Ergonomics Climate Assessment
Ergonomics Climate Assessment
Ergonomics Climate Assessment
Applied Ergonomics
journal homepage: www.elsevier.com/locate/apergo
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 7 May 2014
Accepted 11 March 2015
Available online
Ergonomics interventions have the potential to improve operational performance and employee wellbeing. We introduce a framework for ergonomics climate, the extent to which an organization emphasizes and supports the design and modication of work to maximize both performance and well-being
outcomes. We assessed ergonomics climate at a large manufacturing facility twice during a two-year
period. When the organization used ergonomics to promote performance and well-being equally, and
at a high level, employees reported less work-related pain. A larger discrepancy between measures of
operational performance and employee well-being was associated with increased reports of workrelated pain. The direction of this discrepancy was not signicantly related to work-related pain, such
that it didn't matter which facet was valued more. The Ergonomics Climate Assessment can provide
companies with a baseline assessment of the overall value placed on ergonomics and help prioritize
areas for improving operational performance and employee well-being.
2015 Elsevier Ltd and The Ergonomics Society. All rights reserved.
Keywords:
Ergonomics climate
Well-being
Performance
1. Introduction
To maximize success, organizations must have a climate that
supports operational performance (e.g., production efciency,
product or service quality) as well as employee well-being (e.g.,
health and safety). Researchers (e.g., Aryee et al., 2012; Pousette
et al., 2008) have examined operational performance and
employee well-being climates individually; however, few studies
have considered how to simultaneously promote a climate for both
performance and well-being. One comprehensive way to demonstrate both of these values is by promoting a systems approach as is
frequently done in ergonomics. The eld of ergonomics aims to
increase efciency by designing or modifying the job to eliminate
non-value added processes and hazards that increase the risk of
employee injury (Wickens et al., 2004). Professionals who use ergonomic principles adapt work tasks to the physical and mental
capabilities of the workers. Implementing ergonomic principles in
an occupational environment can directly benet the worker and
the organization by reducing physical and mental strain, lowering
the risk of occupational related injuries and illnesses and improving
* Corresponding author. Present address: Sentis, 116 Inverness Dr. E., Englewood,
CO, 80112, USA. Tel.: 1 719 459 1288.
E-mail address: krista.hoffmeister@gmail.com (K. Hoffmeister).
http://dx.doi.org/10.1016/j.apergo.2015.03.011
0003-6870/ 2015 Elsevier Ltd and The Ergonomics Society. All rights reserved.
161
reect a company's simultaneous value for both employee wellbeing and performance. Ergonomics climate also relates to
employee safety; however, ergonomics climate is a more robust
construct that reects broader values. Ergonomics climate reects a
perceived value for employee well-being, which includes worker
health and safety, quality of work life, and job satisfaction. Ergonomics climate also reects the perceived value that management
places on operational performance. Thus, employees responding to
ergonomics climate measures will be asked to refer to practices that
affect both their well-being and operational performance.
Therefore, we propose ergonomics climate as one example for
how an organization may seek to balance performance and wellbeing objectives. It is possible that an organization might use ergonomics principles to improve one, but not both, of these facets. In
doing so, they do not take full advantage of ergonomic principles.
Thus, we expect the best outcomes for organizations that value
both facets, consistent with the holistic, systems-oriented philosophy of ergonomics.
2. Qualitative analysis of ergonomics climate
If organizations that foster a positive and balanced ergonomics
climate should reap the greatest outcomes from ergonomics initiatives, then organizations who use (or intend to use) such initiatives should benet from assessing, tracking, and understanding
their ergonomics climates. We therefore sought to develop a
measure for ergonomics climate. We began with two focus groups
(N 12) at a manufacturing facility to discuss how an understanding of the relative priorities of performance and employee
well-being might be useful to the company. To analyze the qualitative data from the focus groups, we used open coding and content
analysis processes (Janis, 1965; Weber, 1990; Woike, 2009). Interrater agreement of the measure was not assessed, because only one
researcher coded the responses. The coder read transcripts to
identify common themes that occurred while the group discussed
what an ergonomics climate could look like. This resulted in a total
of eight core themes extracted from the focus groups (Table 1). We
then outlined, from a range of resources, the best practices in
occupational ergonomics. The resources included a review of the
National Institute for Occupational Safety and Health's (NIOSH)
Elements of an Ergonomics Program, the Occupational Safety and
Health Administration's (OSHA) Voluntary Protection Program, the
rescinded federal ergonomics standard promulgated by OSHA, and
state ergonomics guidelines (Washington and California).
