model within the domain of hazardous waste work) and present
the purposes of
two
studies designed to assess the construct validity of
the proposed 4-factor model of general safety performance.
432
PERSONNEL
PSYCHOLOGY
Conceptualizing General &fey Performance
General safety performance is defined as the actions or behaviors
that individuals exhibit in almost all jobs to promote the health and safety
of workers, clients, the public, and the environment. Several basic as-
sumptions concerning the nature and measurement
of
general safety
performance are made. First, consistent with the safety performance
literature (cf. Cheyne et al., 1998; Hofmann
&
Stetzer, 1996; Rudmo,
2000),
we assume that general safety behaviors can be scaled with re-
spect to the frequency that employees engage in the behaviors. Second,
consistent with the job performance literature (cf. Campbell, 1990), we
assume that safety behaviors covary in meaningful ways, yielding an in-
terpretable, multidimensional (correlated) factor structure. Third, we
assume that general safety performance factors are distinguishable in
terms of their determinants (e.g., procedural knowledge and skill) and
covariation with other variables (e.g., absences, accidents, and illnesses).
These assumptions, along with reviews of the relevant literatures and a
content-oriented measurement development study (to be discussed in
the Method section), led to the specification of the a priori 4-factor
model of general safety performance presented in the Appendix. Below,
we describe the respective factors for this model and discuss distinctions
between these factors and similar factors in the safety performance and
safety climate literatures.
Conceptual definitions for the four factors comprising the 4-factor
model of general safety performance are presented in the Appendix for
the factors labeled Using Personal Protective Equipment, Engaging in
Work Practices to Reduce Risk, Communicating Health and Safety In-
formation, and Exercising Employee Rights and Responsibilities. The
factors comprising this model are consistent with performance constructs
specified in the job performance literature, the literature on safety per-
formance, the literature on safety training, and the literature on safety
climate and safety culture. For instance, the factor Engaging in Work
Practices to Reduce Risk conceptually overlaps with measures in the
safety performance literature emphasizing worker safety compliance
(e.g., Cheyne et al., 1998; Griffin
&
Neal,
2000;
Komaki, Heinzmann,
&
Lawson, 1980; McDonald et al.,
2000).
Although the safety performance literature often considers the
proper use of personal protective equipment as an element of safety
compliance, we viewed Using Personal Protective Equipment as an inde-
pendent performance factor. Within a large number of industries, indi-
viduals receive extensive personal protective equipment training that is
believed to be essential to the development of procedural knowledge and
skills with respect to this performance dimension. For instance, within
MICHAEL
J.
BURKE
ET
AL.
433
the nuclear waste cleanup industry, hazardous waste workers need to
be proficient in the use of various types of equipment to protect them-
selves against three types of radioactive wastes: high level wastes (e.g.,
Cesium-137 and Strontium-90), transuranic wastes (i.e., liquid and solid
wastes contaminated with elements above uranium in the chemistry peri-
odic table such as plutonium, americium, and neptunium), and low level
solid and liquid wastes. Therefore, given the criticality of properly
us-
ing personal protective equipment and the unique, requisite procedural
knowledge and skills for using such equipment in a number of industries,
we conceptualized Using Personal Protective Equipment as an indepen-
dent, yet correlated, dimension of general safety performance.
The dimensions Communicating Health and Safety Information and
Exercising Employee Rights and Responsibilities presented in the Ap-
pendix are conceptually similar to dimensions in the safety climate
literature focusing on safety communication
(cf.
Carroll, 1998; Flin,
Mearns, O’Connor,
&
Bryden, 2000) and effective reporting of inci-
dentdaccidents (cf. Flin et al., 2000; Reason, 1998), respectively.
Al-
though these latter climate dimensions are somewhat similar to
two
of
the hypothesized safety performance dimensions, these safety cIimate di-
mensions and other climate dimensions are often operationalized with
respect to employee perceptions of organizational safety policies and
management safety practices or in terms of employee satisfaction with
the respective policies and practices (see Lee
&
Harrison, 2000, for
ex-
amples within the nuclear power industry). That is, in contrast to the
conceptualization and measurement of safety performance factors as ac-
tions or behaviors that employees exhibit, the
focus
of
safety climate is
not
on
particular behaviors of employees. We should
also
note that the
factor Communicating Health and Safety Information is consistent with
conceptual factors of general job performance concerning communica-
tion (e.g., Campbell’s, 1990, factor labeled Written and Oral Communi-
cation Tasks).
Context and Purposes
of
the Overall Study
The purpose
of
Study
1
was to empirically evaluate the a priori
4-factor model of general safety performance as a demonstration project
within the hazardous waste work domain. That is, as an operationaliza-
tion of the 4-factor model, the items serving
as
indicators of the four fac-
tors include a mix of mostly generic or general items that
are
applicable
to many types of work involving safety issues and several items that are
more context-specific (i.e., items that apply primarily to hazardous waste
work).
In
Study 1, we present results of confirmatory factor analytic
tests of the 4-factor model of general safety performance with coworker
434
PERSONNEL PSYCHOLOGY
(peer) appraisals from employees in four organizations. The employ-
ees worked in
23
different jobs at the 560-square mile Hanford nuclear
waste site situated along the Columbia River in Washington State. The
Hanford site is the most contaminated of the
14
large sites in
13
states
that make up the U.S. Nuclear Weapons Complex (Office of Technology
Assessment,
1993).
Using the general safety performance factors confirmed in Study
1,
in Study
2
we examined the extent to which increases in employees’
breadth of knowledge (operationalized as the diversity of training with
respect to a performance factor) and depth of knowledge (operational-
ized in terms of the amount of advanced or refresher training with re-
spect to a performance factor) are predictive
of
supervisory ratings
of
safety performance.
Study One
Method
Participants.
The participants for Study
1
were
574
hazardous waste
workers who provided anonymous ratings of the safety performance of
their coworkers. The targets of the coworker appraisals were employees
of three
DOE
contractors and one subcontractor. The targets
of
the per-
formance appraisals needed to have minimally completed the general
40-hour HAZWOPER (Hazardous Waste Operation and Emergency
Response) course and to be involved in some form of hazardous waste
work at the Hanford nuclear site in Washington State. The jobs held by
employees were representative of jobs at the Hanford site including nu-
clear waste process operator, health physics technician, electrician, field
engineer, material handler, craft supervisor, and plumbedpipe fitter.
Four-factor general safey pe@omance model and measure.
The fol-
lowing
is
a summary of the background work on the development and
administration
of
the general safety Performance measure. First, an ex-
tensive review of over
200
articles in the job performance, safety per-
formance, safety climate, and safety training literatures was conducted.
In addition, we reviewed the lesson plans for general hazardous waste
worker training as conducted by almost all major labor unions that repre-
sent hazardous waste workers (i.e., International Brotherhood of Team-
sters,
1993;
International Union of Operating Engineers, undated; Na-
tional Ironworkers and Employees Apprenticeship Training and Jour-
neyman Upgrading Fund, undated; Oil, Chemical, and Atomic Work-
ers International Union-The Labor Institute, undated; and The United
Brotherhood
of
Carpenters Health and Safety Fund
of
North America,
1992).
This review indi2nted that although existing general job per-
formance models (e.g., Campbell,
1990),
safety performance studies,
MICHAEL
J.
BURKE
ET
AL.
435
and other organizing frameworks for worker behaviors that included
safety-related factors (e.g.,
PAQ;
McCormick, Jeanneret,
&
Mecham,
1972)
could assist in identifying preliminary behavioral performance
constructs and items, these literatures were inadequate for comprehen-
sively
specifying
the behavioral domain of safety performance and, in
particular, hazardous waste worker performance. Therefore, a review
of these literatures along with a content-oriented validation study led to
the development of a preliminary 50-item safety performance measure.
