Professional Documents
Culture Documents
Applied Ergonomics
journal homepage: www.elsevier.com/locate/apergo
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 study samples a broad spectrum of employees from across the United
States. The sampling provides novel and comparably generalizable in-
Wearable technology products are advanced sensor and computing sights for how to enhance the likelihood of a well-accepted wearable
technologies that a person can wear on their body during daily activity technology program in the workplace.
to generate, store, and transmit data. Use cases for wearable technology Survey instruments are often employed to test hypothesis-driven
in the workplace are being developed to, for example: (a) monitor structural models for constructs that associate with attitudes about
productivity, (b) identify and intervene on safety hazards or risks, (c) technology, people's intent to use the technology, and actual use be-
provide augmented instruction to improve task management, and (d) havior. Perhaps the most developed of these models for testing the
facilitate health and wellness (Kalantari, 2017). A key concern with acceptance of workplace technology is the Unified Theory of
adopting wearable technology in the workplace is employee acceptance Acceptance and Use of Technology (UTAUT) model (Venkatesh et al.,
and adherence to its use (Schall et al., 2018). Therefore, the primary 2003; Williams et al., 2015). The model suggests that use behavior is
objective of this study is to identify factors that predict employee ac- predicted by facilitating conditions and a person's behavioral intention
ceptance of work-related wearable technology. Employee acceptance is to use technology. Behavioral intention is then predicted by constructs
represented by the participants’ reported willingness to use work-re- of performance expectancy (belief that the system will meet its objec-
lated wearable technology in the use cases defined in the survey. The tive or help attain gains in their job), effort expectancy (ease of use),
secondary objective of this study is to identify factors related to a po- and social influence (perception that others of importance believe the
sitive experience with work-related wearable technology among parti- individual should use the system). In parallel, gender, age, experience
cipants who already had such experience. This study uniquely con- with the technology, and voluntariness of use all mediate the re-
tributes to the literature through its combined focus on factors related lationships of these constructs with intended and actual use. The model
to the use case of the wearable technology and the workplace en- has been validated across multiple studies evaluating varied types of
vironment in which it is used, in addition to employee characteristics technology, and the strongest predictor of use is behavioral intention
and beliefs. Further, rather than focusing on a specific industry, this 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. Applied Ergonomics 78 (2019) 148–156
people will be more likely to use a fitness band if they believe it will 18 years of age and at least part-time employed. Because online surveys
help them improve their fitness. Other acceptance models have also risk bias, we requested that the survey service set recruitment quotas to
included concepts of trust (e.g., in the safety, security, and use of the equally represent the job functions inquired about in our survey. We
device and its data) and innovativeness (e.g., an individual's propensity also requested recruitment quotas on demographics of gender, age,
to use and accept any novel technology) (Gribel et al.,2016; Kwee- race, and ethnicity to represent the demographics of the United States
Meier et al., 2016; Pfeiffer et al., 2016; Weng, 2016). Thus, the existing employed population. Participant profiles are updated annually within
literature has identified several factors to predict the acceptance of the survey service's participant networks, and the answers to our survey
wearable technology. verified the profiles' accuracy. Participants were recruited from net-
It is not the intent of this study to formally test or extend these works in the United States and received points as a means of re-
models, because many other studies have already applied them to imbursement. Points could be redeemed for money or for gift cards per
wearable technology (Abbasi et al., 2015; Choi et al., 2017; Fensli et al., the pre-negotiated agreement between the survey service and their
2008; Gao et al., 2015; Hong et al., 2017; Huang and Lai, 2016; Hwang, participant networks. The service screened for “straight-line” re-
2014; Kim and Shin, 2015; Kwee-Meier et al., 2016; Lee, 2009; Lin sponding, timed out attempts, and other signals of invalid responses,
et al., 2016; Lunney et al., 2016; Pfeiffer et al., 2016; Rauschnabel and and then backfilled the sample to meet quotas. Participants could re-
Ro, 2016; Spagnolli et al., 2014; Turhan, 2013; Yang et al., 2016). In- spond to the survey only after an informed consent process, which was
deed, these models have been criticized for the heuristic nature of their approved by the Institutional Review Board.
structure as well as for their historically narrow scope to focus on
employee beliefs and characteristics (Benbasat and Barki, 2007;
2.2. The survey
Lunceford, 2009). These criticisms thus suggest a need for an alter-
native modeling approach that is designed to analyze considerations
The survey (provided in the Appendix) was developed by the project
beyond employee characteristics and beliefs. Other needed constructs
team. The survey content was guided by systems and sociotechnical
include task-related factors (e.g., the nature and purpose of the wear-
systems theories (Mele et al., 2010; Robertson et al., 2015) to include
able data collection and the employees' job function) and factors related
items already identified in the literature as valuable predictors of
to the organizational setting (e.g., safety climate, job security, and
technology acceptance as well as other hypothesized items of interest
worker's union membership).
derived from conversations with industrial subject-matter experts. The
When evaluating studies on wearable-technology acceptance in the
survey included Likert scales, ranking, and open-response formats to
workplace, previous results have supported a multifactorial systems
ascertain the participants' characteristics, their employers' organiza-
model of acceptance. Specifically, in addition to employee beliefs, ac-
tional characteristics, past experience with wearable technology, and
ceptance may differ based on cost, type of device, relationship of the
their opinions about wearable technology in the workplace. Because the
user to those who evaluate the data, as well as the employee's job type,
participants were anticipated to have differing levels of experience with
position in the company, and experience with wearable technology
wearable technology, their understanding of the technology was fa-
(Choi et al., 2017; Kwee-Meier et al., 2016; Schall et al., 2018; Spagnolli
cilitated by presenting (a) a definition of the technology; (b) a test
et al., 2014). These previous studies highlight the importance of a
question to identify whether certain examples represented wearable
systems perspective when determining the acceptance of wearable
technology, which was followed by a clarifying explanation about the
technology; however, they are based on very specific sample char-
correct answer; and (c) a concrete scenario for the participants' con-
acteristics and use cases. Thus, research is needed to further examine
sideration when responding to questions about willingness (Appendix).
