Technological Forecasting & Social Change 134 (2018) 246–253

Contents lists available at ScienceDirect

Technological Forecasting & Social Change

journal homepage: www.elsevier.com/locate/techfore

Who will be smart home users? An analysis of adoption and diffusion of T

smart homes☆
⁎
Jungwoo Shina, Yuri Parkb, , Daeho Leec
a
Department of Industrial and Management Systems Engineering, KyungHee University, 1732 Deogyeong-daero, Giheung-gu, Yongin, Gyeonggi 446-701, South Korea
b
Department of ICT Industry Research, Korea Information Society Development Institute, 18 Jeongtong-ro, Deoksan-myeon, Jincheon-gun, Chungchengbuk-do 27872,
South Korea
c
Department of Interaction Science, Sungkyunkwan University, 25-2 Sungkyunwan-ro, Jongno-gu, Seoul, South Korea

A R T I C LE I N FO A B S T R A C T

Keywords: A smart home is considered a primary service of the Internet of Things (IoT), and global leading companies are
Consumer behavior launching smart home services/products based on the IoT. However, the spread of smart homes has been slower
Multivariate probit model than expected, and analysis of smart homes from a demand perspective is required. This study suggests im-
Network externality plications for promoting the smart home market by analyzing factors affecting adoption and diffusion of smart
Smart home
homes. A technology acceptance model was used to describe the adoption of smart homes and a multivariate
Technology acceptance model
probit model was used to describe the diffusion of smart homes. The characteristics of smart homes such as
network effects between services/products and the importance of personal information protection were con-
sidered in addition to demographic variables. The results of this study show that compatibility, perceived ease of
use, and perceived usefulness have significant positive effects on purchase intention. In terms of purchase timing,
unlike other information and communication technology (ICT) services/products, older consumers are more
likely to purchase smart homes within a given time period than are younger consumers. Therefore, a strategy for
promoting smart home purchases by young consumers is required to increase market demand.

1. Introduction allows consumers to control their home appliances through Alexa, an

Interest in various Internet of Things (IoT) services is increasing as
most objects can now be connected to a network through the IoT.
Among these services, smart home services are regarded as killer ap-
plications in terms of marketability and consumer accessibility. A smart
home is defined as “an intelligent environment that is able to acquire
and apply knowledge about its inhabitants and their surroundings in
order to adapt and meet the goals of comfort and efficiency” (Perumal
et al., 2013). Smart home services include the control and automation
of lighting, heating (e.g., smart thermostats), ventilation, air con-
ditioning (HVAC), and security, as well as home appliances such as
washers/dryers, ovens, and refrigerators/freezers.
Smart homes are emerging as a new competitive market for ICT
companies as they seek to find new revenue sources due to saturation of
the smart phone market. Major players leading the ICT ecosystem are
rushing to release smart home services/products. For example, Amazon
artificial intelligence-based voice recognition technology. Google pro-
vides smart home services through ‘Google assistant,’ which includes
personal assistant services based on voice recognition and artificial
intelligence technology. For their smart home services, Amazon and
Google offer smart speakers called ‘Amazon Echo’ and ‘Google Home,’
respectively. Telecommunications companies provide similar smart
home devices such as speakers or set-top boxes.
Despite the competitive launch of smart home services/products,
the global smart home1 penetration rate of households is 7.5%, with
revenue expected to reach $46,252 m in 2018 (Statista, 2018).
Greenough (2016) pointed out that the smart home market is currently
in a chasm between early adopters and the mass market. This chasm, in
spite of suppliers actively providing smart home services, means that it
is necessary to analyze smart homes from a consumer perspective. First,
the adoption rate of smart home services will vary depending on the
consumer utility for automatically controlling connected items such as

☆
This work was supported by the Ministry of Education of the Republic of Korea, the National Research Foundation of Korea (NRF-2017R1C1B5017518) and the Korea Ministry of
Environment as Climate Change Correspondence R&D Program (2014001300001).
⁎
Corresponding author.
E-mail addresses: yrpark@kisdi.re.kr (Y. Park), daeho.lee@skku.edu (D. Lee).
1
According to Statista (2018), smart home services/products include digitally connected and controlled devices, sensors, actuators, and cloud services that support automation, control
hubs to connect sensors and actuators, B2C hardware, and software sales. However, smart TVs, smart gardening devices, and B2B/C2C sales such as hotels are excluded.