Following this review, we examined research literature related to
ergonomics, occupational health and well-being, productivity, and
organizational climate, as outlined in Table 1. The combined results
from the focus groups, national resources and literature review
revealed four common factors central to organizational climates
that involve ergonomics: management commitment, employee
involvement, job hazard analysis, and training and knowledge (see
Table 1). These four factors have been identied many times in a
range of literatures, with the most relevant being the safety climate
literature. Although researchers have yet to agree on a single factor
structure of safety climate (Cox and Cheyne, 2000; Coyle et al.,
1995; Vinodkumar and Bhasi, 2009), many measures share the
common factors of management commitment (e.g., Brown and
Holmes, 1986; Cooper and Phillips, 2004; Grifn and Neal, 2000;
Probst, 2004; Zohar, 1980), employee involvement (Cheyne et al.,
1998; Cox and Cheyne, 2000; Dedobbeleer and Beland, 1991; Neal
and Grifn, 2004), job hazard analysis (Cheyne et al., 1998; Cox
and Cheyne, 2000; Grifn and Neal, 2000; Probst, 2004), and
training and knowledge (e.g., Cooper and Phillips, 2004; Evans
et al., 2007; Grifn and Neal, 2000; Lu and Tsai, 2008). Thus, we
acknowledge that the four factors described here are similar to
162
Table 1
Sources of the four components of ergonomics climate.
Source
Management commitment
Employee involvement
Management commitment
Value of ergonomics
Literature
Assessments
Monitoring the effectiveness
of the ergonomics program
Reporting system
OSHA Job Hazard
Analysis Guidelines
Federal OSHA
Ergonomics Standard
WA Ergonomics Standard
CA Ergonomics Standard
NIOSH Elements
of an Ergonomics Program
OSHA VPP
Chao and Henshaw (2002)
Cheyne et al. (1998)
Cox and Cheyne (2000)
Grifn and Neal (2000)
Morag and Luria (2013)
Munir et al. (2007)
Probst (2004)
Employee knowledge
and training
Communication
Employee Involvement
Value for ergonomics
Reporting system
Federal OSHA
Ergonomics Standard
NIOSH Elements
of an Ergonomics Program
OSHA VPP
Federal OSHA
Ergonomics Standard
WA Ergonomics Standard
CA Ergonomics Standard
NIOSH Elements
of an Ergonomics Program
OSHA VPP
163
5. Methods
5.1. Participants and procedure
The Ergonomics Climate Assessment was rst piloted on 130
employees at an international manufacturer of earth moving
equipment that consisted of both ofce and production employees.
Employees were divided between multiple departments and multiple buildings on the facility's campus. Employees in each
department were responsible for assembling various parts of the
nal product, and each department had their own team supervisor.
Respondents were mostly Caucasian (84%) males (98%) with an
average age of 33 years (SD 11 months). One month following the
pilot survey, 1593 employees from the same manufacturing company were invited to participate in the survey (Sample 1). We
received 1031 responses (65% response rate), of which 84% were
line employees and 16% were ofce employees. The majority of
those surveyed were Caucasian (86.8%) and male (91%) with an
average age of 37 years (SD 11 months). Line employees were
surveyed in person during normal work hours, and ofce employees were provided with an online link to the survey.