Next, coworkers rated the frequency (using a 7-point scale ranging
fromnever to always) with which hazardous waste workers exhibited each
of the
50
behaviors. Third, item analyses and feedback meetings with
line workers and groups of supervisors were conducted. Based on the
results of the item analyses and feedback meetings, responses to the
27
items in the Appendix were retained for the present factor analyses and
grouped into four a priori factors as described above. We should note
that the eliminated behavioral statements were removed because it was
found that they did not apply to a vast majority
of
employees or required
respondents to rate the extent to which an employee had “demonstrated
knowledge” of a particular subject matter.
Coworkers were asked to rate the safety performance of a “typi-
cal coworker.”
A
typical coworker was defined as “the person that you
usually work with or the average performer in your work group.” This
anonymous coworker rating procedure was necessary for ensuring the
participation of various labor union members represented by the Han-
ford Atomic and Metals Trade Council. Coworker appraisals were ob-
tained in group sessions, where the appraisers were informed that their
evaluations would only be used for research purposes. The
4-factor model was specified by restricting each of the
27
items in the
Appendix to load only on the factor it was hypothesized to represent.
The analytic strategy consisted
of
three steps. First, cases in the data-
base with a large percentage
of
missing data (i.e.,
25%
or greater) were
eliminated. This brought the coworker sample size from
574
to
550.
Sec-
ond, in order to retain the maximum amount
of
data for the confirma-
tory factor analysis, missing data (typically data for one or
two
items for
249
cases) were estimated using a regression-based multiple imputation
procedure developed by Schafer
(1997).
‘Third, the confirmatory fac-
tor analytic test of the 4-factor model was conducted using
LISREL
8
(Joreskog
&
Sorbom,
1996).
The fit of the model was judged with
Spec$cation and test of general safety perjotmanee model.
’Data analyses conducted
011
I
he sample after listwise deletion of
cases
with any missing
data
(n
=
301)
yielded very similar levels of fit and factor loadings compared
to
when
using the covariance matrix created after replacing missing values.
For
that reason, the
analyses using the covariance matrix based
on
replacing missing values are presented.
436 PERSONNEL
PSYCHOLOGY
TABLE
1
Descriptive Statistics and Factor Loading Matrix for Four Factor
Model
of
General Safety
Perfomnee
Error
Item
M
SD
UPPE EWPRR
CHSI
EERR
uniqueness
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
Loading
6.16 1.06
5.43 1.36
5.96 1.05
5.80
1.13
5.43 1.33
6.02 1.12
5.80 1.14
5.99 1.00
5.97 1.32
5.74 1.19
5.80
1.14
5.85 1.11
6.27 0.99
6.16 1.10
6.02 1.10
5.93 1.16
5.58
1.25
6.14 1.10
6.03 1.10
5.93 1.25
5.94 1.22
5.66 1.40
.618
.701
326
.841
.741
.724
,720
.761
.690
.704
,796
.798
.721
.625
.770
,750
.741
,768
.795
.742
344
.771
.619
so9
.318
.293
.451
.476
.a1
.421
.525
SO5
.366
,363
.a0
.609
.407
.437
.452
.411
.368
.450
.2a7
.406
MICHAEL
J.
BURKE
ET
AL.
437
TABLE
1
(continued)
Error
Item
hf
SD
UPPE EWPRR CHSI EERR uniqueness
23
5.90
1.30
24 6.10 1.08
25 5.61 1.46
26 5.23 1.51
27 5.32 1.56
Loading
Loading
Loading
Loading
Loading
.I65 ,415
.772
.404
,791 .374
349 .278
.772 .404
Note:
This table represents the completely standardized solution for the factor load-
ings and error uniqueness. UPPE
=
Using Personal Protective Equipment; EWPRR
=
Engaging in
Work
Practices to Reduce Risk; CHSI
=
Communicating Health and Safety
Information; EERR
=
Exercising Employee Rights and Responsibilities. The
27
items are
the same as the respective items in the Appendix.
respect to the chi-square test, overall goodness of fit statistics (i.e.,
CFI and GFI), an analysis of residuals (i.e., RMSEA and Standardized
RMSR), and the magnitudes of item factor loadings.
Results
and
Discussion
The item means, item standard deviations, and factor-loading matrix
for the 4-factor model are presented in Table
1.
All factor loadings for
the model were greater than or equal to .62 and statistically significant.
The overall fit statistics for the 4-factor model (i.e.,
x2[318,
N
=
5501
=
1,377,
p
<
.01; CFI
=
.90;
GFI
=
34) and resid-
ual statistics (RMSEA
=
.078; Standardized RMSR
=
.048) indicated
that the model provided an acceptable fit
to
the data. The correlations
among the four factors ranged from .62
to
.79.
In spite of high correla-
tions among the four factors, the 4-factor model fit the data considerably
better than a single first factor, representing a possible common meth-
ods factor (x2[324,
N
=
5501
=
2,295,
p
<
.01;
CFI
=
.81,
GFI
=
.71;
RMSEA
=
.11, Standardized RMSR
=
.062).
Overall, the results from Study
1
provided support for a correlated,
4-factor model of general safety performance. Notably, the results indi-
cated that the four factors were highly correlated, leaving open the pos-
sibility that a common higher order factor may underlie general safety
performance. In addition, although the LISREL analyses identified sev-
eral complex items that could be loaded on other factors or eliminated
to improve the overall fit statistics for the 4-factor model, our decision
was to retain all items
as
originally hypothesized. This decision yields
a parsimonious general safety performance model that includes a com-
438
PERSONNEL PSYCHOLOGY
prehensive set of general behaviors representing this construct domain.
In a subsequent section, we will discuss how the measure in the Appen-
dix might be modified to measure general safety performance in other
occupations and types
of
work.
As
noted above, the four confirmed safety factors would be hypoth-
esized to have different knowledge and skill determinants. Empirical
support for hypothesized relationships between scores on safety dimen-
sions and indicators of knowledge and skill constructs would provide
fur-
ther support for the construct validity of the 4-factor safety performance
model. We now turn to Study
2,
which was designed to test hypothesized
relationships between knowledge and skill constructs and general safety
performance factors.
Study
Two
In the fields of applied psychology and organizational behavior, per-
formance is frequently expressed as a function of its determinants. A
longstanding proposition has been that job performance is a multiplica-
tive function of ability and motivation. This functional relationship, first
specified by Maier
(1955),
has received a fair amount of research atten-
tion. The literature has produced mixed results concerning the interac-
tion of ability and motivation in predicting performance. More recently,
Sackett, Gruys, and Ellingson
(1998)
examined this multiplicative propo-
sition when motivation was conceptualized and measured in terms of
personality. Basedon reanalysis of four large datasets they concluded that
ability-personalityinteractionsare
notdetectedatorabovechancelevels.
Campbell
(1990)
and Hunter and colleagues (Hunter,
1983;
Schmidt,
Hunter,
&
Outerbridge,
1986)
have proposed alternative models of per-
formance determinants. Both sets of authors argue that declarative
knowledge and procedural knowledge and skills (as indicated by work
sample performance in Hunter and colleagues’ research [Hunter,
1983;
Schmidt et al.,
19861)
are direct antecedents of job performance. Declar-
ative knowledge is defined as an understanding of the task requirements
or the ability to state the facts, rules, and principles that are prerequisite
for successful task performance (Anderson,
1985;
Kanfer
&
Ackerman,
1989).
Procedural knowledge and skill is the capability attained when
declarative knowledge (knowing what to do) is combined with knowing
how and being able to perform a task.