https://doi.org/10.1016/j.techfore.2018.06.029
Received 12 January 2018; Received in revised form 18 May 2018; Accepted 13 June 2018
Available online 22 June 2018
0040-1625/ © 2018 Published by Elsevier Inc.
J. Shin et al.
home appliances. Therefore, it is important to analyze consumers' in-
tentions to adopt smart home services and the factors influencing the
adoption rate (diffusion) of smart homes in order to determine how to
increase smart home demand.
In addition, smart homes have attributes that are different from
other ICT services or products, such that specific analysis of smart
homes is required. A smart home has individual products or services
interconnected to one other. Therefore, compatibility is an important
factor when choosing smart home services compared to other IoT ser-
vices since smart home services require connections and communica-
tions among various home appliances. It is also not easy to purchase a
smart home all at once because different services/products have dif-
ferent replacement cycles, many of which are long. The choice of one
smart home service might depend on other smart home services as
smart home services require communication within a smart home.
Consequently, analysis of smart home services from a system perspec-
tive consisting of interworking (or combining) individual smart devices
is appropriate rather than considering the selection of each device in-
dividually.
Few previous studies have analyzed smart homes from a user per-
spective, while a lot of research has occurred on smart home technol-
ogies. Some studies that have analyzed smart home technologies based
on IoT are as follows. Soliman et al. (2013) suggested an approach to
develop smart home applications based on IoT and Cloud. Li and Yu
(2011) presented smart home system design based on IoT. Chong et al.
(2011) analyzed the characteristics and disadvantages of smart home
systems. Stojkoska and Trivodaliev (2017) presented a framework for
narrowing the gap between current and future smart homes based on
IoT. Feng et al. (2017) suggested how to apply IoT and a cognitive
dynamic system to smart homes. Hui et al. (2017) provided seven major
requirements for establishing a smart home. Mao et al. (2017) analyzed
a design scheme of an intelligent home system.
There have been attempts to analyze smart homes from the user's
perspective. Yang et al. (2016) analyzed user adoption of smart home
services using a partial least squares method. Park et al. (2018) ana-
lyzed adoption of smart homes using a technology acceptance model.
Kim et al. (2016) estimated willingness to pay for smart home services
using a contingent valuation method. Richter and Pollitt (2016) ana-
lyzed the choice of smart electricity service using a discrete choice
model. While these studies empirically analyzed smart home services
from the user's point of view, none of them simultaneously analyzed
adoption and diffusion. In addition, previous studies analyzed in-
dividual smart home services/products without considering a system
perspective.
This study suggests a strategic direction to overcome the chasm
following early adopters by analyzing how acceptance depends on the
characteristics of smart home services and factors affecting the diffusion
of smart home services. The remainder of this paper is organized as
follows. Section 2 describes the research models, and section 3 de-
scribes the survey design and empirical data for adoption and diffusion
of smart homes. Section 4 presents an analysis of the main factors in-
fluencing adoption and diffusion of smart home devices as part of a
smart home system. Lastly, section 5 presents a summary of our results
and provides a market revitalization strategy for smart home systems.
2. Methodology
2.1. Data
Micro-level consumer data is required to analyze factors affecting
consumer adoption and diffusion of smart home services. However,
revealed preference data for smart home service markets, especially
market data on individual consumer choices, are difficult to obtain
since the smart home market is in its early stages. Therefore, instead of
using revealed preference data, this study used stated preference data
from surveys that included usage behavior for smart home services.
247
Technological Forecasting & Social Change 134 (2018) 246–253
Two different surveys were used in this study. One attempted to
clarify the intention to adopt IoT services, while the other investigated
usage behavior of IoT services. The two surveys were used together as
each individually was not designed to analyze both adoption and dif-
fusion factors simultaneously. Characteristics of smart homes were re-
quired to analyze adoption according to the attributes of smart homes,
while purchase timing data was needed to analyze diffusion. The first
survey was designed to analyze consumer adoption and did not include
purchase timing data. On the other hand, the second survey included
data on the intention to use, but the characteristics of services were not
investigated. Therefore, the first survey was used to analyze adoption,
while the second survey was used for diffusion despite the lack of a
uniform survey population.
The respondents of the first survey used for adoption analysis were
310 South Korean smartphone users who could be potential smart home
service users. Data were collected through a mobile survey. All 310
individuals were smartphone owners, 152 of whom (49%) were male,
and 158 (51%) were female. By age, 23% were in their teens, 19% in
their 20s, 22% in their 30s, 19% in their 40s, and 18% in their 50s or
older. The respondents of the second survey used for the diffusion
analysis were 2113 smartphone users, aged 20 to 65. The survey was
conducted online. Of the respondents, 1059 (50%) were male, and 1054
(50%) were female. By age, 22% were in their 20s, 30% were in their
30s, 26% were in their 40s, 16% were in their 50s, and 5% were in their
60s. The samples were not significantly different from the age and
gender distributions of South Korea. In the second survey, smart home
services were divided into three categories based on home appliance
size. Respondents were asked about status of use, intention to use, and
when to purchase (if the respondents were willing to become users) of
large, medium, and small home appliances.
< 4% of respondents currently use a smart home service, but 46.4%
of respondents showed high intention to use in the future. This study
suggests approaches for activating smart home services by analyzing
influential factors.
2.2. Model specifications
2.2.1. Extended technology acceptance model for analyzing adoption of
smart home services
The technology acceptance model (TAM), originally developed by
Davis (1989), is one of the most popular methodologies for analyzing
consumer acceptance intentions. TAM analyzes consumer acceptance
intentions through perceived usefulness and perceived ease of use.
According to Davis (1989), perceived usefulness is defined as a user's
ability to utilize the system to improve his/her performance, while
perceived ease of use is the degree to which the user can use the system
without effort. TAM has been sufficiently explanatory for analyzing
user acceptance in previous studies. However, Legris et al. (2003)
pointed out that it is necessary to extend TAM according to the char-
acteristics of the technology analyzed because consumer objectives for
adopting ICT are different for each technology. In this study, therefore,
TAM was extended by additionally considering compatibility and
privacy protection factors beyond the perceived usability and perceived
ease of use of the basic TAM. The extended TAM was used to analyze
consumer acceptance of smart home services as shown in Fig. 1.
As pointed out earlier, compatibility is a critical factor in adopting
smart home services since it is important to assess how smart home
services are interoperable with various home appliances and external
services. Compatibility was considered an important factor for adoption
of ICT products in Chen et al. (2009), Corrocher (2011), and Wu and
Wang (2005). In addition, Liébana-Cabanillas et al. (2014), Hwang
(2014), and Yang et al. (2016) analyzed the effects of compatibility on
acceptance of smart homes, fintech, and wearable services, respec-
tively. Therefore, this study assumes that high compatibility of smart
home services could be a factor enhancing consumer convenience.
Personal information protection has also been recognized as an
J. Shin et al. Technological Forecasting & Social Change 134 (2018) 246–253

dependent variable for analyzing the diffusion of smart home products

is the intent to purchase timing of those smart home products. The
purchase timing has five options (within 6 months, 1 year, 2 years,
3 years, and beyond 3 years). The dependent variable has a value of 1 if
there is intent to buy at least one smart home product during that
timeframe.