One year later, the same manufacturing facility was surveyed
again using the same methodology. A total of 706 employees
responded, resulting in a response rate of 53% (Sample 2). Of those
that responded, 87% were line employees and 13% were ofce
employees. Most of those surveyed were Caucasian (85%) and male
(90.6%) with an average age of 38 years (SD 12 months). Due to
concerns about condentiality expressed by the management and
employee representatives, we were not able to link responses at the
individual level from Survey 1 to Survey 2. Although the samples
are not independent, it is likely that they are also not completely
overlapping, and so the samples were analyzed separately to provide additional insight.
5.2. Measures
The rst version of the Ergonomics Climate Assessment consisted of 20 parallel items per each of the four core factors (e.g., 10
management commitment for PE and 10 management commitment for WE), resulting in 80 questions total. Participants were
asked to rate the extent to which they agreed or disagreed that the
survey statements were accurate for their department on a scale of
1 (strongly disagree) to 5 (strongly agree). Pilot testing (results
described below) allowed us to reduce the Ergonomics Climate
Assessment to a 40-item measure that was used in the subsequent
surveys. All survey items are provided in the supplementary
materials.
To measure self-reported pain, individuals were also asked if
they had experienced any work-related pain in the past 12 months
in nine different areas of their body. Each of these nine items was a
dichotomous variable recorded as yes (1) or no (0), and a total pain
score (ranging from 0 to 9) was calculated by summing the number
of body parts reported to be in pain. The self-reported pain survey
was similar to other surveys used in epidemiological research
among occupational populations (Dickinson et al., 1992;
Rosecrance et al., 2002).
6. Results
6.1. Pilot study results
The pilot study allowed us to identify the most informative of
the original survey items based on preliminary factor analyses,
following recommendations from Mcdonald (1999). The proposed 4-factor models fell somewhat short of conventional
164
Table 2
Means, standard deviations, and correlations.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
21
Variable
M(Sample 2)
SD (Sample 2)
Wage or Salary
Age
Gender
Time with Supervisor (Months)
Time with Company (Months)
PE Climate
WE Climate
PE e Management Commitment
PE e Employee Involvement
PE e Job Hazard Analysis
PE e Training & Knowledge
WE e Management Commitment
WE e Employee Involvement
WE e Job Hazard Analysis
WE e Training & Knowledge
Pain
e
37.54 (38.21)
e
8.48 (8.52)
68.57 (71.21)
3.36 (3.25)
3.34 (3.22)
3.35 (3.21)
3.32 (3.22)
3.42 (3.34)
3.33 (3.22)
3.31 (3.17)
3.33 (3.25)
3.35 (3.26)
3.35 (3.21)
2.15 (2.24)
e
11.60 (11.96)
e
13.58 (13.00)
96.40 (103.88)
.90 (.84)
.93 (.90)
1.01 (.94)
.98 (.92)
.92 (.87)
.93 (.88)
1.04 (1.00)
.96 (.93)
.97 (.95)
.94 (.91)
2.48 (2.24)
e
.05
.17
.18
.16
.23
.23
.21
.20
.23
.23
.23
.23
.22
.22
.20
.16
e
0
.13
.49
.09
.09
.05
.03
.08
.08
.10
.08
.12
.14
.03
.11
.01
e
.04
.05
.11
.10
.08
.07
.10
.09
.11
.10
.12
.11
.09
.12
.13
.04
e
.26
.01
.02
.01
.00
.00
.01
.02
.02