Although the research of Hunter and his colleagues (Hunter,
1983;
Schmidt et al.,
1986)
has focused
on
overall job performance as the cri-
terion, Campbell’s
(1990)
work has emphasized a taxonomy of perfor-
mance factors as relevant criteria.
An
important feature
of
Campbell’s
(1990)
framework is that knowledge (and measures thereof) should be
isomorphic or similar in content with the respective performance factors.
MICHAEL
J.
BURKE
ET
AL.
439
However, to date, almost all research examining relationships between
knowledge and job performance has linked overall job knowledge tests
or composite measures of job knowledge with overall performance rat-
ings or composite measures of job performance (e.g., Borman, Hanson,
Oppler, Pulakos,
&
White, 1993; Borman, White, Pulakos,
&
Oppler,
1991; Schmidt
et
al., 1986).
Importantly, declarative knowledge and procedural knowledge and
skill
are hypothesized within Campbell’s (1990) model
of
performance
determinants to be, in part, a function of education and training.
In
the hazardous waste worker domain, safety training is conducted in a
manner that would be expected to enhance knowledge and skill acqui-
sition and, thus, performance with respect to each factor.
As
a manda-
tory job requirement, all hazardous waste workers are initially required
to take extensive general safety training. This training is highly struc-
tured and designed to not only provide trainees with a broad health and
safety knowledge base with respect to central health and safety concepts,
safety rules and regulations, and safe work behaviors, but also develop
trainees’ understanding of interrelationships among these factors. The
training often incorporates multiple training methods including lecture
and discussion, followed by behavioral modeling, and then practice in
simulated contexts.
Increases in the diversity or breadth of the above types of training
within a knowledge area (e.g., training related to the use of various types
of personal protective equipment such as different types of respirators)
would be primarily expected to enhance declarative knowledge (which
we will refer to as
breadth
ofknowledge).
In particular, the enhancement
of declarative knowledge would provide hazardous waste workers with
a broad knowledge base for responding to new circumstances, changing
situations, and emergencies. Novel and changing situations in hazardous
waste work are similar to the notion of inconsistent tasks in the learning
theory literature (cf. Weiss, 1990). Responding
to
inconsistent tasks is
expected to be heavily dependent on one’s available resources (declara-
tive knowledge and cognitive abilities; Kahneman, 1973; Kanfer
&
Ack-
erman, 1989). Therefore, consistent with resource allocation models of
skill acquisition and performance (Kanfer
&
Ackerman, 1989), we hy-
pothesize that for each performance factor:
Hypothesis
1:
Breadth of knowledge will be positively related
to
safety
performance.
Increases in training in the form of safety refresher
or
recertifica-
tion courses would be primarily expected to enhance procedural knowl-
edge and skills through knowledge and skill updating and repeated-
440
PERSONNEL
PSYCHOLOGY
opportunities to practice with feedback. In this sense, skills are expected
to become proceduralized or automatized with increased refresher in-
struction, leading to more consistent exhibition of
routine
safety behav-
iors on the job. We will refer to the mastery and maintenance of pro-
cedural knowledge and skills as
depth ofknowledge and skills.
In terms
of stage theories
of
learning, refresher training would be expected to en-
hance knowledge compilation (Anderson, 1982), where individuals inte-
grate sequences of cognitive and motor processes required to perform a
task, and mastery of procedural skills (Anderson, 1982). Furthermore,
refresher training that exposes individuals to a variety
of
related or sim-
ilar material across different presentations should permit individuals to
more easily integrate the new materials with that already in memory
(Baldwin
&
Ford, 1988). Yet, the mastery and maintenance of proce-
dural knowledge and skill is expected to enhance the transfer
of
train-
ing to the job only when individuals have the opportunity to exhibit the
knowledge and skills they have learned (Ford, Quiiiones, Sego,
&
Sorra,
1992). In sum, consistent with stage theories of skill acquisition and per-
formance (cf. Anderson, 1982, 1985; Kanfer
&
Ackerman, 1989) and
research on transfer of training, we hypothesize the following for each
for each general safety performance factor:
Hypothesis
2:
Depth
of
knowledge and skill will be positively
related to
safety
performance.
In addition, given that general hazardous waste worker training in-
volves knowledge and skill acquisition with respect to all safety perfor-
mance factors, we also tested Hypothesis 2 with respect to depth of gen-
eral health and safety knowledge and skill and each performance fac-
tor (and a composite measure of general safety performance). Further-
more, noting that depth of knowledge and skill with respect to general
safety issues and the performance factors (i.e., Using Personal Protective
Equipment, Engaging in Work Practices to Reduce Risk, Communicat-
ing Health and Safety Information, and Exercising Employee Rights and
Responsibilities) become translated into greater work experience over
time, we will also test Hypothesis 2 with respect to a composite mea-
sure of depth of safety knowledge and skill and a composite measure
of general safety performance.
A
test of the latter expectation will pro-
vide a linkage to the literature examining relationships between over-
all job knowledge tests or composite measures
of
job knowledge and
overall performance ratings or composite measures of job performance
(cf. Borman, White, Pulakos,
&
Opler, 1991; Borman, Hanson, Oppler,
Pulakos, &White, 1993; Schmidt et al., 1986).
MICHAEL
J.
BURKE
ET
AL.
441
Given that most general safety behaviors or tasks in the Appendix are
consistent or routine tasks, we expect that depth of knowledge and skill
will be more strongly related to performance on each factor than breadth
of knowledge. In addition, over time, the number of safety refresher
courses that one takes becomes translated into greater depth
of
work
experience in a given safety performance area. Therefore, the expecta-
tion that depth of knowledge and skill
will
be more strongly related than
breadth of knowledge to performance on each factor also is supported
from work experience perspectives (Quiiiones, Ford,
&
Teachout, 1995;
Tesluk
&
Jacobs, 1998).
Thus,
for each safety performance factor, the
third hypothesis is as follows:
Hypothesis
3:
Depth of knowledge and skill
will
be more strongly related
to safety performance than breadth of knowledge.
In addition, we will test Hypothesis
3
with composite measures of depth
and breadth of knowledge and a composite measure of general safety
performance.
Method
Participants
The participants for Study
2
varied between
127
and
133
hazardous
waste workers whose performance had been rated by their supervisors.
The variations in sample size represent individuals who had usable data
(on a pairwise basis) for examining relationships between training histo-
ries and safety performance. For these workers, complete safety train-
ing histories were available. The primary jobs and number of workers in
these jobs were nuclear waste process operator
(n
=
44),
health physics
technician
(n
=
14),
electrician
(n
=
lo),
field engineer
(n
=
8),
mate-
rial handler
(n
=
6),
craft supervisor
(n
=
6),
plumber/pipefitter
(n
=
5),
and
16
other jobs
(n
=
40).
Measures
of
Depth
of
Knowledge and Skill, Breadth
of
Knowledge, and
Safety Pegormame
Depth
of
knowledge and skill.
For each performance factor (with the
exception of Exercising Employee Rights and Responsibilities), indica-
tors of depth of knowledge and skill were computed as the sum of the
number of refresher training courses taken and passed in the respective
knowledge and skill area. For instance, depth of knowledge and skill
with respect to Using Personal Protective Equipment was computed as
the sum of the number of times an individual had taken five equipment-
442
PERSONNEL
PSYCHOLOGY
related refresher courses, such as the annual self-contained breathing
apparatus course and the quantitative mask fit refresher course. In
ad-
dition, indicators of depth
of
knowledge and skill were computed for
general health and safety training as the sum of the number of general
employee training courses and &hour annual HAZWOPER training
courses a worker had taken and passed. Finally, an unweighted compos-
ite (sum) of the number of refresher courses a worker had taken across
knowledge and skill areas was computed. We should note that the safety
refresher courses involved lecture presentations of new material (e.g.,
legal developments, advances in equipment design) followed in many
cases by demonstrations and practice with feedback.
