Applied Ergonomics 78 (2019) 148–156

Contents lists available at ScienceDirect

Applied Ergonomics
journal homepage: www.elsevier.com/locate/apergo

Employee acceptance of wearable technology in the workplace T

a,b,∗ a a,c a
Jesse V. Jacobs , Lawrence J. Hettinger , Yueng-Hsiang Huang , Susan Jeffries ,
Mary F. Lescha, Lucinda A. Simmonsa, Santosh K. Vermaa, Joanna L. Willettsa
a
Liberty Mutual Insurance, 157 Berkeley St, Boston, MA, USA
b
Rehabilitation and Movement Science, University of Vermont, 106 Carrigan Dr, Burlington, VT, USA
c
Oregon Institute of Occupational Health Science, Oregon Health & Science University, 3181 SW Sam Jackson Park Rd, Portland, OR, USA

ARTICLE INFO ABSTRACT

Keywords: Wearable technology has many industrial applications. Optimal use adherence and outcomes largely depend on
Technology acceptance employee acceptance of the technology. This study determined factors that predict employee acceptance of
Wearable technology wearables. An online survey of 1273 employed adults asked about demographics, job and organizational
Safety climate characteristics, experience with and beliefs about wearables, and willingness to use wearables. Use cases focused
Industrial workplace ergonomics
on workplace safety elicited the highest acceptance. An employee's performance expectancy and their organi-
Work measurement
zational safety climate were common predictors of acceptance across use cases. Positive past experiences co-
incided with involving employees in choosing the device and adequately informing them about data use.
Organizations intending to implement wearable technology should (a) focus its use on improving workplace
safety, (b) advance a positive safety climate, (c) ensure sufficient evidence to support employees' beliefs that the
wearable will meet its objective, and (d) involve and inform employees in the process of selecting and im-
plementing wearable technology.

1. Introduction
Wearable technology products are advanced sensor and computing
technologies that a person can wear on their body during daily activity
to generate, store, and transmit data. Use cases for wearable technology
in the workplace are being developed to, for example: (a) monitor
productivity, (b) identify and intervene on safety hazards or risks, (c)
provide augmented instruction to improve task management, and (d)
facilitate health and wellness (Kalantari, 2017). A key concern with
adopting wearable technology in the workplace is employee acceptance
and adherence to its use (Schall et al., 2018). Therefore, the primary
objective of this study is to identify factors that predict employee ac-
ceptance of work-related wearable technology. Employee acceptance is
represented by the participants’ reported willingness to use work-re-
lated wearable technology in the use cases defined in the survey. The
secondary objective of this study is to identify factors related to a po-
sitive experience with work-related wearable technology among parti-
cipants who already had such experience. This study uniquely con-
tributes to the literature through its combined focus on factors related
to the use case of the wearable technology and the workplace en-
vironment in which it is used, in addition to employee characteristics
and beliefs. Further, rather than focusing on a specific industry, this
study samples a broad spectrum of employees from across the United
States. The sampling provides novel and comparably generalizable in-
sights for how to enhance the likelihood of a well-accepted wearable
technology program in the workplace.
Survey instruments are often employed to test hypothesis-driven
structural models for constructs that associate with attitudes about
technology, people's intent to use the technology, and actual use be-
havior. Perhaps the most developed of these models for testing the
acceptance of workplace technology is the Unified Theory of
Acceptance and Use of Technology (UTAUT) model (Venkatesh et al.,
2003; Williams et al., 2015). The model suggests that use behavior is
predicted by facilitating conditions and a person's behavioral intention
to use technology. Behavioral intention is then predicted by constructs
of performance expectancy (belief that the system will meet its objec-
tive or help attain gains in their job), effort expectancy (ease of use),
and social influence (perception that others of importance believe the
individual should use the system). In parallel, gender, age, experience
with the technology, and voluntariness of use all mediate the re-
lationships of these constructs with intended and actual use. The model
has been validated across multiple studies evaluating varied types of
technology, and the strongest predictor of use is behavioral intention
via performance expectancy (Williams et al., 2015). For example,

∗
Corresponding author. Liberty Mutual Insurance, 157 Berkeley Street, 4C, Boston, MA, 02116, USA.
E-mail address: jesse.jacobs@libertymutual.com (J.V. Jacobs).