.02
.03
.07
.01
.49
.08
.22
e
.11
.10
.08
.06
.11
.10
.10
.11
.13
.12
.09
.33
.14
.11
.06
.17
(.97)
.96
.93
.90
.94
.91
.96
.92
.95
.92
.37
.31
.13
.10
.05
.16
.96
(.98)
.89
.94
.89
.95
.91
.96
.92
.96
.37
Variable
10
11
12
13
14
15
16
Wage or Salary
Age
Gender
Time with Supervisor (Months)
Time with Company (Months)
PE Climate
WE Climate
PE e Management Commitment
PE e Employee Involvement
PE e Job Hazard Analysis
PE e Training & Knowledge
WE e Management Commitment
WE e Employee Involvement
WE e Job Hazard Analysis
WE e Training & Knowledge
Pain
.30
.12
.10
.08
.15
.93
.89
(.91)
.93
.82
.82
.85
.82
.83
.81
.35
.32
.14
.09
.02
.17
.92
.87
.80
(.91)
.81
.85
.83
.86
.82
.84
.36
.33
.11
.12
.03
.16
.95
.90
.85
.81
(.91)
.91
.86
.85
.87
.84
.33
.29
.14
.12
.09
.15
.94
.90
.81
.83
.87
(.91)
.86
.90
.87
.89
.35
.27
.11
.01
.07
.14
.89
.94
.93
.77
.82
.80
(.93)
.91
.91
.88
.35
.30
.14
.07
.04
.17
.90
.93
.81
.90
.83
.85
.83
(.91)
.89
.92
.36
.32
.12
.10
.02
.14
.91
.96
.83
.80
.90
.85
.87
.85
(.93)
.93
.36
.28
.14
.11
.06
.16
.91
.95
.80
.81
.86
.92
.82
.85
.90
(.92)
.35
.21
.03
.01
.02
.05
.31
.32
.28
.30
.28
.29
.29
.30
.33
.31
e
Note. Correlations greater than or equal to .08 are signicant at the p < .05 level.
Sample 1 correlations are on the lower diagonal, Sample 2 correlations are on the upper diagonal. Average reliabilities across samples are reported on the diagonal; across
years all reliabilities were within .02 of each other.
PE Performance-focused ergonomics climate, WE Well-being-focused ergonomics climate.
Table 3
Model t statistics for performance and well-being.
Description
Pilot
1 general factor
2 facet factors
PE 1 factor
WE 1 factor
PE 4 factor
WE 4 factor
Sample 1
1 general factor
2 facet factors
PE 1 factor
WE 1 factor
PE 4 factor
WE 4 factor
Sample 2
1 general factor
2 facet factors
PE 1 factor
WE 1 factor
PE 4 factor
WE 4 factor
165
Chi-square
df
CFI
RMSEA
e
e
410.09
492.95
355.47
413.69
e
e
170
170
164
164
e
e
.87
.85
.90
.88
e
e
.10
.12
.10
.11
8129.78
7995.85
1691.93
1802.46
1226.87
1335.57
740
739
170
170
164
164
.84
.84
.92
.92
.94
.94
.10
.10
.10
.10
.08
.09
5981.09
5912.46
1212.50
1512.23
912.97
1146.12
740
739
170
170
164
164
.83
.83
.91
.90
.94
.93
.10
.10
.09
.10
.08
.09
Highest
factor r
Dc2
1.00
1.00
54.62**
79.26**
.80
133.93**
1.00
1.00
116.61**
66.39**
.71
68.63**
.98
.98
180.46**
133.34**
Note. The tests for general factor and 2 facet factor models for the pilot data resulted
in a non-positive denite rst-order derivative product matrix; thus, results are not
presented here. PE Performance facet of ergonomics climate, WE Well-being
facet of ergonomics climate.
**p < .01.
Table 4
Polynomial regression results for work-related pain.
Model
Variable
F-Value
df
R-Square
SE
Beta
t-value
Sample 1
Constant
Wage or Salary
Time with Sup
Tenure
PE
WE
PE2
PE x WE
WE2
24.35**
.17
3.26
1.02
.02
.00
.32
.62
.57
1.49
1.02
.27
.23
.01
.00
.32
.30
.44
.74
.39
.14
.08
.07
.12
.23
.28
.72
.51
12.11**
4.49**
2.54*
2.17*
1.01
2.06*
1.30
2.02*
2.65**
Constant
Wage or Salary
PE
WE
PE2
PE x WE
WE2
15.07**
3.22
.75
.11
.79
.33
1.11
.77
.30
.25
.35
.33
.58
.88
.39
.12
.04
.31
.15
.51
.38
10.89**
3.01**
.32
2.42*
.57
1.26
1.97*
Sample 2
.13
166
Table 5
Response Surface Tests.