Breadth
of
knowledge.
For the performance factors, Using Personal
Protective Equipment and Engaging in Work Practices to Reduce Risk,
indicators of breadth of knowledge were computed by summing the num-
ber of different courses a worker had taken and passed in each area. For
instance, breadth of knowledge concerning Engaging in Work Practices
to Reduce Risk involved summing the number of different initial (nonre-
fresher) training courses an individual had taken dealing with confined
space
(2
possible courses), radiological control
(12
possible courses),
as-
bestos control
(5
possible courses), lead control
(3
possible courses),
lock and tag
(7
possible courses), packaging and transport
(9
possible
courses), and bloodborne pathogens
(1
possible course). These training
courses often included a mix of lecture-based instruction and behavioral
role modeling.
An
unweighted composite (sum) of the number of dif-
ferent initial (nonrefresher) training courses a worker had taken across
knowledge and skill areas was also computed.
Breadth of knowledge could not be computed for knowledge related
to Exercising Employee Rights and Responsibilities and Communicating
Health and Safety Information. Material related to exercising one’s
rights and responsibilities is only presented in the general HAZWOPER
course and there is only one specific hazard communication course.
Safetyye$ormance measures.
The performance measures for Study
2
were based on the four confirmed factors of Study
1.
Foremen who di-
rectly supervised the line employees provided the performance ratings
in group sessions, where the appraisers were informed that their evalu-
ations would only be used for research purposes.
We should note that we considered conducting a common factor
analysis on the supervisory ratings to examine the invariance
of
the con-
firmed 4-factor model of safety performance. However, due to the pat-
tern of missing data in the supervisory ratings of employee performance,
we would have needed to impute data (on one or more items) for
79
of the
133
hazardous waste workers who were rated by their supervisor.
Therefore, we relied
on
the confirmatory factor analyses
of
the coworker
MICHAEL
J.
BURKE
ET
AL.
443
data for defining the performance scales (i.e., measures). However, sta-
tistical tests of differences between the coworker and supervisory means
and standard deviations for the respective performance factors were not
significant and both sets of distributions were negatively skewed. Perfor-
mance on each general safety performance factor was computed as the
average of the respective item ratings for the performance factor.
Data Analyses
Initially, Pearson correlations were computed for tests of the hypoth-
esized relationships between breadth of knowledge, depth
of
knowledge
and skill, and performance on each factor. In addition, overall associ-
ations among safety knowledge and safety performance measures were
estimated by computing
q
coefficients.
Results and Discussion
The variable means, standard deviations, skewness values, and inter-
correlations among the observed variables are reported in Table 2. The
tests of the study hypotheses are reported in Table
3,
along with overall
measures of association
(q
coefficients).
As
shown in Table
3,
Hypothe-
sis
1
was not supported in that breadth of knowledge with respect to the
use of personal protective equipment and breadth of knowledge with re-
spect to work practices had small, nonsignificant linear relationships with
the respective performance factors. In addition, a composite measure of
breadth of knowledge had a low relationship with an overall (composite)
measure of safety performance.
The hypothesis (Hypothesis 2) that depth of knowledge and skill
would be positively related to performance on each factor was gener-
ally supported. The correlation between depth of knowledge and skill
relating to Using Personal Protective Equipment and performance
on
this factor was marginally significant
(1.
=
.16,
p
<
.lo)
and depth of
knowledge and skill relating to Engaging in Work Practices to Reduce
Risk correlated significantly with performance on the respective fac-
tor
(T
=
.21,
p
<
.05).
Also,
depth of knowledge and skill in general
safety correlated significantly with a composite measure
of
Performance
(T
=
.25,
p
<
.Ol). Finally, a composite measure of depth of knowl-
edge and skill correlated significantly with the composite safety perfor-
mance measure
(T
=
.22,
p
<
.Ol). The finding that depth of knowledge
and skill did not correlate with Communicating Health and Safety In-
formation is understandable, given the nature of the behaviors (indi-
cators)
of
this performance factor. That is, several indicators
of
this
performance factor would be considered as more inconsistent tasks
or
TABLE
2
Variable Means, Standard Deviations and Observed Intercorrelations
(N
=133)
~~ ~~
Variable
Variable"
M
SD
SKb
1
2 3 4
5
6 7
8
9 10111213
1.
Using personal protective equipment"
6.11 0.72 -.98 (39)
2.
Engaging in work practices to reduce risk
6.11 0.70 -1.28 .74 (.89)
3.
Communicating health and safety information
6.19 0.74 -1.52 .53 .65 (33)
4.
Exercising employee rights and responsibilitieP
5.66 1.29 -1.25
51
.69
.60
(33)
5.
Composite performance
6.08 0.67
-1.11
$5
.93 .78 .75
-
6.
General safety DK
8.05
4.12 -.09 .18 .29 .20 .20 .25
-
7.
Using personal protective equipment DK
8.56 4.67
.lo
.16 .26 .17 .14 .21 .75
-
8.
Using personal protective equipment BK
4.52 0.96 -2.78 .01 .04 .03
.02
.03
.15
.48
-
9.
Engaging inworkpractices to reduce riskDK
9.08 5.42
SO
.09 .21 .14
.08
.16 .69 .72 .27
-
10.
Engaging in work practices to reduce risk BK
7.46 3.03
51
.09 .05
.05
-.14 .07 .47 .54 .28 .68
-
11.
Communicating health and safety information DK
0.74 1.07 1.43 -.07
-.01
.01 -.14 -.03 .40 .31
.08
.34 .45
-
12.
Composite safety DK
26.44 14.02 .37 .14 .27 .18 .13 .22 39 .91 .33 .90 .65 .45
-
13.
Composite safety BK
11.98 3.43 -.02
.08
.06
.05
-.I2 .07 .46 .61
.53
.68 .96 .42 .67
-
Note:
Coefficient alphas are reported in parentheses
on
the diagonal.
For
rs
2.23,
P
<
.01;
for
T~
2.18,
p
<
.05
"DK
=
Depth
of
knowledge and skill;
BK
=
Breadth
of
knowledge.
*SK=skewness.
N=132.
dN=127
TABLE
3
Tests
of
Hypothesized Relationships Between Breadth of Knowledge, Depth
of
Knowledge and Skill,
and
Safely Performance
Performance
UPPE EWPRR
CHSI
EERR Composite
Knowledge and skill variable
rrl
r77
r77
r77
r
77
Using personal protective equipment (UPPE)
Breadth of knowledge"
.01
.19
Depth
of
knowledge"
.16'
.3S
Breadth
of
knowledge"
.05
.32
Engaging in work practices to reduce risk (EWPRR)
Depth of knowledge" .21* .39
Depth
of
knowledge .01
.08
Communicating health and safety information
(CHSI)
Composite breadth
of
knowledge"
.07
.48
Composite depth
of
knowledge'
,22**
.62
General health and safety
Depth of knowledge
.18*
.SO
.29*
*
.44
20'
.38
.20*
.44
.25'
*
.47
The difference between the respective correlations involving breadth and depth of knowledge measures is statistically significant at
p
<
.OS,
one-tailed.
EERR
=
Exercising Employee Rights and Responsibilities
tp
<
.10
*p
<
.os
**p
<
.01
P
446 PERSONNEL
PSYCHOLOGY
low base rate behaviors (i.e., communicating information related to
emergencies, accidents, and exposures), which would be expected to re-
late more to breadth of knowledge.