3. Results

3.1. Adoption of a smart home device as part of a smart home system

The reliability coefficients for the measurement items of each

variable were calculated using the SPSS statistic 23 program in order to
examine the relevance of the measurement items included in the ex-
tended TAM. As shown in Table 1, the values of Cronbach's alpha for all
factors exceed 0.7, which is the general acceptance threshold. There-
Fig. 1. Extended TAM model. fore, all variables used in this analysis are statistically reliable.

3.1.1. Entire model

important factor related to intention to accept. Chen (2006) verified
To estimate the measurement model fit, we used four common
that anxiety about the collection and usage of personal information is
model-fit measures: Root Mean Square Error of Approximation
an important factor in mobile payment services. Yousafzai et al. (2010)
(RMSEA), Incremental Fit Index (IFI), Comparative Fit Index (CFI), and
and Yap et al. (2010) emphasized the importance of privacy in con-
Chi-square/degree of freedom (χ2/df). All of the model-fit indices sa-
sumer acceptance. This study defined the variable of personal in-
tisfy the acceptance criteria suggested in the previous literature
formation as the user's concern about the infringement of personal in-
(Table 2).
formation that occurs while using a service. These concerns include the
The estimation results for the entire sample show that compatibility,
possibility of leakage of personal information during the personal in-
ease of use, and usability had a significant effect on attitude (Table 3).
formation process such as the collecting, storing, using, and disclosing
In particular, the more compatible and more useful is the smart home
of information, as well as unwanted advertisements (Featherman and
service, the more enthusiastic are the respondents, and the greater is
Pavlou, 2003). Accordingly, this study assumed that concerns about
the intention to accept. Interestingly, privacy had no significant effect
personal information will have a negative impact on the acceptance of
on attitude. As Kraemer and Flechais (2018) point out, only a few
smart home services.
studies have investigated privacy in the context of smart homes. Ac-
Demographic characteristics of users (i.e., age, gender, income, and
cording to Zeng et al. (2017), users consider physical security more
education level) that affect acceptance intention were also considered
important than privacy when using smart homes.
in this study as moderating effects.
3.1.2. Analysis of differences according to demographic characteristics
2.2.2. Multivariate probit model
The main purpose of this study is to examine factors affecting the
The purchase of smart home products is expected to vary by pro-
adoption and diffusion of smart homes. Therefore, the effects of de-
duct, with respondents having different purchasing timeframes for each
mographic characteristics such as age, gender, education level, and
smart home product. Therefore, a multivariate probit model (MVP)2
income on the adoption and diffusion of smart homes were analyzed. In
was used to consider multiple choices, allowing users various purchase
the TAM analysis, the moderating effect was analyzed by dividing the
timing options for different smart home products. Previous studies have
310 respondents from the first survey into junior/senior, male/female,
used the MVP model to analyze relationships between over-the-top
high/low education, and high/low income groups.
services and traditional broadcasting services (Shin et al., 2016), smart
First, in order to analyze age effects, the total sample was divided
car and smart home services (Hong et al., 2016), and cloud service and
into junior (193 people) and senior (117 people) groups according to an
terminal devices (Shin et al., 2014). The utility function of the MVP
age cut-off of 40 years based on Serenko et al. (2006). The results are
model is as follows:
shown in Table 4. In the senior group, perceived usefulness had a
Uij = γj + ∑ βjd′ Yid + εij greater effect on attitude than did perceived ease of use. On the other
d (1) hand, in the junior group, the effect of perceived ease of use on attitude
was 0.513, which was significantly larger than in the senior group
Here, γj is an alternative specific constant (ASC) for each alternative
(0.206). In general, the junior group was more familiar with advanced
j, and Yid is a social demographic variable. The MVP model assumes that
ICT devices and more aware of the importance of user experience and
the stochastic εij has a multivariate normal distribution with zero mean
user interface.
and covariance matrix Σ. A Bayesian estimation method is used to re-
Second, the analysis of gender effects showed that personal in-
duce the complexity of the calculation process (Allenby and Rossi,
formation did not have a significant effect on attitude in males
1999; Edwards and Allenby, 2003; Huber and Train, 2001; Train,
(Table 4.). While significant for females, the effect size was very small
2003), which has advantages such as solving the integration problem
(−0.117). There were significant differences in compatibility and
for the multivariate probability density function and overcoming the
perceived usefulness by gender. The effects of compatibility on attitude
local optimization and initial value problem of the maximum likelihood
showed that men (0.310) were less affected by compatibility than
estimation method.
women (0.489). On the other hand, the influence of perceived useful-
The smart home products considered in this study include re-
ness was significantly higher in males (0.525) than females (0.310). In
frigerators, washing machines, TVs, air conditioners, boilers, dehumi-
other words, males tended to purchase smart home products even if
difiers, robot cleaners, smart meters, plugs, smart lighting, smart ther-
they were less compatible given that they were useful. On the other
mostats, smart switches, home CCTV, gas locks, and door locks. The
hand, women seemed to value the compatibility of the smart home
product with other products already owned even if the product by itself
2
See Edwards and Allenby (2003) for more detail about multivariate probit models. was useful.