https://doi.org/10.1016/j.apergo.2019.03.003
Received 5 November 2018; Received in revised form 8 January 2019; Accepted 4 March 2019
Available online 13 March 2019
0003-6870/ © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/BY/4.0/).
J.V. Jacobs, et al.
people will be more likely to use a fitness band if they believe it will
help them improve their fitness. Other acceptance models have also
included concepts of trust (e.g., in the safety, security, and use of the
device and its data) and innovativeness (e.g., an individual's propensity
to use and accept any novel technology) (Gribel et al.,2016; Kwee-
Meier et al., 2016; Pfeiffer et al., 2016; Weng, 2016). Thus, the existing
literature has identified several factors to predict the acceptance of
wearable technology.
It is not the intent of this study to formally test or extend these
models, because many other studies have already applied them to
wearable technology (Abbasi et al., 2015; Choi et al., 2017; Fensli et al.,
2008; Gao et al., 2015; Hong et al., 2017; Huang and Lai, 2016; Hwang,
2014; Kim and Shin, 2015; Kwee-Meier et al., 2016; Lee, 2009; Lin
et al., 2016; Lunney et al., 2016; Pfeiffer et al., 2016; Rauschnabel and
Ro, 2016; Spagnolli et al., 2014; Turhan, 2013; Yang et al., 2016). In-
deed, these models have been criticized for the heuristic nature of their
structure as well as for their historically narrow scope to focus on
employee beliefs and characteristics (Benbasat and Barki, 2007;
Lunceford, 2009). These criticisms thus suggest a need for an alter-
native modeling approach that is designed to analyze considerations
beyond employee characteristics and beliefs. Other needed constructs
include task-related factors (e.g., the nature and purpose of the wear-
able data collection and the employees' job function) and factors related
to the organizational setting (e.g., safety climate, job security, and
worker's union membership).
When evaluating studies on wearable-technology acceptance in the
workplace, previous results have supported a multifactorial systems
model of acceptance. Specifically, in addition to employee beliefs, ac-
ceptance may differ based on cost, type of device, relationship of the
user to those who evaluate the data, as well as the employee's job type,
position in the company, and experience with wearable technology
(Choi et al., 2017; Kwee-Meier et al., 2016; Schall et al., 2018; Spagnolli
et al., 2014). These previous studies highlight the importance of a
systems perspective when determining the acceptance of wearable
technology; however, they are based on very specific sample char-
acteristics and use cases. Thus, research is needed to further examine
more generalizable use cases and environmental features on a broad-
spectrum sample of employees. To address this need, this study seeks to
explore how factors related to the organizational setting, the individual
employee, and the purpose or use case of the wearable technology as-
sociate with attitudes toward its use in the workplace. We hypothesized
that acceptance would (a) significantly differ by use case, and (b) be
best predicted from a combination of environmental factors with factors
of employee characteristics and beliefs.
2. Methods
2.1. Participants
Participants were recruited by email through an independent online
survey system (SurveyGizmo; www.surveygizmo.com). Individuals had
opted into the survey system's participant network and consented to
receive such emails. Data collection occurred from February 24th to
March 29th, 2017. Inclusion criteria required participants to be at least
Table 1
Use-case scenarios to assess willingness to use wearable technology at work.
Use Case Function
1 Record activity and physical status
2 Record activity and physical status
3 Record activity and physical status
4 Record activity and physical status
5 Record activity and physical status
6 Monitor environmental hazards
7 Provide information on work process
149
Applied Ergonomics 78 (2019) 148–156
18 years of age and at least part-time employed. Because online surveys
risk bias, we requested that the survey service set recruitment quotas to
equally represent the job functions inquired about in our survey. We
also requested recruitment quotas on demographics of gender, age,
race, and ethnicity to represent the demographics of the United States
employed population. Participant profiles are updated annually within
the survey service's participant networks, and the answers to our survey
verified the profiles' accuracy. Participants were recruited from net-
works in the United States and received points as a means of re-
imbursement. Points could be redeemed for money or for gift cards per
the pre-negotiated agreement between the survey service and their
participant networks. The service screened for “straight-line” re-
sponding, timed out attempts, and other signals of invalid responses,
and then backfilled the sample to meet quotas. Participants could re-
spond to the survey only after an informed consent process, which was
approved by the Institutional Review Board.
2.2. The survey
The survey (provided in the Appendix) was developed by the project
team. The survey content was guided by systems and sociotechnical
systems theories (Mele et al., 2010; Robertson et al., 2015) to include
items already identified in the literature as valuable predictors of
technology acceptance as well as other hypothesized items of interest
derived from conversations with industrial subject-matter experts. The
survey included Likert scales, ranking, and open-response formats to
ascertain the participants' characteristics, their employers' organiza-
tional characteristics, past experience with wearable technology, and
their opinions about wearable technology in the workplace. Because the
participants were anticipated to have differing levels of experience with
wearable technology, their understanding of the technology was fa-
cilitated by presenting (a) a definition of the technology; (b) a test
question to identify whether certain examples represented wearable
technology, which was followed by a clarifying explanation about the
correct answer; and (c) a concrete scenario for the participants' con-
sideration when responding to questions about willingness (Appendix).
Responses to questions regarding confidence that wearables would
improve the participants’ safety, productivity, or health were con-
sidered to reflect their performance expectancy. Innovativeness was
determined by the level of agreement that they are the first to try new
technology among their peers. Trust was represented by items re-
garding levels of concern over data security and misuse. Safety climate
was represented by the short scale defined by Huang and colleagues.
This scale was found reliable and highly correlated (Pearson r of at least
0.95) with the original safety climate scales known to be predictive of
important safety outcomes (Huang et al., 2017).
To address our primary objective, we asked participants to state
their level of agreement on a 5-point ordinal scale (strongly disagree to
strongly agree) about whether they would voluntarily use a wearable
under seven use cases that varied the function, location, and purpose of
the data collection (Table 1). To address our secondary objective,
participants were asked to state on a 5-point ordinal scale (very bad to
very good) the quality of their prior experience using wearables for
work-related or work-sponsored purposes.
Location Purpose
Only at work Monitor and improve safety
In and outside of work Monitor and improve safety
Only at work Monitor and improve productivity
In and outside of work Monitor and improve productivity
In and outside of work Monitor and improve health and fitness
Only at work Alert for hazard proximity or level
Only at work Instruct how and when to perform tasks
J.V. Jacobs, et al. Applied Ergonomics 78 (2019) 148–156

Fig. 1. Sample Characteristics. Distributions of the sample by (A) state – 47 of 50 states elicited sample; (B) job function; (C) age bracket; (D) sex; (E) race and
ethnicity; (F) education attained. For C-F, gray bars represent the study sample, and black bars represent the employed population of the United States in 2017 (year
of data collection) according to the Bureau of Labor Statistics.