The findings concerning depth of knowledge and skill are consistent
with theoretical rationales for a causal linkage between skill updating
or refresher training and the proceduralization of more routine, task-
oriented safety behaviors. More specifically, the findings concerning
depth of knowledge and skill are in line with learning theory expecta-
tions that refresher training and advanced training enhance the integra-
tion of information and mastery of procedural skills and, thus, improve
performance. Furthermore, these findings are consistent with Ford and
his colleagues’ (Baldwin
&
Ford,
1988;
Ford et al.,
1992)
expectations
that exposing individuals to a variety of related or similar material (i.e.,
increasing “stimulus variability”) across different presentations, such as
refresher training courses, can improve the transfer of training to the job
of more routine, task-oriented behaviors.
All
of the correlations between depth of knowledge and skill and per-
formance reported in Table
3
were significantly greater than the respec-
tive correlations between breadth of knowledge and performance, con-
firming our expectations (Hypothesis
3)
about the dominance of depth
of knowledge and skill over breadth of knowledge in predicting perfor-
mance on more consistent safety tasks. Given the nature of the general
safety behaviors studied (i-e., more routine types of tasks that should be
consistently engaged in), the correlational findings concerning the dom-
inance
of
depth of knowledge and skill over breadth
of
knowledge in the
prediction of safety performance is understandable. Breadth
of
knowl-
edge would be expected to relate more to safety performance when the
tasks are inconsistent in nature and depend heavily
on
one’s available
resources (declarative knowledge and cognitive abilities). For instance,
there are unforeseen critical incidents that occur in hazardous waste
work that would be considered as inconsistent, nonroutine tasks or emer-
gencies. Here, we would expect breadth of knowledge to have a stronger
linear relationship with safety performance. However, the present over-
all study focused on general, routine tasks that are carried out
in
a more
consistent manner, when requisite skills are proceduralized.
Notably, followup tests
(7
coefficients) for the hypothesized safety
knowledge-safety performance relationships indicated that the safety
performance criteria could also be predicted from the respective knowl-
edge and skill variables
by
a
“best fitting” line that was nonlinear. That is,
in all cases, the
7
coefficients were meaningfully greater than the respec-
tive observed Pearson correlation coefficients and greater than all but
one of the respective corrected correlations. The average for the
7
coef-
ficients across the twelve hypothesized relationships was
.39;
whereas the
MICHAEL
J.
BURKE
ET
AL.
447
average of the 12 observed Pearson correlations was
.15
and the average
corrected correlation was .26.
The form
of
nonlinear relationship was consistent with respect to the
depth of knowledge-safety performance relationships. That is, a mono-
tonically increasing prediction line that showed an initially steep slope
and then tended to plateau with greater amounts of depth of knowledge
adequately described the overall relationships. Importantly, followup
quadratic regression analyses indicated that the nonlinear relationships
involving depth of general health and safety knowledge and composite
safety performance and the composite depth of knowledge measure and
composite safety performance were statistically significant
(F
=
5.08,
p
<
.01;
F
=
3.95,~
<
.05,
respectively). With the exception
of
depth of
knowledge concerning communication, the nonlinear, quadratic equa-
tion for all other relationships involving depth of knowledge measures
either approached statistical significance or was marginally significant.
In contrast, the curves for the breadth
of
knowledge-safety performance
relationships were flatter throughout the range of the predictor and
tended to be slightly negatively accelerating at higher levels of breadth.
Given that depth of knowledge in comparison to breadth of knowl-
edge is more closely associated with greater work experience, the over-
all depth of knowledge-safety performance relationships are consistent
with findings reported in Vineberg and Taylor
(1972),
Schmidt et al.
(1986),
Jacobs, Hofmann, and Kiska
(1990),
and Farrell and McDaniel
(2001).
Using Vineberg and Taylor's data, Schmidt et al.
(1986)
showed
that for individuals having up to 60 months of experience in medium
complexity military jobs, job knowledge and work sample scores in-
creased linearly with tenure; beyond this point the relations appeared
to plateau. In addition, for a sample of firefighters, Jacobs et al.
(1990)
found a very similar nonlinear relationship between months of job tenure
and supervisory ratings of performance with the plateau occurring at
the 60-month point. We encourage researchers to examine the nature
of safety knowledge-safety performance relationships with respect to
breadth and depth of knowledge.
General
Discussion
The findings from the overall study offer progress toward a taxon-
omy of general safety performance
by
directly addressing the criterion
dimensionality issue.
Four
general safety performance factors were
con-
firmed within Study
1
and labeled Using'Personal Protective Equipment,
Engaging in Work Practices to Reduce Risk, Communicating Health and
Safety Information, and Exercising Employee Rights and Responsib-
ilities. In addition to being grounded in the extant literature, the four
448
PERSONNEL PSYCHOLOGY
safety performance factors are consistent with safety training content ar-
eas and lesson plans (in terms of being segmented into training modules)
for a number of major labor unions (e.g., International Brotherhood of
Teamsters,
1993;
The United Brotherhood of Carpenters Health and
Safety Fund of North America,
1992).
In essence, the four safety perfor-
mance factors and behaviors that compose the domain
of
general safety
performance are closely associated with an implicit theory of the safety
performance domain. Importantly, the four general safety factors pro-
vide a starting point for future research on the dimensionality
of
general
safety performance in other types of work.
Notably, the four general safety factors were highly correlated, with
correlations among the factors ranging from
.62
to
.79
in Study
1
and
from
51
to
.74
in Study
2.
This positive manifold, coupled with the fact
that a test of a single, unitary factor model provided a poor fit to the data
in Study
1,
implies that a common higher order factor underlies the four
first-order general safety performance factors. Along with a general lack
of discriminant validity in the prediction of the four factors in Study
2,
this positive manifold would suggest the need for future research to ad-
dress both the possible nature and presence of a higher order safety per-
formance factor. In effect, a composite or overall safety performance
score for the General Safety-Performance Scale
(GSS)
may be mean-
ingful as an indicator of a higher order safety performance factor.
The degree to which the confirmed factors of the
GSS
generalize
to work domains other than nuclear waste is both a theoretical and
an empirical question. Due to similar actions that workers engage
in across many types
of
safety-related work and the regulatory nature
of safety-related work, we anticipate that the four general safety fac-
tors would generalize
to
safety behavior in other industries, such as
agriculture, mining, energy, manufacturing, construction, transporta-
tion, and public services including health care. However, we expect
that the four subscales used
in
this study to measure the dimensions
of general safety performance will, in some cases, require slight modifi-
cation for use in other work domains. For optimal use in other types of
work, we suggest that researchers and practitioners address the follow-
ing three questions for the respective items in each of the general safety
performance subscales: (a) does the item apply to this type
of
work (if
not, delete), (b) does the language of the item need to be modified (if
yes, modify accordingly), and (c)
do
items need to be added to reflect
relevant safe work behaviors in the work domain (if yes, add appropri-
ate items to the respective scales). Items that are more context-specific
or items that may require editing prior to use in another type of work
are identified in the Appendix.
MICHAEL
J.
BURKE
ET
AL.
449
As
examples, several dimension-specific suggestions for modifying
items of the
GSS
for use in other types of work are as follows. First,
Items
6
and
9
of Using Personal Protective Equipment may not apply
to some types of work that are more individualized or work that does
not require the use of pressurized air respirators, respectively. Second,
Item
22
(“Contacts appropriate personnel when faced with questions
and/or issues regarding HAZWOPER”) would require minor editing.