248
J. Shin et al. Technological Forecasting & Social Change 134 (2018) 246–253

Table 1
Cronbach's alpha reliability and regression weights of observed variables.
Component Cronbach's α Unstandardized regression weight Standardized regression weight S.E. C.R. P

Compatibility
1 0.887 1.000 0.647
2 1.018 0.642 0.082 12.373 < 0.001
3 1.020 0.637 0.105 9.735 < 0.001
4 1.379 0.808 0.117 11.782 < 0.001
5 1.384 0.826 0.116 11.967 < 0.001

Privacy
1 0.828 1.000 0.717
2 1.034 0.721 0.067 15.426 < 0.001
3 1.302 0.893 0.090 14.443 < 0.001
4 1.292 0.886 0.090 14.411 < 0.001

Perceived ease of use

1 0.813 1.000 0.791
2 0.831 0.695 0.065 12.743 < 0.001
3 0.918 0.717 0.069 13.223 < 0.001
4 0.777 0.584 0.075 10.396 < 0.001

Perceived usefulness
1 0.830 1.000 0.819
2 1.034 0.857 0.059 17.584 < 0.001
3 0.864 0.772 0.055 15.654 < 0.001
4 0.998 0.832 0.059 16.787 < 0.001

Attitude
1 0.875 1.000 0.863
2 0.973 0.844 0.049 19.668 < 0.001
3 0.946 0.791 0.054 17.578 < 0.001
4 0.960 0.814 0.052 18.440 < 0.001

Intention to use
1 0.855 1.000 0.816
2 0.962 0.745 0.067 14.461 < 0.001
3 1.000 0.832 0.059 16.900 < 0.001
4 0.898 0.730 0.063 14.159 < 0.001

Note: S.E.E = standard error, C.R. = composite reliability.

Table 2 Finally, this study divided the respondents into high and low edu-
Model-fit indices. cation levels and examined differences by education level. The high
RMESA IFI CFI Chi-square/df education group showed a higher acceptance rate for smart home
products when compatibility was high, while the low education group
Level 0.059 0.949 0.949 2.064 was more likely to adopt when product usefulness was high (Table 4).
Acceptance level < 0.06 > 0.9 > 0.9 <3
Reference Bentler and Bagozzi and Bagozzi and Bentler and
Bonett (1980) Yi (1988) Yi (1988) Bonnet (1980)
3.2. Diffusion of smart home products

Table 3 The 2113 respondents from the second survey were used in the
Estimation results. diffusion analysis. The definitions of variables used are shown in
Table 5. The demographic variables were gender, age, education level,
Entire
monthly average income, number of household members, and type of
Unstandardized Standardized S.E. C.R. P value housing. In addition, the amount of ICT devices and home appliances
estimate estimate possessed is considered an independent variable because the current
status of ICT devices and home appliances is expected to have a sig-
Compatibility → 0.315⁎⁎⁎ 0.226 0.120 2.621 0.009
nificant impact on the purchase of smart home products. The level of
attitude
Privacy → −0.022 −0.020 0.030 −0.735 0.462 recognition and usability of smart home appliances and the importance
attitude of personal information protection were also considered as independent
⁎⁎⁎
PEoU → attitude 0.496 0.449 0.129 3.851 < 0.001 variables.
PU → attitude 0.355⁎⁎⁎ 0.340 0.077 4.583 < 0.001 According to the survey results, the most preferred intention to
Attitude → IU 0.954⁎⁎⁎ 0.952 0.056 17.082 < 0.001
purchase timeframe was ‘within one year,’ followed by ‘within two
Note: S.E. = standard error, C.R. = composite reliability; ⁎⁎⁎ indicates a sig- years,’ ‘within three years,’ within six months,’ and ‘after three years.’
nificance level of 99%; PEoU, PU, and IU indicate perceived ease of use, per- The percentage of respondents who want to purchase smart home
ceived usefulness, and intention to use, respectively. products within one year is much smaller than those who want to buy
after one year. Therefore, it is necessary to examine which factors affect
Third, looking at the differences according to income, compatibility timing purchase in order to promote the diffusion of smart home pro-
and perceived usefulness were more important to the low income ducts. In this context, we conducted two analyses to analyze the main
group, while perceived ease of use was more important to the high factors affecting purchase time: 1) purchase timing for overall smart
income group. However, this difference was not statistically significant home products and 2) purchase timing for smart home products by size.
(Table 4). We divided smart home products into three categories: 1) large home
appliances including refrigerators, washing machines, TVs, air

249
J. Shin et al. Technological Forecasting & Social Change 134 (2018) 246–253

Table 4
Estimation results moderated by age, gender, income, and education.
Group Junior Senior Difference

Unstandardized estimate S.E. C.R. Standardized estimate Unstandardized estimate S.E. C.R. Standardized estimate

⁎⁎⁎ ⁎⁎⁎
Compatibility → 0.433 0.064 6.770 0.482 0.406 0.079 5.172 0.412 0.027
Attitude (−0.262)
Privacy → attitude −0.093⁎ 0.055 −1.703 −0.100 −0.124⁎⁎ 0.061 −2.034 −0.137 0.031
(−0.318)
PEoU → attitude 0.513⁎⁎⁎ 0.076 6.785 0.536 0.206⁎⁎ 0.085 2.407 0.175 0.307⁎⁎⁎
(−2.694)
⁎⁎⁎ ⁎⁎⁎
PU → attitude 0.286 0.050 5.668 0.361 0.523 0.063 8.288 0.709 −0.237⁎⁎⁎
(2.936)
Attitude → IU 0.852⁎⁎⁎ 0.089 9.591 0.876 1.039⁎⁎⁎ 0.112 9.309 0.948 −0.187
(1.307)

Group Male Female Difference

Unstandardized estimate S.E. C.R. Standardized estimate Unstandardized estimate S.E. C.R. Standardized estimate

⁎⁎⁎ ⁎⁎⁎
Compatibility → attitude 0.310 0.061 5.064 0.364 0.489 0.077 6.322 0.513 −0.179⁎
(1.817)
Privacy → attitude −0.073 0.055 −1.334 −0.084 −0.117⁎ 0.060 −1.956 −0.128 0.044
(−0.540)
PEoU → attitude 0.401⁎⁎⁎ 0.087 4.625 0.367 0.375⁎⁎⁎ 0.076 4.914 0.387 0.026
(−0.228)
PU → attitude 0.525⁎⁎⁎ 0.065 8.084 0.658 0.310⁎⁎⁎ 0.050 6.152 0.432 0.215⁎⁎⁎
(−2.617)
⁎⁎⁎ ⁎⁎⁎
Attitude → IU 0.872 0.096 9.059 0.866 0.934 0.102 9.170 0.918 −0.062
(0.440)