Model
Surface test
Slope of agreement
Curvature of agreement
Curvature of disagreement
Slope of disagreement
a1
a2
a3
a4
Coefcient
.94**
.11
.30
3.08**
SE
.10
.87
.61
.10
Slope of agreement
Curvature of agreement
Curvature of disagreement
Slope of disagreement
a1
a2
a3
a4
.67**
.00
.90
2.21
.11
.08
.67
1.14
Note. a1 (b1 b2), where b1 is beta coefcient for performance-focused ergonomics climate (PE) and b2 is beta coefcient for well-being-focused ergonomics climate (WE).
a2 (b3 b4 b5), where b3 is beta coefcient for PE squared, b4 is beta coefcient for the cross-product of PE and WE, and b5 is beta coefcient for WE squared.
a3 (b1 b2), a4 (b3 b4 b5), SE standard error.
**p < .01 *p < .05 p .054.
167
effectiveness of organizational efforts focused on improving ergonomics as well as a measure to assist in the prioritization of ergonomic interventions during periods of limited resources.
7.1. Limitations and suggestions for further research
This study is based on cross-sectional and self-reported data,
and included only a single outcome measure. As an initial step in
the development of this measure, this outcome provided evidence
needed to move forward with this framework and test more
detailed models. Other outcomes of employee well-being such as
the quality of work life and presenteeism, and operational
performance-based outcomes such as production efciency, product quality, and errors or deviations in tolerances should be added
to subsequent models. Nevertheless, based on the results of the
present study, it appears that the Ergonomics Climate Assessment
may be very useful in understanding how an agreement or
disagreement between the PE and WE facets of ergonomics climate
is related to a variety of individual and organizational outcomes.
The results of this study also suggest that the PE and WE facets
were very highly correlated. Although it was most appropriate to
combine the sub-factors within PE and WE facets for the purposes
of analyzing our particular research questions, additional research
should further examine the model by which ergonomics operates
in organizations and the relationship between PE and WE facets.
Our results also suggested that ergonomics climate was best
assessed at the individual level for our sample, because of a lack of
between-group variation. Together, these results suggest that PE
and WE facets were both equally acknowledged within this organization, and that this perception was facility-wide as opposed to
varying across departments. Therefore, the use of this particular
organization, which had a longstanding tradition of highly visible
ergonomics programs targeting both performance and well-being,
could have inuenced the results. Future research should examine
ergonomics climate within companies that may not place a strong
and equal emphasis on both PE and WE facets.
Additional research is also needed to further investigate the
associations between the Ergonomics Climate Assessment tool and
individual and organizational outcomes. For example, it appears
logical that when an organization focuses primarily on operational
performance, the value for well-being decreases, resulting in
increased pain. However, it is less clear why an increased focus on
well-being (over performance) would result in increased complaints of pain. We proposed that lack of adoption of ergonomic
improvements or work-related stress might help explain this result,
but researchers should explore and test these mechanisms further.
Furthermore, it is possible that the performance and/or well-being
facets of ergonomics climate could be differentially related to other
organizational outcomes such as quality, productivity, and other
aspects of workplace health and safety. An understanding of these
relationships may help organizations recognize the importance of
valuing both performance and well-being facets of ergonomics
climate.
7.2. Conclusion
The Ergonomics Climate Assessment is a practical tool that researchers and practitioners can use to understand the value for
performance and well-being that exists within an organization and
to help identify areas for improvement, as well as a method to
benchmark ergonomic process goals. Our ndings demonstrated
that the framework of ergonomics climate can be used for
designing and modifying work to improve both operational performance and employee well-being. Furthermore, our results
indicate that it is not only possible, but also preferred, for
168
169