That is, the context-specific term HAZWOPER could be replaced with
the word “safety.” Third, an item such as, “Exercises rights to review
MSDS and to use other reference materials that may provide additional
health and safety information,” could be reworded to include the par-
ticular safety manuals and reference guides for another type of work.
Here, an MSDS (Materials Safety Data Sheet) may not be applicable to
other types of work.
Although the above recommendations may enhance the application
of the GSS to other types of work, the above suggestions are not intended
to preclude efforts to study more specific safety performance factors in
either hazardous waste work or other types of work. Certainly, the pos-
sibility that more specific safety performance factors could assist in ex-
plaining safety behaviors in the present set of jobs or other
jobs
is very
likely. For instance, recognizing and evaluating hazards and responding
to emergencies are two possible performance factors that could be the
subject of confirmatory factor analytic efforts
to
explain performance
variability in other types of safety-related work, such as firefighting or
emergency medical services. In addition, the study of contextually ori-
ented performance in the safety domain would be a useful supplement
to the present effort, which examined worker actions relative to more
consistent, routine tasks.
Although a detailed discussion and empirical examination of the
relation between general safety performance and routine task perfor-
mance is beyond the scope of this study, the potential overlap between
general safety performance and task performance is an important con-
ceptual and practical issue. Across jobs, general safety performance and
task performance are likely to overlap along a continuum. Within criti-
cal skills occupations where safety concerns are an important aspect of
the work, the four dimensions that comprise general safety performance
described here are likely reflected in task performance. ask perfor-
mance has been defined as behavior involved in transforming raw ma-
terials into products and services or servicing or maintaining the tech-
nical core (Borman
&
Motowidlo,
1993).
General safety performance
is inherent in these latter types of behaviors. For instance, task per-
formance, such as operating a production machine in a manufacturing
plant, will likely require engaging in certain actions, such as wearing
450
PERSONNEL PSYCHOLOGY
any necessary personal protective equipment (Using Personal Protec-
tive Equipment dimension), properly disposing of any dangerous scrap
material (Engaging in Work Practices to Reduce Risk dimension), and
reporting any safety incidents/accidents while working on the machine
(Communicating Health and Safety Information dimension). In many
ways, the notion that safety performance is an inherent aspect of task
performance in more safety-related types of work parallels Borman and
Motowidlo’s
(1993)
observation that contextual performance might be
considered task (routine) performance in more service-oriented work.
Indeed, safety behaviors have been included when measuring task per-
formance (e.g., Motowidlo
&
Van Scotter,
1994).
Another likely factor influencing the relationship between general
safety and task performance is how effectiveness is defined, monitored,
and rewarded with respect to task performance. For instance,
as
is well
documented in the safety literature (e.g., Cohen,
1977;
Pate-Cornell,
1990;
Zohar,
1980),
following safety procedures (e.g., taking the time
to put on safety goggles) can add steps to the work process that can re-
duce efficiencies (e.g., how quickly a machine operator completes a work
operation). Whether task performance is defined with safety consider-
ations in mind is likely due to factors in the broader social environment
like safety climate (e.g., Zohar,
2000)
and supervisory messages relat-
ing to the important goals of the work group (Hofmann, Morgeson,
&
Gerras,
2001).
In addition to future research efforts aimed at clarifying the con-
ceptual and empirical distinction between general safety performance
factors and task performance factors, other research directions might
include confirmatory factor analytic tests of safety performance factor
structures with data from other sources (e.g., supervisors). Recall, the
first study relied on employee ratings
of
the “typical” coworkers per-
formance, which may have affected the attained factor solutions. In
addition, examinations of other hypothesized individual difference an-
tecedents of safety performance (ability and personality variables) in-
cluding an assessment of whether or not these individual difference
variables differentially predict the four dimensions of safety pecfor-
mance would further add to assessments of the construct validity
of
the
GSS.
Finally, an assessment of the degree to which individual differ-
ence-safety performance relationships are moderated by organizational
climate would be a meaningful addition to the literature.
In regard to studying antecedents of safety performance, the present
findings add to our understanding of how workers’ training experiences
contribute to safety performance. In particular, the present findings
contribute to our limited understanding of how breadth and depth
of
knowledge and skill relate to safety performance. The magnitudes and
MICHAEL
J.
BURKE
ET
AL.
45
1
nonlinear nature of most relationships between depth of knowledge and
skill and safety performance are consistent with relationships reported
in the literature between job tenure, job knowledge, work sample per-
formance, and overall supervisory performance ratings (e.g., Jacobs et
al., 1990; Schmidt et al., 1986). In addition, the present linear depth of
knowledge-safety performance relationships are likely lower than
com-
parable estimates of job knowledge-performance relationships reported
in the literature due to the fact that depth of knowledge and skill scores
are restricted in range. If an individual completed a safety refresher
course (i.e., updated his or her skills), then he or she would have ex-
ceeded a minimum passing score on a refresher course test. In addition,
we believe that the moderate to high relationships between the various
knowledge and skill measures are more reflective of the development
of well organized safety knowledge structures (Stout, Salas,
&
JSraiger,
1997) and the fact that, over time, increases in safety knowledge and
skill become closely associated with greater depth of work experience
(cf. Tesluk
&
Jacobs, 1998).
We should also note that the present observed safety knowledge-
safety performance relationships are likely lower than might be found
in other types of work as a result of substantive range restriction (and
consequent negative skew) on the performance measures. That is, the
distributions for the safety performance measures were restricted and
negatively skewed, likely owing to the highly regulated nature of perfor-
mance in the domain of hazardous waste work.
Although our discussion to this point has focused on issues con-
cerning the construct validity of general safety performance factors and
breadth and depth of knowledge and skill variables, the present findings
have several important practical implications. First, the magnitudes of
the depth of knowledge-safety performance relationships indicate that
the continuance of annual refresher training courses and advanced train-
ing is meaningfully related to safety performance. In terms of capitaliz-
ing on economies of scale, our findings support across-job safety training
in the areas of general health and safety, personal protective equipment,
and work practices aimed at reducing risk.
Although the four safety performance factors are highly correlated,
another practical implication concerns the value of the 4-dimensional
model in settings where specific aspects of safety performance are em-
phasized. Examples might include emphasizing the Using Personal Pro-
tective Equipment dimension to evaluate trainee safety performance fol-
lowing a training program for firefighters on the proper practices for us-
ing and inspecting their personal protective equipment. Alternatively,
one could employ the Communicating Health and Safety Information
dimension and measure to study the influence of leadership and orga-
452
PERSONNEL
PSYCHOLOGY
nization support on safety communication behaviors (e.g., Hofmann et
al.,
2001).
However, in other situations, a single composite measure may
be more appropriate, for instance, when seeking to determine an em-
ployee’s overall level of safety performance
as
one factor in conducting
a performance evaluation, validating a selection procedure with respect
to safety performance, or making a promotional decision.
Conclusion
Although the economic and societal benefit of efforts to clean up
hazardous waste sites and consequently the utility of hazardous waste
worker training
are
increasingly issues of public debate and commentary
(cf. Murphy,
2000),
an important message from this study is that progress
is being made with respect to the training and performance
of
those
responsible for cleaning up the nation’s nuclear weapons complex. The
extensive hazardous waste worker training at the Hanford nuclear waste
site, the site of this overall study, occurs within a safety culture that values
continuous safety education, the exhibition
of
safe work behaviors, and
a concern for the health and well being of workers and the public. In
our opinion, the Hanford safety culture undergirds the associated
DOE
contractor safety climates and the effectiveness of the present safety
training and worker safety performance. The failure to make progress
in these areas and the lack of a safety culture or safety climate have
been singled out as primary precipitating factors in foreign nuclear waste
disasters at Chernobyl and Tokaimura (cf. International Atomic Energy
Agency, 1996, 1999). In closing, this study provides a theoretical and
empirical foundation for understanding general safety performance and
the role of safety training histories that can be used in guiding research
and practice efforts concerned with general safety performance.