Group Low income High income Difference

Unstandardized estimate S.E. C.R. Standardized estimate Unstandardized estimate S.E. C.R. Standardized estimate

Compatibility → attitude 0.406⁎⁎⁎ 0.065 6.286 0.430 0.370⁎⁎⁎ 0.085 4.373 0.416 0.036
(−0.334)
Privacy → attitude −0.109⁎ 0.060 −1.830 −0.106 −0.090⁎ 0.053 −1.682 −0.131 0.019
(0.239)
⁎⁎⁎ ⁎⁎⁎
PEoU → attitude 0.337 0.064 5.234 0.354 0.550 0.154 3.567 0.398 −0.213
(1.278)
⁎⁎⁎ ⁎⁎⁎
PU → attitude 0.462 0.054 8.514 0.594 0.343 0.067 5.151 0.459 0.119
(−1.383)
Attitude → IU 0.901⁎⁎⁎ 0.086 10.418 0.879 0.916⁎⁎⁎ 0.131 6.984 0.927 −0.015
(0.097)

Group Low education High education Difference

Unstandardized estimate S.E. C.R. Standardized estimate Unstandardized estimate S.E. C.R. Standardized estimate

⁎⁎⁎ ⁎⁎⁎
Compatibility → attitude 0.311 0.082 3.805 0.304 0.501 0.066 7.574 0.575 −0.190⁎
(1.809)
Privacy → attitude −0.044 0.071 −0.619 −0.045 −0.156⁎⁎⁎ 0.049 −3.180 −0.180 −0.093
(−1.299)
PEoU → attitude 0.548⁎⁎⁎ 0.111 4.922 0.460 0.300⁎⁎⁎ 0.062 4.835 0.319 0.248⁎
(−1.948)
PU → attitude 0.394⁎⁎⁎ 0.067 5.862 0.467 0.403⁎⁎⁎ 0.049 8.154 0.559 −0.009
(0.098)
⁎⁎⁎ ⁎⁎⁎
Attitude → IU 0.707 0.091 7.780 0.791 1.104 0.101 10.972 0.991 −0.397⁎⁎⁎
(2.926)

Note: S.E. = standard error, C.R. = composite reliability; ⁎, ⁎⁎, and ⁎⁎⁎ indicate significance levels of 90%, 95%, and 99% respectively; PEoU, PU, and IU indicate
perceived ease of use, perceived usefulness, and intention to use, respectively.

conditioners, boilers, and dehumidifiers; 2) small home appliances in-
cluding robot cleaners, smart meters, plugs, smart lighting, smart
thermostats, and smart switches; and 3) safety and security appliances
including home CCTV, gas locks, and door locks.
The results of analyzing the diffusion factors of smart home pro-
ducts and product size using the MVP model are shown in Tables 6 and
7, respectively. The effects of demographic characteristics on purchase
timing show that males tend to purchase more smart home products
within two years, but purchase timing varied depending on product
size. Older people are more likely to purchase smart home products
within six months or one year but prefer to purchase ‘small home ap-
pliances’ and ‘safety and security’ products within six months.
According to Kooti et al. (2016), the results of analyzing online
purchasing behavior showed that purchase time was delayed in the
younger age group when the purchase price was higher. Even though
young people are willing to purchase due to their curiosity regarding
new technology, purchasing smart home systems could be delayed in
younger people since it consists of household appliances that are

250
J. Shin et al. Technological Forecasting & Social Change 134 (2018) 246–253

Table 5 purchase ‘safety and security products’ within six months is relatively
Definitions of variables and sample statistics. high. Therefore, it is expected that early recognition of smart home
Variables Explanation Mean services will have little impact on the initial diffusion of most smart
(Std) home markets, except for safety and security products. Inconsistency in
attitude and behavior has been shown in previous research (e.g., Chung
Dependent Within_6M 1 = intention to purchase 0.125
and Leung, 2007; Cho and Shin, 2017; Claudy et al., 2013; Lapiere,
variables products within 6 months, (0.331)
otherwise 0
1934; Gupta and Ogden, 2006; Claudy et al., 2013). This tendency also
Within_1Y 1 = intention to purchase 0.258 appears in the spread of most smart home systems. On the other hand,
products within 1 year, otherwise (0.438) respondents who believe that smart homes are useful tend to buy within
0 one or two years, although safety and security products are commonly
Within_2Y 1 = intention to purchase 0.241
purchased earlier. These results suggest that strategies to increase re-
products within 2 years, (0.428)
otherwise 0 cognition of the usefulness rather than awareness of smart home ser-
Within_3Y 1 = intention to purchase 0.152 vices will be more effective for increasing the smart home market for
products within 3 years, (0.359) most smart home products. However, for safety and security products,
otherwise 0
strategies to increase recognition of the usefulness and awareness of
In_3Y 1 = intention to purchase 0.118
products after 3 years, otherwise (0.322)
smart home services will be effective for the diffusion of smart home
0 products.
Independent GENDER Male = 1, Female = 0 0.501 Finally, analysis of the effect of personal information protection
variables (0.500) shows that people who think that personal information is important
AGE Age 40.046
tend to postpone smart home appliance purchases. Therefore, in order
(11.314)
EDU Education level 15.525 to improve the smart home market, the personal information protection
(years) (2.081) issue should be resolved first.
INCOME Monthly income 477.840
(10 thousand Korean Won) (250.625)
N_Family Number of household members 3.378 4. Conclusion and remarks
(persons) (1.161)
T_Residence Residence type 0.668
For initial activation of the smart home market, it is necessary to
(apartment = 1, otherwise = 0) (0.471)
N_Device_I Number of information devices 1.920 establish a marketing strategy by setting the target consumers as those
(0.762) with high purchase intention and willingness to purchase in the near
N_Device_L Number of large home appliances 4.993 future. According to Rogers (2003), these consumers are likely to be
(1.307) early adopters and the early majority of the smart home market in
N_Device_S Number of small home appliances 1.556
(1.304)
terms of diffusion of innovation. Therefore, it is important to determine
A_Home Recognition of smart home 3.480 who will be the consumers with high and early adoption in order to
(5-point Likert scale) (0.792) overcome demand stagnation until popularized.
V_Home Usability of smart home 3.886 The results of this study show that compatibility, perceived ease of
(5-point Likert scale) (0.684)
use, and perceived usefulness had positive effects on purchase inten-
I_Personal Importance of personal 4.329
information protection (0.759) tion. Perceived usefulness was greater in the older and male groups,
(5-point Likert scale) while compatibility was more important in the high education and fe-
male groups. In terms of purchasing time, people who are older, have a
higher income level, and have more large household appliances are
relatively higher priced than other smart devices for individuals. The more likely to buy a smart home within one year. In addition, the
effects of the income variable indicate that the purchase of smart home usefulness of smart home services rather than pre-recognition had a
appliances will be faster at higher income levels. On the other hand, positive effect on purchase time. Unlike other advanced ICT services/
education level and housing type do not have a significant effect on the products, older consumers were more likely to purchase smart home
adoption time of smart home appliances. Adoption time of overall smart services and wanted to purchase relatively earlier than younger people.
home products is not statistically significant depending on family size. However, for continuous growth of the smart home market, market
However, adoption time of smart home products shows that relatively strategies for consumers with low purchase intention and who are late
small families postpone their purchase time of smart home products for in purchasing are also needed. In other words, it is necessary to plan the
two or three years depending on product size. Overall, older people diffusion of smart home services with strategies for the laggards group,
with higher income tend to purchase smart home appliances relatively which is the long tail part of the innovation diffusion curve. Among
early, while young people and small families tend to buy relatively late. consumers with relatively low purchase intention, the importance of
Analysis of the effect of prior recognition of smart home appliances perceived usefulness was low in younger generations and women, while
revealed that people who are aware of smart home appliances are likely the importance of compatibility was low in the low education group
to purchase them within two or three years, while the tendency to and men. Those consumers who wanted to delay purchases were