2.3. Data processing and analysis
To ensure our target variables of willingness and quality of prior
experience had sufficient sample for each level of response, we binned
the responses from a 5-point ordinal scale to binary scales of positive
versus neutral-or-negative responses (i.e., scores greater than 3 versus
scores of 3 or less). McNemar's tests of proportions were used to de-
termine whether acceptance differed across use cases. To facilitate
running only one single-variable analysis, rather than on each use case,
responses to the questions on willingness were averaged across sce-
narios prior to binning. Ordinal scales of the predictor variables were
also binned for single-variable analysis. We report the odds ratios of a
positive willingness or a positive prior experience for participants with
versus without a predictor characteristic. Fisher's Exact tests or Chi-
Square tests were used to determine the statistical significance of the
relationship.
Once significant individual predictors were identified, the unbinned
responses to the significant predictor variables were entered into
elastic-net generalized linear models (GLMs) using R software packages
(e.g., glmnet). Both the binned and unbinned versions were tested, but
performance improved with the unbinned predictors, so we report on
those models. Modeling included the following steps: (a) random down
sampling of over-represented target-variable responses to balance the
data set and prevent model overfitting; (b) stratified random sampling

150
J.V. Jacobs, et al.
Fig. 2. The percentage of the participants that identified a positive willingness
(score of 4 or 5 on a 5-point ordinal scale) to use wearable technology at work
for each use case.
to split 80% of the data for model training and 20% of the data as hold-
out data for model testing; (c) 4-fold cross validation on the training
data to optimize the lasso regularization; and (d) model testing on the
hold-out sample. A stable random seed was identified to ensure com-
parability across models, and the robustness of the models was tested by
repeating the process across 5 unique random seeds. We evaluated the
data with and without down sampling, and model performance on hold-
out data was best after balancing the training data; we therefore report
on the findings with down sampling. Focus was placed on optimizing
the lasso regularization, rather than the ridge regularization, because
the intent was to identify the most efficient complimentary set of in-
dependent predictors, rather than keeping all predictors in the model
with diminished coefficients. The lasso penalty selects relatively poorly
predictive features out of the model.
The area under the receiver operating characteristic curve (AUC)
was utilized to assess model performance. This metric represents the
level at which model performance balances the false positive rate (1-
specificity) and the true positive rate (recall or sensitivity); perfect
performance is represented by a value of 1.0, and chance performance
is represented by a value of 0.5. Because the AUC can provide biased
results on unbalanced samples, we also report the F1 score. The F1
score is the harmonic mean of recall (defined above) and precision (also
called the positive predictive value). We report both the best-per-
forming model and the most-efficient model. The most efficient model
contained the fewest number of predictors that elicit cross-validation
performance within a standard error of the best-performing model's
performance. These models were generated to test the average will-
ingness across all use cases as well as to test the willingness of each of
the seven use cases.
When modeling our secondary objective on the participants’ quality
of prior experience with workplace wearables, the sample size was
significantly reduced to only those participants who had prior experi-
ence. The sample was also heavily biased to a positive prior experience.
The sample, therefore, did not allow for a robust predictive analysis, so
the results on the quality of prior experience are derived from a 5-fold
cross validation on all available data without a hold-out sample. Thus,
the secondary analysis on the quality of prior experience with work-
place wearables represents a relational analysis rather than a predictive
analysis.
151
Applied Ergonomics 78 (2019) 148–156
3. Results
3.1. Sample characteristics
We received 3048 responses, of which 163 were immediately dis-
qualified due to lack of consent; 1406 were not analyzed because the
respondents didn't meet inclusion criteria or because responses were
identified as invalid by the survey service. For example, median re-
sponse time for the invalid sub-sample was 1.5 min compared to
9.8 min for the valid responses. Finally, 206 samples were not analyzed
due to partial completion, generally of only the initial demographic
questions. Therefore, a total of 1273 participants were analyzed with
consented, valid responses.
Fig. 1 illustrates the geographic distribution, job types, and demo-
graphic characteristics of the sample. Attempts to use sampling quotas
to mimic the United States’ employed population (according to the
Bureau of Labor Statistics tables for 2017; https://www.bls.gov/cps/
demographics.htm) elicited directionally similar demographic dis-
tributions, however, some statistical variations existed. Specifically,
people 25–34 years old were over-sampled, and people 65 years or
older were under-sampled (chi-squared = 120, p < 0.001); people
with some college experience were over-sampled, and people with less
than a high-school diploma were under-sampled (chi-squared = 6.59,
p = 0.010); females were under-sampled (chi-squared = 5.15,
p = 0.023); and people of Hispanic or Latino ethnicity were under-
sampled (chi-squared = 8.42, p = 0.0037). The proportions of White,
African-American, and Asian participants were not statistically different
from those reported by the Bureau of Labor Statistics (chi-
squared = 3.51, p = 0.17). The Bureau did not report on every race
category collected by the study, so comparisons focused only on those
three categories.
In addition to the features illustrated in Fig. 1, 56% of the partici-
pants identified themselves as frontline workers or individual con-
tributors, 21% as direct supervisors of frontline workers or individual
contributors, and 23% as middle or executive managers. Twelve per-
cent had been employed with their current organization for less than
one year, 38% for 1–5 years, 23% for 6–10 years, and 27% for greater
than 10 years.
3.2. Comparing employee acceptance across use cases
Qualitatively, use cases focused on safety had a higher proportion of
positive willingness to use wearables than those focused on pro-
ductivity. In addition, use cases in which data are only collected at work
also had a higher proportion of positive willingness to use the wear-
ables than those for which data are collected in and out of work (Fig. 2).
Statistically, identifying environmental hazards for safety purposes
elicited a higher proportion of positive willingness than all other use
cases, and monitoring activity or status in and out of work for either
safety or productivity elicited significantly lower proportions of posi-
tive willingness than the other use cases (p-value < Bonferroni-cor-
rected threshold of 0.0024).
When asked to identify their most-willing and least-willing use
cases, the participants most often selected the scenario of monitoring
activity or status at work for safety as their most-willing use case (31%
versus 6–18% for each of the other use cases), and the participants most
often selected the scenario of monitoring activity or status in and out of
work for productivity as their least willing use case (29% versus 6–16%
for each of the other use cases). Participants indicated they would de-
sire the following level of incentive for their most-willing and least-
willing use cases, respectively: median = 100 versus 200 USD with a
95%-trimmed mean = 255 versus 822 USD (Wilcoxon Signed-Rank
Z = 11.23, p < 0.001). Further, the percentage of participants willing
to use the wearable for no incentive was 12.4% for their most-willing
use case versus 9.6% for their least-willing use case (McNemar's Test
statistic = 7.15, p = 0.008).
J.V. Jacobs, et al.