REFERENCES
Anderson JR. (1982). Acquisition
of
cognitive skill.
Psychological Review,
89,369406.
Anderson JR. (1985).
Cognitive psychology and its implications
(2nd ed.). New
York
Freeman.
Austin JT, Villanova
P.
(1992). The criterion problem: 1917-1992.
Journal
of
Applied
Psychology.
77,836-874.
Baldwin
‘IT,
Ford JK. (1988). Transfer of training:
A
review and directions for future
research.
PERSONNELPSYCHOLOGY,
41,63-105.
Borman WC, Hanson MA, Oppler
SH,
Pulakos ED, White LA. (1993). Role of early
supervisory experience in supervisor performance.
Jouml
of
Applied PvchoIom,
78,443-449.
Borman WC, Motowidlo SJ. (1993). Expanding the criterion domain
to
include elements
of
contextual performance. In Schmitt
N,
Borman WC (Eds.),
Personnel selection
in
organizations
(pp. 71-98). San Francisco: Jossey-Bass.
Borman WC, White LA, Pulakos ED, Oppler SH.
(1991).
Models of supervisory job
performance ratings.
Journal
of
Applied Psychology,
76,863-872.
MICHAEL
J.
BURKE
ET
AL.
453
Campbell JI?
(1990).
Modeling the performance prediction problem in industrial and
organizational psychology. In Dunnette MD, Hough LM (Eds.),
Handbook of
industrial and
otganizationalpJychology:
2nd ed.,
(pp.
687-732).
Palo Alto, CA
Consulting Psychologists Press.
Campbell
JP,
McHenry JJ, Wise
LL. (1990).
Modeling job performance in a population
of
Cantor JA.
(1992).
Evaluation of human performance in critical-skills occupations: Cri-
Carroll JS.
(1998).
Safety culture as an ongoing process: Culture surveys as opportunities
Cheyne A,
Cox
S,
Oliver A, Tomas JM.
(1998).
Modeling safety climate in the prediction
Chhokar JS.
(1990).
Behavioral safety management.
nkaka,
15,
15-22.
Cohen A.
(1977).
Factors in successful occupational safety programs.
Journal
of
Safety
Research,
9,
168-178.
Farrell JN, McDaniel MA.
(2001).
The stability
of
validity coefficients over time: Acker-
man’s
(1988)
model and the general aptitude test battery.
Journal of Applied
Psy-
chology,
86,60-79.
Flin
R,
Mearns
K,
O’Connor
P,
Bryden
R.
(2000).
Measuring safety climate: Identifying
the common features.
Safety Science,
34,
177-192.
Ford
JK,
Quiiiones
MA,
Sego
D,
Sorra
J.
(1992).
Factors
affecting
the
opportunity
to
perform trained tasks
on
the job.
PERSONNELPSYCHOLOGY,
45,511-527.
Griffin MA, Neal
A. (2000).
Perceptions
of
safety at work A framework for linking safety
climate to safety performance.
Journal of Occupational Health Psychology,
5,
347-
358.
Hofmann DA, Jacobs
R,
Landy
R.
(1995).
High reliability process industries: Individual,
micro, and macro organizational influences
on
safety performance.
Journal
of
Safety Research,
26,131-149.
Hofmann DA, Stetzer A.
(1996).
A cross-level investigation
of
factors influencing unsafe
behaviors and accidents.
PERSONNEL PSYCHOLOGY,
49,307-339.
Hofmann DA, Morgeson
FP,
Gerras SJ.
(2001).
Cfimate
as
a moderator of the relationship
between
LMX
and content specific citizenship behavior: Safety climate as an exemplar.
Unpublished manuscript.
Hunt
ST.
(1996).
Generic work behavior:
An
investigation into the dimensions of entry-
level, hourly job performance.
PERSONNEL PSYCHOLOGY,
49,5143.
Hunter
JE. (1983).
A causal analysis of cognitive ability, job knowledge,
job
performance,
and supervisory ratings.
In
Landy F, Zedeck
S,
Cleveland
J
(Eds.),
Performance
measurement and theory
(pp.
257-266).
Hillsdale, NJ: Erlbaum.
International Atomic Energy Agency.
(1996).
Highlights
of
conclusions and recommenda-
tions of “International Conference: One Decade Ajler Chernobyl.
”
Vienna, Austria:
IAEA.
lnternational Atomic Energy Agency.
(1999).
Report on the preliminuly fact fnding mis-
sion following the accident at the nuclear fuel processing facility in Tokaimura, Japan.
Vienna, Austria: IAEA.
International Brotherhood
of
Teamsters.
(1993).
International Brotherhood of Teamsters
hazardous waste worker trainingprogram: Student manual.
Washington,
DC
Inter-
national Brotherhood
of
Teamsters.
International Union
of
Operating Engineers. (undated).
fntemnfional Union of Operating
Engineers national
HA
ZMATprogram: Department of Energy hazardous waste oper-
ations and emergency response trainingguide.
Washington, DC International Union
of Operating Engineers.
jobs.
PERSONNEL PSYCHOLOGY,
43,313-333.
teria and issues.
Performance Improvement Quarter&
5,3-15.
for enquiry and change.
Work
&
Stress,
12,272-284.
of
levels of safety activity.
Work
&
Stress,
12,255-271.
454
PERSONNEL
PSYCHOLOGY
Jacobs
R,
Hofmann DA, Kriska SD.
(1990).
Performance and seniority.
Human Peifor-
munce,
3, 107-121.
Joreskog
KG,
Sorbom D.
(1996).
LISREL
8:
User’s
reference
guide.
Chicago: Scientific
Software International, Inc.
Kahneman
D.
(1973).
Attention and effort.
Englewood Cliffs, NJ: Prentice-Hall.
Kanfer
R,
Ackerman
PL.
(1989).
Motivation and cognitive abilities:
An
integrative/
aptitude-treatment interaction approach to skill acquisition.
Journal
ofApplied
Psy-
chology,
74,657-690.
Komaki J, Heinzmann AT, Lawson L.
(1980).
Effects of training and feedback: Com-
ponent analysis
of
a behavioral safety program.
Journal
of Applied Psychology,
65,
Lee
T,
Harrison K.
(2000).
Assessing safety culture in nuclear power stations.
Safefy
Science,
34,61-97.
Lingard H, Rowlinson
S.
(1997).
Behavior-based safety management in Hong Kong’s
construction industry.
Journal
of
Safefy Research,
28,243-256.
Maier NRF.
(1955).
Psychology in indusby.
Boston: Houghton-Mifflin.
McCormick EJ, Jeanneret PR, Mecham
RC.
(1972).
A study
of
job characteristics and
job dimensions as based
on
the Position Analysis Questionnaire (PAQ).
Journal
of
Applied Psychology,
56,347-368.
McDonald N, Corrigan
S,
Daly
C,
Cromie
S.
(2000).
Safety management systems and
safety culture in aircraft maintenance organizations.
Safety Science,
34, 151-176.
McHenry JJ, Hough LM, Toquam JL, Hanson MA, Ashworth
S.
(1990).
Project Avalidity
results: The relationship between predictor and criterion domains.
PERSONNEL
Motowidlo SJ, Van Scotter JR.
(1994).
Evidence that task performance should be distin-
guished from contextual performance.
Journal
of
Applied Psychology,
79,47540.
Murphy
K.
(2000,
March
12).