Table 6
The estimation results of diffusion.
Alternative MVP model

ASC GENDER AGE INCOME EDU N_Family T_Residence N_Device_I N_Device_L N_Device_S A_Home V_Home I_Personal

1 Within_6M −4.875** 0.142* 0.006* 0.0003** −0.003 0.003 0.042 0.0350 0.045* 0.072** 0.136** 0.411** −0.121**
2 Within_1Y −7.557** 0.191 0.010* 0.0003** −0.0008 0.007 0.035 0.0350 0.030 0.118** 0.192** 0.754** −0.182**
3 Within_2Y −7.030** 0.431** 0.009 0.0003* −0.022 −0.017 0.085 0.001 −0.047 0.124** 0.276** 0.754** −0.140*
4 Within_3Y −5.973** 0.335** −0.003 −0.0001 −0.011 −0.037 0.089 0.007 −0.008 0.045* 0.266** 0.595** −0.120*
5 In_3Y −5.148** 0.254* −0.012** −0.0002 −0.020 0.039 0.050 −0.001 −0.025 −0.042* 0.091* 0.572** 0.078

⁎⁎ ⁎
Note: Significant at a 5% level; Significant at a 10% level.

251
J. Shin et al. Technological Forecasting & Social Change 134 (2018) 246–253

younger, had fewer large home appliances, and cared about privacy

−0.1447**

−0.1244**
−0.1331**

−0.1455**
−0.1211**
protection. Overall, younger consumers had lower intention to pur-

−0.0891*

−0.0796*
−0.0532*

−0.2011*
I_Personal

−0.0024
−0.0267

−0.0882
0.0760*
0.0454

0.0894
chase.
Income and the importance of protecting personal information had
0.2222** a significant effect on purchase timing but not on purchase intention.
0.3414**
0.3243**
0.2662**
0.3644**
0.4026**
0.4054**
0.3212**
0.2874**
0.3215**
0.9314**
0.6638**
0.3989**
0.4499**
0.4970**
Personal information is emerging as an important issue as the number
V_Home

of IoT-based services, including smart homes, increases. However, ac-

cording to this study, potential smart home users are not concerned
about personal information protection in their decision on when to
0.0792**
0.0855**
0.1113**

0.1054**
0.1240**
0.1293**
0.0847**

0.2950**
0.1899**
0.2186**
0.2721**
A_Home

0.0799*

0.0512*

adopt smart home services. Therefore, more detailed analysis is re-

0.0366

0.0202
quired to understand the different factors influencing purchase inten-
tion and purchase timing.
The following strategic implications can be obtained from this
−0.0619**
N_Device_S

−0.1181*
0.1250**
0.1685**
0.1372**

0.1223**
0.1379**
0.1212**

0.2557**
0.1704**
−0.0389

0.0860*
study. Smart home operators need to establish cooperative relationships
0.0329

0.0371

0.0345
with a wide range of providers as well as other smart home operators to
ensure interoperability and compatibility between services/products, as
compatibility had a significant positive effect on purchase intention. In
N_Device_L

−0.0582*

−0.0568*
−0.0164

−0.0039
−0.0233

−0.0066
−0.0761
−0.0933

fact, many smart home operators are already working together on

0.0752*
0.0489
0.0339

0.0084

0.0407

0.0976
0.0637

building smart home ecosystems. For example, Nest Labs, one of the
leading smart home operators, provides various smart home products
such as learning thermostats, security cameras, and door-bells run by a
N_Device_I

−0.0285

−0.0702

program called ‘Works with Nest’ in order to collect as many partners as

0.1286*
0.0476
0.0265
0.0356
0.0414

0.0604
0.0052
0.0182
0.0218
0.0848
0.1051

0.0946
0.0274

possible.
Usefulness was a significant factor affecting both the adoption and
purchase timing of smart homes; thus, smart home operators should
T_Residence

also consider how smart homes can create a direct utility to consumers.
−0.0105

−0.0015
0.0358

0.0578
0.0308

0.0008
0.0014
0.0166
0.0636
0.0108
0.0568
0.0447
0.0029
0.0326
0.0500

Unlike other ICT products, consumers already own a large number of

home appliances, so they are likely to have a low incentive to purchase
a new smart home product unless the smart home product offers dif-
−0.0719*

−0.0660*

−0.1046*
−0.0235
−0.0074
−0.0096

−0.0231

−0.0843

−0.0662
N_Family

ferentiated value. Therefore, operators need to create additional value

0.0672*
0.0508
0.0084
0.0001

0.0158

0.0675

beyond simply connecting to the Internet.