Table 2
Single-variable analysis of factors related to willingness to use wearables.
Factor Significantly Related to Positive % of Sample with
Willingness to Use Wearables (Fisher's Exact or Attribute
Chi-Square Tests p < 0.05)
Positive Performance Expectancy 48%
Positive Safety Climate 55%
Positive Innovativeness 52%
Past use of wearable for work 31%
Age 25–34yr relative to 55–64yr 29%, 14%
respectively
Confident work will not be terminated in next 12 76%
months
Union Member 17%
Past use of wearable for personal use 59%
Confident can meet employer's expectations 85%
Race is Black, African American 10%
Confident will not be injured at work in next 12 76%
months
Male relative to Female 57%
Frontline worker relative to direct supervisor 56%, 21%
respectively
Annual Income $35,000-$54,999 relative to 22%, 19%
> $100,000 respectively
Concern: information used against me 65%
Race is Asian 6%
Concern: employer access to private info 66%
Concern: data not secure, non-permissible access 69%
3.3. Factors that predict employee willingness to use wearable technology
The single-variable analysis identified 18 factors significantly re-
lated to willingness to use wearables for work (Table 2). These factors
spanned features of employee characteristics and beliefs (e.g., perfor-
mance expectancy, innovativeness, trust, prior experience with wear-
ables, age, sex, race, income, and position in the company), as well as
the organizational setting (safety climate, worker union membership,
job security). It is notable that binned average willingness did not sig-
nificantly relate to job functions of construction-utility-field, manu-
facturing-material handling, service-standing, or office-sedentary work
(Chi-square = 7.13, p = 0.068). Unbinned average willingness, how-
ever, demonstrated that employees in construction-utility-field work
were significantly more willing than employees in service-standing or
office-sedentary job functions (Kruskal-Wallis Statistic = 15.79,
p < 0.001). Mean (95% confidence interval) willingness was 3.91
(3.78–4.05) for construction-utility-field work, 3.86 (3.73–3.99) for
manufacturing-materials handling work, 3.63 (3.51–3.75) for service-
standing work, and 3.61 (3.48–3.73) for office-sedentary work.
The best-performing predictive models consistently included per-
formance expectancy and safety climate as the strongest predictors of
willingness to use wearables across the tested use cases (Table 3). The
average AUC performance on the hold-out test dataset ranged from
0.808 to 0.878 for the best-performing models, and from 0.788 to 0.864
for the most-efficient models. Average F1 scores ranged from 0.733 to
0.791 for the best-performing models, and from 0.673 to 0.791 for the
most-efficient models (Fig. 3).
3.4. Factors related to a positive prior experience with wearables for work
Of the 1273 participants, 31% had prior experience using a wear-
able at work. Of the participants with prior experience, they reported
many purposes of their use: 48% to monitor productivity, 44% for
health promotion, 42% for safety, 29% to provide job instruction, 6%
for other purposes, and 6% identified the purpose as unknown to them
(numbers do not add to 100% because some participants had experi-
ence with multiple purposes). Common types of wearable devices in-
cluded activity-tracking wristbands, smartwatches, smart glasses, and
air-quality or noise sensors. In addition, of the participants with prior
Applied Ergonomics 78 (2019) 148–156
experience, 84% reported a positive prior experience, 80% identified
that the use was voluntary, and 40% received an incentive. Among the
Odds participants who received an incentive, 57% indicated the incentive
Ratio affected their willingness to use the wearable. Common types of in-
centives included money, insurance discounts, gift cards, or time off.
The single-variable analysis identified 18 variables significantly
11.73
6.51
related to a positive prior experience with wearables at work (Table 4).
3.40 The best-performing model to determine positive versus neutral-or-
2.73 negative prior experience was defined by an adequate level of in-
2.65 formation on data use (coefficient = 1.23), being involved in choosing
the device (coefficient = 1.13), adequate information on who has data
2.40
access (coefficient = 0.31), and safety climate (coefficient = 0.01),
1.95 with an intercept of 0.15. The mean (95% confidence interval) AUC of
1.84 the best-performing models was 0.932 (0.905–0.959). The most-effi-
1.80
cient model was defined by being involved in choosing the device
1.58
1.48
(coefficient = 1.94) and an adequate level of information on data use
(coefficient = 1.81), with an intercept of −0.39. The mean (95% con-
1.46 fidence interval) AUC of the most-efficient models was 0.884
0.61 (0.811–0.956).
0.56
4. Discussion
0.53
0.52 The results support our hypotheses that employee acceptance of
0.48 wearable technology differs across use cases and is predicted by factors
0.48
related to the organizational setting as well as employee characteristics
and beliefs. Specifically, use cases that focus on safety from recordings
made only at work elicited the highest acceptance, whereas tracking
information in and out of work to improve productivity elicited the
lowest acceptance. If the use case is inflexible, the single-variable
analysis revealed many other factors related to employee acceptance of
wearable technology, but the best-performing multi-variable models
often included safety climate along with performance expectancy and
innovativeness. For some scenarios, best-performing models also in-
cluded features related to job security, union membership, trust in the
technology, and demographic characteristics. The most-efficient models
included performance expectancy and sometimes included safety cli-
mate, although the confidence intervals for safety climate in the most-
efficient models overlapped zero. These results are consistent with
previous findings (Williams et al., 2015) that performance expectancy
represents the primary factor to predict technology acceptance. The
results of the best-performing models, however, also highlight the im-
portance of a strong safety climate, even for use cases of wearable
technology that are not specifically focused on safety.
Our results suggest that a well-accepted use case may also be im-
portant to the economic costs of wearable technology. Our sample in-
dicated that organizations often implement an incentive for using
wearable technology, and that the incentive often affected their will-
ingness to use the wearable. Thus, incentives may facilitate adherence
to using a wearable device, and adherence is fundamental to realizing
any potential benefits of its use. The results also demonstrated that only
a minority of participants would participate in even their most-willing
use case for no incentive (12.4%), and this percentage significantly
decreased for the least-willing use case. In addition, the median desired
incentive for the participants' most-willing use case was half of that for
the least-willing use case (100 versus 200 USD). A recent survey study
(Schall et al., 2018) asked industrial health and safety professionals to
estimate the amount their organization would be willing to spend on a
wearable device, and the mean was 72 USD per employee. Together,
these findings suggest that incentives would improve acceptance of
workplace wearable technology, but an economic gap may exist be-
tween organizations’ willingness to invest in wearable technology and
the expectations of employees. Thus, given this economic misalign-
ment, it becomes even more important to implement a well-accepted
use case for workplace wearable technology, because well-accepted use
cases require less incentive.
Regarding our secondary objective, a positive prior experience with
152
 Table 3
Mean (95% confidence interval) coefficients from 5 random seeds of models to predict willingness across use cases.
Average of Average of Monitor Monitor Monitor Monitor Monitor Monitor Monitor Monitor Monitor Monitor Alert for Alert for Provide Provide
Scenarios: Scenarios: Safety at Safety at Productivity Productivity Safety In Safety In Productivity Productivity Fitness In Fitness In External External Information Information
J.V. Jacobs, et al.