more generalizable use cases and environmental features on a broad-
Responses to questions regarding confidence that wearables would
spectrum sample of employees. To address this need, this study seeks to
improve the participants’ safety, productivity, or health were con-
explore how factors related to the organizational setting, the individual
sidered to reflect their performance expectancy. Innovativeness was
employee, and the purpose or use case of the wearable technology as-
determined by the level of agreement that they are the first to try new
sociate with attitudes toward its use in the workplace. We hypothesized
technology among their peers. Trust was represented by items re-
that acceptance would (a) significantly differ by use case, and (b) be
garding levels of concern over data security and misuse. Safety climate
best predicted from a combination of environmental factors with factors
was represented by the short scale defined by Huang and colleagues.
of employee characteristics and beliefs.
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
2. Methods important safety outcomes (Huang et al., 2017).
To address our primary objective, we asked participants to state
2.1. Participants their level of agreement on a 5-point ordinal scale (strongly disagree to
strongly agree) about whether they would voluntarily use a wearable
Participants were recruited by email through an independent online under seven use cases that varied the function, location, and purpose of
survey system (SurveyGizmo; www.surveygizmo.com). Individuals had the data collection (Table 1). To address our secondary objective,
opted into the survey system's participant network and consented to participants were asked to state on a 5-point ordinal scale (very bad to
receive such emails. Data collection occurred from February 24th to very good) the quality of their prior experience using wearables for
March 29th, 2017. Inclusion criteria required participants to be at least work-related or work-sponsored purposes.
Table 1
Use-case scenarios to assess willingness to use wearable technology at work.
Use Case Function Location Purpose
1 Record activity and physical status Only at work Monitor and improve safety
2 Record activity and physical status In and outside of work Monitor and improve safety
3 Record activity and physical status Only at work Monitor and improve productivity
4 Record activity and physical status In and outside of work Monitor and improve productivity
5 Record activity and physical status In and outside of work Monitor and improve health and fitness
6 Monitor environmental hazards Only at work Alert for hazard proximity or level
7 Provide information on work process Only at work Instruct how and when to perform tasks
149
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 positive willingness or a positive prior experience for participants with
versus without a predictor characteristic. Fisher's Exact tests or Chi-
To ensure our target variables of willingness and quality of prior Square tests were used to determine the statistical significance of the
experience had sufficient sample for each level of response, we binned relationship.
the responses from a 5-point ordinal scale to binary scales of positive Once significant individual predictors were identified, the unbinned
versus neutral-or-negative responses (i.e., scores greater than 3 versus responses to the significant predictor variables were entered into
scores of 3 or less). McNemar's tests of proportions were used to de- elastic-net generalized linear models (GLMs) using R software packages
termine whether acceptance differed across use cases. To facilitate (e.g., glmnet). Both the binned and unbinned versions were tested, but
running only one single-variable analysis, rather than on each use case, performance improved with the unbinned predictors, so we report on
responses to the questions on willingness were averaged across sce- those models. Modeling included the following steps: (a) random down
narios prior to binning. Ordinal scales of the predictor variables were sampling of over-represented target-variable responses to balance the
also binned for single-variable analysis. We report the odds ratios of a data set and prevent model overfitting; (b) stratified random sampling
150
J.V. Jacobs, et al. Applied Ergonomics 78 (2019) 148–156
3. Results
151
J.V. Jacobs, et al. Applied Ergonomics 78 (2019) 148–156
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
Work:
0.02 (0.07)
Work: Best
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
of Work:
(0.09)
0.18
employee experience.
Best
regarding response rate and sample bias (Evans and Mathur, 2005). We,
at Work:
Efficient
Monitor
−0.01 (0.03)
Productivity
Efficient
Work:
(0.05)
Work:
the participants were used to improve response rate. Lastly, the online
0.04
Best
0 (0.02)
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
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
155
J.V. Jacobs, et al. Applied Ergonomics 78 (2019) 148–156
safety: research needs and opportunities. Ergonomics 58, 650–658. 104, 375–395.
Schall, M.C., Sesek, R.F., Cavuoto, L.A., 2018. Barriers to the adoption of wearable sensors Venkatesh, V., Morris, M.G., Davis, G.B., Davis, F.D., 2003. User acceptance of informa-
in the workplace: a survey of occupational safety and health professionals. Hum. tion technology: toward a unified view. MIS Q. 425–478.
Factors 60, 351–362. Weng, M., 2016. The Acceptance of Wearable Devices for Personal Healthcare in China,
Spagnolli, A., Guardigli, E., Orso, V., Varotto, A., Gamberini, L., 2014. Measuring User Information Processing Science. University of Oulu.
Acceptance of Wearable Symbiotic Devices: Validation Study across Application Williams, M.D., Rana, N.P., Dwivedi, Y.K., 2015. The unified theory of acceptance and use
Scenarios, International Workshop on Symbiotic Interaction. Springer, pp. 87–98. of technology (UTAUT): a literature review. J. Enterp. Inf. Manag. 28, 443–488.
Turhan, G., 2013. An assessment towards the acceptance of wearable technology to Yang, H., Yu, J., Zo, H., Choi, M., 2016. User acceptance of wearable devices: an extended
consumers in Turkey: the application to smart bra and t-shirt products. J. Text. Inst. perspective of perceived value. Telematics Inf. 33, 256–269.
156