According to this study, the main target customers for smart homes
are older people with higher income. Thus, it is necessary to think about
−0.0297*
−0.0090

−0.0166
−0.0285
−0.0207
−0.0259
−0.0233
−0.0147
−0.0143
−0.0598
−0.0324
−0.0335

−0.0461

making services/products intuitive and convenient for the older age

0.0080

0.0034

group who are relatively unfamiliar with new technologies. Smart TVs
EDU

are a good example. Various types of remote controls have been de-
veloped and provided for convenient use of smart TVs, but the smart
−0.0005**

−0.0002**

functions of the TVs are not used much.

0.0005**
−0.0001

−0.0001
−0.0001

−0.0001
INCOME

0.0005*
0.0003*
0.0003*

0.0004*
0.0002

0.0003

0.0002
0.0001

This study has limitations in that it does not reflect the revealed
preference of consumers since survey data was used. However, as the
smart home market is in an early stage, it is almost impossible to obtain
−0.0110**

−0.0145**

−0.0285**

market data, especially personal microdata. In this situation, it is ex-

−0.0003

−0.0022

−0.0033
0.0085*

0.0142*
0.0126*
0.0026
0.0036
0.0036

0.0054
0.0004

0.0046

pected that this study can provide useful information on the smart
AGE

home market. In addition, it is difficult to generalize the results to

global consumers as the sample of this study was South Korean con-
MVP model

sumers. Despite these limitations, this study is meaningful as the first

0.1389**

0.2249**

0.2563**
GENDER

0.1458*

0.1254*

0.1281*
0.2229*
0.1828*

0.1506*

0.2235*
0.0795

0.0440
0.2666
0.0771

0.0959

attempt to simultaneously analyze the adoption and purchase timing of

Significant at a 10% level.

smart home systems rather than individual smart home services/pro-

ducts. With the evolution of smart homes, new characteristics and is-
−10.5684**
−4.1523**
−3.5435**
−3.1925**
−3.0251**
−3.2176**
−5.2821**
−4.3606**
−3.5990**
−2.9999**
−3.4431**

−6.8875**
−4.2168**
−5.1161**
−3.9581**

sues that were not considered in this study may emerge. We hope that
The estimation results of diffusion by product size.

this study will trigger more discussion and further studies on the smart
ASC

home market.
Within_6M

Within_6M

Within_6M

References
Within_1Y
Within_2Y
Within_3Y

Within_1Y
Within_2Y
Within_3Y

Within_1Y
Within_2Y
Within_3Y
In_3Y

In_3Y

In_3Y

Allenby, G.M., Rossi, P.E., 1999. Marketing models of consumer heterogeneity. Journal of
Significant at a 5% level;

Econometrics 89, 57–78.

Bagozzi, R.P., Yi, Y., 1988. On the evaluation of structural equation models. J. Acad.
Safety and security appliance

Mark. Sci. 16 (1), 74–94.

Bentler, P.M., Bonett, D.G., 1980. Significance tests and goodness of fit in the analysis of
covariance structures. Psychol. Bull. 88 (3), 588.
Small home appliance
Large home appliance

Chen, L.D., 2006. A theoretical model of consumer acceptance of payment. In: AMCIS
2006 Proceedings, pp. 247.
Chen, J.C.V., Yen, D.C., Chen, K.C., 2009. The acceptance and diffusion of the innovative
smart phone use: a case study of a delivery service company in logistics. Information
Alternative

Management 46, 241–248.

Table 7

⁎⁎

Cho, M., Shin, J., 2017. Market strategy for promoting green consumption: consumer
Note:

preference and policy implications for laundry detergent. International Journal of

Consumer Studies 41 (3), 283–290.