Best Efficient Work: Work: at Work: Best at Work: and Out and Out In and Out of In and Out of and Out and Out Hazards Hazards on Work on Work
Best Efficient Efficient of Work: of Work: Work: Best Work: of Work: of Work: at Work: at Work: Process: Best Process:
Best Efficient Efficient Best Efficient Best Efficient Efficient

Intercept −5.21 −2.91 −3.25 −1.32 −2.81 −0.79 (0.79) −3.4 −1.99 −3.46 (1.08) −0.95 (1.03) −3.5 −1.95 −3.36 −1.62 −4.27 −1.77
(1.05) (1.14) (2.04) (1.19) (0.43) (0.59) (1.73) (1.21) (1.46) (1.31) (1.81) (2.23) (0.63)
Performance 1.1 (0.11) 0.7 (0.22) 0.55 0.29 0.74 (0.15) 0.24 (0.25) 0.76 0.41 0.74 (0.16) 0.26 (0.28) 0.86 0.5 0.66 0.35 0.87 (0.24) 0.51 (0.19)
Expectancy (0.27) (0.24) (0.18) (0.33) (0.3) (0.36) (0.22) (0.31)
Safety Climate 0.93 0.37 (0.47) 0.53 0.23 0.58 (0.21) 0 1.01 0.48 0.48 (0.33) 0.04 (0.07) 0.66 0.21 0.73 0.24 0.5 (0.29) 0.1 (0.23)
(0.32) (0.51) (0.41) (0.42) (0.56) (0.35) (0.39) (0.5) (0.49)
Innovativeness 0.25 0.04 (0.09) 0.27 0.1 0.09 (0.12) 0 0.22 0.09 0.16 (0.1) 0 0.11 0.03 0.28 0.07 0.21 (0.2) 0.03 (0.03)
(0.22) (0.26) (0.19) (0.17) (0.12) (0.19) (0.09) (0.26) (0.11)
Concern: Data −0.14 0 −0.28 −0.02 0 0 −0.2 −0.01 −0.01 (0.04) 0 −0.15 −0.02 −0.01 0 −0.22 −0.01 (0.03)
Not Secure (0.11) (0.23) (0.07) (0.16) (0.04) (0.23) (0.07) (0.02) (0.17)
Confident Will 0.08 0 0.08 0 0 0 0 0 −0.01 (0.02) 0 0 0 0.2 (0.24) 0.04 0.14 (0.21) 0.01 (0.03)
Meet (0.07) (0.14) (0.11)
Employer
Expectations
Race is Asian 0.26 (0.54) 0 0.34 0 0 0 0.23 0.07 0.01 (0.03) 0 0.16 0 −0.02 0.01 −0.1 (0.59) 0
(0.46) (0.22) (0.18) (0.43) (0.71) (0.02)
Concern: −0.18 −0.01 −0.18 −0.03 −0.03 (0.05) 0 −0.05 0 −0.1 (0.1) 0 −0.07 −0.02 −0.18 −0.01 −0.05 (0.12) 0
Employer (0.27) (0.04) (0.21) (0.07) (0.06) (0.19) (0.06) (0.27) (0.02)
Access to
Private
Information