Radioactive waste seeps toward
Columbia
River.
tos
An-
geles Times [Online]. Available:
www.latimes.com/news/science/environ/2000312/
t000023730.
html
National Ironworkers and Employees Apprenticeship Training and Journeyman Upgrad-
ing Fund. (undated).
Hazardous
materials training for ironworker contractors and
emergency response training guide.
Washington,
DC
International Association
of
Bridge, Structural and Ornamental Iron Workers.
Neal A, Griffin MA, Hart PM.
(2000).
The impact
of
organizational climate
on
safety
climate and individual behavior.
Safety Science,
34,99-109.
Office
of
Technology Assessment.
(1993).
Hazardsahead: Manugingcleanup workerhealth
and
safety
at
the nuclear weapons complex
(DHHS Publication No. OTA-BP-0-85).
Washington, DC:
U.S.
Government Printing Office.
Oil, Chemical, and Atomic Workers International Union-The Labor Institute. (undated).
Emergency response and prevention workbook: Awareness level
and
operations level.
New York: The Labor Institute.
Pate-Cornell ME.
(1990).
Organizational aspects of engineering system safety: The case
of
offshore platforms.
Science,
250,1210-1217.
Quiiiones MA, Ford JK, Teachout MS.
(1995).
The relationship between work experi-
ence and job performance: A conceptual and meta-analytic review.
PERSONNEL
PSYCHOLOGY,
48,887-910.
Reason J.
(1998)
Achieving a safe culture: Theory and practice
Work& Stress,
12,293-306.
Rudmo
7:
(2000).
Safety climate, attitudes, and risk perception in Norsk Hydro.
Safev
Sackett
PR,
Gruys ML, Ellingson JE.
(1998).
Ability-personality interactions when pre-
Schafer JL.
(1997).
Analysis
of
incomplete multivariate data.
London: Chapman
&
Hall.
261-270.
PSYCHOLOGY,
43,335-366.
Science,
34, 47-59.
dicting job performance.
Journal
of Applied Psychology,
83,545-556.
MICHAEL
J.
BURKE
ET
AL.
455
Schmidt FL, Hunter JE, Outerbridge
AN.
(1986). Impact of job experience and ability
on
job knowledge, work sample performance and supervisory ratings
of
job perfor-
mance.
Journal of Applied Psychology,
71,432-439.
Stout RJ, Salas
E,
Kraiger
K.
(1997). The role of trainee knowledge structures in aviation
team environments.
The International Journal of Aviation Psychology,
7,235-250.
Tesluk PE, Jacobs RR. (1998). Toward an integrated model
of
work experience.
PERSON-
NELPSYCHOLOGY,
51,321-355.
The United Brotherhood
of
Carpenters Health and Safety Fund
of
North
America.
(1992).
Hazardous waste participants munuul.
Cincinnati,
OH:
Midwest Consortium
for
Hazardous Waste Worker Training.
Vineberg R,Taylor EN. (1972).
Petformunce in fourmilitaryjobsbymen ofdifferentaptitude
(AFQT) levels: The relationship
of
AFQT and job experience
to
job performunce.
Alexandria, VA. Human Resources Research Organization.
Weiss HM. (1990). Learning theory and industrial and organizational psychology.
In
Dun-
nette MD, Hough LM (Eds.),
Handbook
of
industrial and
organizationalpsychology:
2nd ed.
(Vol.
1,
pp.171-221). Palo Alto, CA: Consulting Psychologists Press.
Zohar D. (1980). Safety climate in industrial organizations: Theoretical and applied
implications.
Journal
of
Applied Psychology,
65,96-102.
Zohar
II
(2000)
A
group-level model of safety climate: Testing the effect
of
group climate
on microaccidents in manufacturing
jobs
JournuLofApplied Psychology,
85,587-596
Appendix
Factor Definitions and Hypothesized Indicators of the Four-Factor-Model
of the General Safety-Pe$ormance Scale
(GSS)
*
Using
Personal Protective Equipment (UPPE):
Using respiratory
equipment and protective clothing to shield or isolate individuals from
chemical, physical, and biological hazards that may be encountered,
when engineering and work controls are not feasible to control expo-
sure.
1.
Uses the appropriate personal protective equipment as indicated by
2. Correctly inspects and tests all personal protective equipment.'
3.
Dons all personal protective equipment correctly.
4.
Doffs all personal protective equipment correctly.
5.
Correctly stores all personal protective equipment.
6.
When required, properly assists partner in checking, donning, and
removing personal protective equipment or breathing apparatus.
7.
Appropriately communicates with other workers while wearing per-
sonal protective equipment.l
8.
Properly performs work while wearing personal protective equip-
ment.'
9.
Conducts positive and negative pressure tests to ensure proper
fit
of air purifying respirator according to procedures outlined in the
training manual.'
the site health and safety plan.
456
PERSONNEL
PSYCHOLOGY
Engaging
in
Work
Practices
to
Reduce
Risk
(EWPRR):
Performing
tasks to assure safety to those involved, the environment, and the nearby
community using barriers, isolation, equipment, and other methods to
minimize hazards.
10.
Makes appropriate decisions about use of monitoring equipment and
interpretation of instrument readings.'
11.
Correctly uses applicable hazard controls and equipment (e.g., ven-
tilation, physical barriers, remotely operated equipment).
12.
Applies the appropriate work practices to reduce exposures to haz-
ards including applicable SOP'S relating to operations and construc-
tion.'
13.
Takes general precautions and meets permit requirements for con-
fined space work.2
14.
Properly uses lockout/tagout procedures.2
15.
Properly disposes of materials and/or equipment that pose a health
16.
Practices safe spill handling procedures.
17.
Inspects engineering controls as dictated by conditions?
18.
Takes action to avoid contamination.2
19.
Accurately follows established decontamination procedures.
20.
Takes appropriate action to prevent recurrence
of
injuries, illnesses,
accidents, and/or near misses.'
Communicating Health and Safety Information (CHSI):
Communi-
cating hazard, incident, accident, and/or illness information to appropri-
ate personnel.
21.
When necessary, communicates potential exposure(s) to key person-
22.
Contacts appropriate personnel when faced with questions and/or
23.
Appropriately reports incidents, accidents, and/or illnesses.
24.
Engages in the appropriate methods to notify workers, supervisors,
and/or emergency coordinators of emergency conditions.
Exercising Employee Rights and Responsibilities
(EERR):
Exercis-
ing rights and responsibilities pertaining to laws and regulations.
25.
Appropriately uses MSDS and other reference materials that may
provide additional health and safety information (e.g., the National
Institute for Occupational Safety and Health Pocket Guide,
DOT
Handbook, Emergency Response Guide, 4-Agency Manual, and
CHRIS
Manual).2
26.
When necessary, exerts the employees rights and responsibilities to
access and provide input into altering the site HASP?
risk.
'
nel responsible for site health and safety.'
issues regarding HAZWOPER.2
MICHAEL
J.
BURKE
ET
AL.
457
27.
Takes the appropriate steps if prevented from or punished
for
exer-
cising one’s rights under
OSHA
and
DOE
policies and procedures.2
Nore.
SOP
=
Standard Operating Procedure; HAZWOPER
=
Hazardous Waste
Op
erations and Emergency Response; MSDS
=
Materials Safety Data Sheet; DOT
=
De-
partment of llansportation; CHRIS
=
Chemical Hazard-Response Information System;
HASP
=
Health and Safety Plan;
OSHA
=
Occupational Safety and Health Administra-
tion; DOE
=
Department of Energy.
zMore context-specific item that may require editing
or
consideration of relevance for
a
particular type of safety-related
work.
Generic item, applicable to almost all types
of
safety-related
work;