252
J. Shin et al.
Chong, G., Zhihao, L., Yifeng, Y., 2011. The research and implement of smart home
system based on internet of things. In: Electronics, Communications and Control
(ICECC), 2011 International Conference on. IEEE, pp. 2944–2947.
Chung, S., Leung, M.M., 2007. The value-action gap in waste recycling: the case of un-
dergraduates in Hong Kong. Environmental Management 40, 603–612.
Claudy, M.C., Peterson, M., O'Driscoll, A., 2013. Understanding the attitude-behavior gap
for renewable energy systems using behavioral reasoning theory. J. Macromark. 33
(4), 273–287.
Corrocher, N., 2011. The adoption of Web 2.0 services: an empirical investigation.
Technological Forecasting and Social Change 78, 547–558.
Davis, F.D., 1989. Perceived usefulness, perceived ease of use, and user acceptance of
information technology. MIS Q. 13 (3), 319–340.
Edwards, Y.D., Allenby, G.M., 2003. Multivariate analysis of multiple response data.
Journal of Marketing Research 40 (3), 321–334.
Featherman, M.S., Pavlou, P.A., 2003. Predicting e-services adoption: a perceived risk
facets perspective. International Journal of Human Computer Studies 59, 451–474.
Feng, S., Setoodeh, P., Haykin, S., 2017. Smart home: cognitive interactive people-centric
internet of things. IEEE Communications Magazine 55 (2), 34–39.
Greenough, J., 2016. The US smart home market has been struggling — here's how and
why the market will take off. In: Business Insider, Retrieved from. http://www.
businessinsider.com/the-us-smart-home-market-report-adoption-forecasts-top-
products-and-the-cost-and-fragmentation-problems-that-could-hinder-growth-
2015-9.
Gupta, S., Ogden, D., 2006. The attitude-behavior gap in environmental consumerism.
APUBEF Proc. 3 (1), 199–206.
Hong, J., Lee, D., Shin, J., 2016. Strategic management of next-generation connected life:
focusing on smart key and car-home connectivity. Technological Forecasting and
Social Change 106, 11–20.
Huber, J., Train, K., 2001. On the similarity of classical and Bayesian estimates of in-
dividual mean partworths. Marketing Letters 12, 259–269.
Hui, T.K., Sherratt, R.S., Sánchez, D.D., 2017. Major requirements for building smart
homes in smart cities based on internet of things technologies. Future Generation
Computer Systems 76, 358–369.
Hwang, C., 2014. Consumers' acceptance of wearable technology: examining solar-pow-
ered clothing. In: Graduate Theses. Iowa State University.
Kim, Y., Lim, S.E., Choi, J., 2016. Estimation of willingness to pay for smart home service
by contingent valuation method. J. Korean Soc. Qual. Manag. 44 (4), 833–843.
Kooti, F., Lerman, K., Aiello, L.M., Grbovic, M., Djuric, N., Radosavljevic, V., 2016.
Portrait of online shopper: understanding and predicting consumer behavior. In:
Proceedings of the 9th ACM International Conference on Web Search and Data
Mining, San Francisco, CA, USA, 22–25 February 2016.
Kraemer, M.J., Flechais, I., 2018. Researching privacy in smart homes: a roadmap of
future directions and research methods. In: The 2018 IET Living in the Internet of
Things: Cybersecurity of the IoT Conference.
Lapiere, R.T., 1934. Attitudes vs. actions. Social Forces 13 (2), 230–237.
Legris, P., Ingham, J., Collerette, P., 2003. Why do people use information technology? A
critical review of the technology acceptance model. Inf. Manag. 40 (3), 191–204.
253
Technological Forecasting & Social Change 134 (2018) 246–253
Li, B., Yu, J., 2011. Research and application on the smart home based on component
technologies and Internet of Things. Process Engineering 15, 2087–2092.
Liébana-Cabanillas, F.J., Sánchez-Fernández, J., Muñoz-Leiva, F., 2014. Role of gender on
acceptance of mobile payment. Industrial Management & Data Systems 114 (2),
220–240.
Mao, X., Li, K., Zhang, Z., Liang, J., 2017. Design and implementation of a new smart
home control system based on internet of things. In: Smart Cities Conference (ISC2),
2017 International. IEEE, pp. 1–5.
Park, E., Kim, S., Kim, Y., Kwon, S.J., 2018. Smart home services as the next mainstream
of the ICT industry: determinants of the adoption of smart home services. Universal
Access in the Information Society 17 (1), 175–190.
Perumal, T., Sulaiman, M.N., Sharif, K.Y., Ramli, A.R., Leong, C.Y., 2013. Development of
an embedded smart home management scheme. Int. J. Smart Home 7 (2), 15–26.
Richter, L.L., Pollitt, M.G., 2016. Which smart electricity service contracts will consumers
accept? In: The Demand for Compensation in a Platform Market.
Rogers, E.M., 2003. Diffusion of Innovations, 5th ed. Free Press, New York.
Serenko, K., Turel, O., Yol, S., 2006. Moderating roles of user demographics in the
American customer satisfaction model within the context of mobile services. J. Inf.
Technol. Manag. 17 (4), 20–32.
Shin, J., Cho, M., Lee, J., Lee, D., 2014. Strategic management of cloud computing ser-
vices: focusing on consumer adoption behavior. IEEE Transactions on Engineering
Management 61 (3), 419–427.
Shin, J., Park, Y., Lee, D., 2016. Strategic management of over-the-top services: focusing
on Korean consumer adoption behavior. Technological Forecasting and Social
Change 112, 329–337.
Soliman, M., Abiodun, T., Hamouda, T., Zhou, J., Lung, C.H., 2013. Smart home: in-
tegrating internet of things with web services and cloud computing. In: Cloud
Computing Technology and Science (CloudCom). 2013 IEEE 5th International
Conference on 2. IEEE, pp. 317–320.
Statista, 2018. Statistics Portal: Smathome. Retrieved from. https://www.statista.com/
outlook/279/100/smart-home/worldwide.
Stojkoska, B.L.R., Trivodaliev, K.V., 2017. A review of internet of things for smart home:
challenges and solutions. Journal of Cleaner Production 140, 1454–1464.
Train, K., 2003. Discrete Choice Methods with Simulation. Cambridge University Press,
Cambridge.
Wu, J.H., Wang, S.C., 2005. What drives mobile commerce? An empirical evaluation of
the revised technology acceptance model. Information Management 42, 719–729.
Yang, H., Yu, J., Zo, H., Choi, M., 2016. User acceptance of wearable devices: an extended
perspective of perceived value. Telematics and Informatics 33, 256–269.
Yap, K.B., Wong, D.H., Loh, C., Bak, R., 2010. Offline and online banking-where to draw
the line when building trust in e-banking? The International Journal of Bank
Marketing 28 (1), 27–46.
Yousafzai, S.Y., Foxall, G.R., Pallister, J.G., 2010. Explaining internet banking behavior:
theory of reasoned action, theory of planned behavior, or technology acceptance
model? Journal of Applied Social Psychology 40, 1172–1202.
Zeng, E., Mare, S., Roesner, F., 2017. End user security & privacy concerns with smart
homes. In: Symposium on Usable Privacy and Security (SOUPS).