153
Experience With 0.31 (0.4) 0.02 (0.07) 0.25 0.07 0.19 (0.31) 0 0.04 0 0.13 (0.15) 0 0.07 0 0.05 0 0.71 (0.27) 0.14 (0.19)
Wearable for (0.41) (0.2) (0.08) (0.11) (0.13)
Work
Female −0.1 0 −0.14 0 −0.04 (0.11) 0 −0.16 −0.01 −0.01 (0.03) 0 −0.09 −0.01 −0.16 −0.01 −0.4 (0.2) −0.04 (0.1)
(0.24) (0.29) (0.19) (0.02) (0.26) (0.01) (0.31) (0.04)
Race is Black, −0.1 0 −0.15 0 0 0 0.06 −0.01 0.09 (0.24) 0 0.08 0 0.09 0 0.13 (0.44) 0
African- (0.13) (0.76) (0.36) (0.04) (0.22) (0.18)
American
Experience With −0.09 0 0.14 0.04 0 0 −0.24 −0.03 −0.01 (0.02) 0 −0.03 0 0.13 0 0.11 (0.19) 0
Wearable for (0.27) (0.38) (0.12) (0.33) (0.1) (0.07) (0.17)
Personal Use
Confident Will 0.06 (0.16) 0.01 (0.03) 0.11 0 0 0 0.02 0 0.08 (0.09) 0 0.01 0 0.04 (0.1) 0 0.11 (0.23) 0
Not Be (0.11) (0.04) (0.02)
Terminated
Confident Will 0.05 (0.09) 0 0.03 0 0 0 0 (0.1) 0 0 (0.05) 0 0.02 0 −0.03 0 0.09 (0.32) 0
Not Be (0.06) (0.07) (0.08)
Injured
Position in the −0.06 0 −0.19 −0.01 −0.04 (0.11) 0 −0.11 −0.04 0 0 0 0 −0.12 0 −0.06 (0.11) 0
Company (0.16) (0.22) (0.04) (0.2) (0.1) (0.17)
Concern: 0.04 (0.12) 0 0.34 0 0 0 −0.03 0 0 0 0.07 0 0 0 0.06 (0.15) 0
Information (0.31) (0.08) (0.19)
Will Be Used
Against Me
Union Member 0.03 (0.05) 0 −0.09 −0.04 0 0 0.11 0.05 0.18 (0.15) 0 0.02 0 −0.18 0 −0.02 (0.04) 0
(0.66) (0.1) (0.52) (0.13) (0.07) (0.39)
Age Bracket −0.01 0 −0.02 0 0 0 −0.01 0 (0.01) 0 0 0 0 0 0 0.06 (0.13) 0
(0.03) (0.09) (0.03)
(continued on next page)
Applied Ergonomics 78 (2019) 148–156
J.V. Jacobs, et al. Applied Ergonomics 78 (2019) 148–156

Information
on Work

Efficient
Process:
Provide

0
Process: Best
Information

0.01 (0.09)
on Work
Provide

at Work:
Alert for

Efficient
External
Hazards

0
at Work:
Alert for
External

0 (0.01)
Hazards

Best
Fitness In

of Work:
Efficient
and Out
Monitor

(0.09)
0.03
Fitness In

of Work:
and Out
Monitor

(0.16)

Fig. 3. Model Performance. The area under the receiver operating character-
0.06
Best

istic curve (AUC) and F1 scores for classifying a positive willingness to use
wearables for each use case. Squares with error bars represent the mean and
In and Out of
Productivity

95% confidence interval of the AUC during test performance on a hold-out

sample; circles represent the same features of the F1 scores. Black symbols
Efficient
Monitor

Work:

represent performance of the best-performing models; gray symbols, the most-

efficient models.
0
In and Out of
Productivity

0.02 (0.07)
Work: Best

wearable technology was associated with the employees reported in-

Monitor

volvement in selecting the device and whether they were well informed
about data use. In the best-performing model, a strong safety climate
was also a valuable factor. The single-variable analysis once again in-
Bold text indicates non-zero coefficients, as determined by a 95% confidence interval that does not overlap zero.
Safety In

of Work:
Efficient
and Out
Monitor

dicated a long list of correlated factors that span features of the orga-
(0.03)
0.01

nizational setting and the employees’ characteristics and beliefs. Thus,

involving and informing employees during the selection and im-
Safety In

of Work:

plementation of a wearables program would likely facilitate a positive

and Out
Monitor

(0.09)
0.18

employee experience.
Best

Some study limitations must be considered. Of primary concern, our

study utilized an online survey, which associates with known concerns
Productivity

regarding response rate and sample bias (Evans and Mathur, 2005). We,
at Work:
Efficient
Monitor

however, employed several recommended methods to mitigate these

concerns and enhance the strengths afforded by online surveys (Evans
0

and Mathur, 2005). First, to standardize contextual understanding

at Work: Best

−0.01 (0.03)
Productivity

among participants, we (a) provided definitions and examples, (b)

Monitor

employed a test question with clarifying rationale, and (c) constructed a

concrete scenario regarding the form factor and data capabilities of the
wearable devices. Second, we utilized sampling quotas and a national
Safety at

Efficient

sample to minimize bias. Third, we used pre-populated participant

Monitor

Work:

networks of a survey service to minimize misattribution of the parti-

0

cipants’ characteristics, technological variations, and effects of online

experience. Fourth, small incentives from an opt-in network familiar to
Safety at
Monitor

(0.05)
Work:

the participants were used to improve response rate. Lastly, the online
0.04
Best

medium enabled control over question and response ordering, as well

as visual formatting. Nevertheless, signs of bias that likely relate to the
Average of
Scenarios:
Efficient

method of recruitment remained evident, such that the sample had a

smaller proportion of older adults, participants who did not complete
0

high school, participants of Hispanic or Latino ethnicity, and females in

Average of

comparison to the employed population of the United States. Further,

Scenarios:

0 (0.02)

online-service recruitment with a voluntary consent process signifies

Best

that the recruited individuals had access to technology resources, in-

Table 3 (continued)

terest in the research topic, and trust to use their data. Indeed, our
average score on participant innovativeness was similar to that of an-
Annual Income

other online survey study on wearable technology (Hong et al., 2017),

buthigher than that from another study that utilized in-person re-
cruitment at shopping centers (Rauschnabel and Ro, 2016). These signs
of bias in our study may indicate an over-estimate of acceptance.

154
J.V. Jacobs, et al. Applied Ergonomics 78 (2019) 148–156

Table 4
Single-variable analysis of factors related to a positive past experience with wearables.
Factor Significantly Related to Positive Past Experience (Fisher's Exact or Chi-Square Tests p < 0.05) % of Sample with Past Experience with Attribute Odds Ratio

Adequate Information On Who has Data Access 78% 12.9

Involved In Choosing Device 66% 12.7
Adequate Information On Data Use 79% 10.5
Safety Climate 70% 9.3
Innovativeness 73% 5.2
Adequate Information On What Is Measured 85% 5
Supervisor Responsive To Feedback 76% 5
Direct Supervisor Relative To Frontline Worker 28%, 47% respectively 4.7
Confident Wearable Can Improve Fitness 79% 4.1
Confident Will Not Be Injured At Work 78% 3.9
Attained Bachelor's Degree Or Higher 54% 3.5
Confident Wearable Can Improve Safety 75% 3.4
Confident Wearable Can Improve Productivity 71% 3.2
Confident Will Meet Employer Expectations 84% 2.6
Purpose Is For Health Promotion 44% 2.6
Confident Will Not Be Terminated From Work 76% 2.3
Personal-Use Experience With Wearable 85% 2.3
Use Was Voluntary 80% 2.2

5. Conclusions representing the authors' employer at the time of the study.

Our results indicate that wearable technology is already actively Funding

used in the workplace. To realize the potential benefits of wearable
technology, however, employees will need to adopt and adhere to its
use, and use behavior is strongly dependent on employee acceptance.
Past literature has focused on employee characteristics related to the
acceptance of wearable technology, and this study extends the litera-
ture with additional focus on the use cases and work environments that
predict acceptance. Amidst other indicators, our results suggest that
organizations interested in implementing a work-sponsored program of
wearable technology should (a) prioritize use cases of workplace safety
in which data are recorded only at work, (b) advance a positive safety
climate, (c) ensure sufficient evidence to support employees’ beliefs
that the wearable will meet its objective, and (d) involve and inform
employees in the process of selecting and implementing the wearable
technology.
The optimal scenario appears to be the voluntary and incentivized
implementation of a wearable for construction, utility, or field work in
which data are collected only at work to monitor and improve safety.
The environment would have a strong safety climate, be unionized, and
have a low risk for employee termination. The employee population
would be predominantly techno-curious, experienced with wearable
technology, and the direct supervisors can serve as relevant participants
in data collection. The implementation should facilitate performance
expectancy by communicating evidence that the wearable will meet its
objective (i.e., the device would be proven effective), the employees
should be involved in the process of selecting the device, and the pro-
gram should engender trust by clearly informing employees about why,
how, and by whom the data will be used and protected.
Acknowledgments
We thank our co-workers for piloting and commenting on an early
draft of the survey in order to improve wording, formatting, and con-
tent prior to implementation.
Appendix A. Supplementary data
Supplementary data related to this article can be found at https://
doi.org/10.1016/j.apergo.2019.03.003.
Declaration of Interest
The work was privately funded by Liberty Mutual Insurance,
This study was funded by Liberty Mutual Insurance.
References
Abbasi, M.S., Tarhini, A., Hassouna, M., Shah, F., 2015. Social, organizational, demo-
graphy and individuals'technology acceptance behaviour: a conceptual model. Eur.
Sci. J. 11.
Benbasat, I., Barki, H., 2007. Quo Vadis, TAM? J. Assoc. Inf. Syst. 8, 211–218.
Choi, B., Hwang, S., Lee, S., 2017. What drives construction workers' acceptance of
wearable technologies in the workplace?: indoor localization and wearable health
devices for occupational safety and health. Autom. ConStruct. 84, 31–41.
Evans, J.R., Mathur, A., 2005. The value of online surveys. Internet Res. 15, 195–219.
Fensli, R., Pedersen, P., Gundersen, T., Hejlesen, O., 2008. Sensor acceptance mod-
el–measuring patient acceptance of wearable sensors. Methods Inf. Med. 47, 89–95.
Gao, Y., Li, H., Luo, Y., 2015. An empirical study of wearable technology acceptance in
healthcare. Ind. Manag. Data Syst. 115, 1704–1723.
Gribel, L., Regier, S., Stengel, I., 2016. Acceptance factors of wearable computing: an
empirical investigation. In: INC, pp. 67–72.
Hong, J.-C., Lin, P.-H., Hsieh, P.-C., 2017. The effect of consumer innovativeness on
perceived value and continuance intention to use smartwatch. Comput. Hum. Behav.
67, 264–272.
Huang, F.-F., Lai, Y.-H., 2016. The Acceptance of Smart Wearable Devices through Health
Cognitive, MATEC Web of Conferences. EDP Sciences, pp. 05005.
Huang, Y.H., Lee, J., Chen, Z., Perry, M., Cheung, J.H., Wang, M., 2017. An item-response
theory approach to safety climate measurement: the Liberty Mutual Safety Climate
Short Scales. Accid. Anal. Prev. 103, 96–104.
Hwang, C., 2014. Consumers' Acceptance of Wearable Technology: Examining Solar-
Powered Clothing. Iowa State University.
Kalantari, M., 2017. Consumers' adoption of wearable technologies: literature review,
synthesis, and future research agenda. Int. J. Technol. Mark. 12, 274–307.
Kim, K.J., Shin, D.-H., 2015. An acceptance model for smart watches: implications for the
adoption of future wearable technology. Internet Res. 25, 527–541.
Kwee-Meier, S.T., Bützler, J.E., Schlick, C., 2016. Development and validation of a
technology acceptance model for safety-enhancing, wearable locating systems.
Behav. Inf. Technol. 35, 394–409.
Lee, H.-M., 2009. A study on the acceptance of wearable computers based on the extended
technology acceptance model. Res. J. Costume Cult. 17, 1155–1172.
Lin, B.S., Wong, A.M., Tseng, K.C., 2016. Community-based ECG monitoring system for
patients with cardiovascular diseases. J. Med. Syst. 40, 80.
Lunceford, B., 2009. Reconsidering technology adoption and resistance: observations of a
semi-luddite. Explor. Media Ecol. 8, 29–47.
Lunney, A., Cunningham, N.R., Eastin, M.S., 2016. Wearable fitness technology: a
structural investigation into acceptance and perceived fitness outcomes. Comput.
Hum. Behav. 65, 114–120.
Mele, C., Pels, J., Polese, F., 2010. A brief review of systems theories and their managerial
applications. Serv. Sci. 2, 126–135.
Pfeiffer, J., von Entress-Fuersteneck, M., Urbach, N., Buchwald, A., 2016. Quantify-Me:
Consumer Acceptance of Wearable Self-Tracking Devices.
Rauschnabel, P.A., Ro, Y.K., 2016. Augmented reality smart glasses: an investigation of
technology acceptance drivers. Int. J. Technol. Mark. 11, 123–148.
Robertson, M.M., Hettinger, L.J., Waterson, P.E., Ian Noy, Y., Dainoff, M.J., Leveson,
N.G., Carayon, P., Courtney, T.K., 2015. Sociotechnical approaches to workplace

155
J.V. Jacobs, et al.
safety: research needs and opportunities. Ergonomics 58, 650–658.
Schall, M.C., Sesek, R.F., Cavuoto, L.A., 2018. Barriers to the adoption of wearable sensors
in the workplace: a survey of occupational safety and health professionals. Hum.
Factors 60, 351–362.
Spagnolli, A., Guardigli, E., Orso, V., Varotto, A., Gamberini, L., 2014. Measuring User
Acceptance of Wearable Symbiotic Devices: Validation Study across Application
Scenarios, International Workshop on Symbiotic Interaction. Springer, pp. 87–98.
Turhan, G., 2013. An assessment towards the acceptance of wearable technology to
consumers in Turkey: the application to smart bra and t-shirt products. J. Text. Inst.
156
Applied Ergonomics 78 (2019) 148–156
104, 375–395.
Venkatesh, V., Morris, M.G., Davis, G.B., Davis, F.D., 2003. User acceptance of informa-
tion technology: toward a unified view. MIS Q. 425–478.
Weng, M., 2016. The Acceptance of Wearable Devices for Personal Healthcare in China,
Information Processing Science. University of Oulu.
Williams, M.D., Rana, N.P., Dwivedi, Y.K., 2015. The unified theory of acceptance and use
of technology (UTAUT): a literature review. J. Enterp. Inf. Manag. 28, 443–488.
Yang, H., Yu, J., Zo, H., Choi, M., 2016. User acceptance of wearable devices: an extended
perspective of perceived value. Telematics Inf. 33, 256–269.