Determinants of information communication and smart home automation technology adoption for aging-in-place

Abstract

Purpose –

Smart home (SH) internet of things can promote home safety, health monitoring and independence of older adults to age-in-place. Despite its commercial growth, low adoption rates of the technology among aging consumers remain a major barrier. The purpose of this study is to examine SH technology ownership of older adults and its causal pathways with demographics, health and functioning, home safety and information communication technology (ICT) use.

Design/methodology/approach –

A survey on technology-mediated aging-in-place was completed by 447 respondents, 65 years and older. Structural equation modeling was used to underscore the causal pathways among demographics, health, independence and home safety, ICT and home automation technology adoption.

Findings –

The study found that ICT usability, home security and independence have a significant direct effect on SH ownership. Demographics had no significant direct effect, but its influence was mediated through ICT usability. With home safety as mediator, physical impairment, falls and accidents and independence had a significant association with SH ownership. Similarly, increased social support (mediated through home security) decreased the probability of SH automation ownership.

Originality/value –

The findings signify the perceived usefulness of SH automation as theorized in technology acceptance models.

Introduction

As innovations in smart home (SH) automation technology surge, there is a growing interest in their potential for older adults to age-in-place. From the popular voice-activated assistants, body-worn health monitoring devices and smart hubs to control the home environment, SH devices can promote safety and security, emergency response and independence of the elderly. A globalized economy combined with widespread access to the internet and wireless technologies (such as wifi and Bluetooth) have made SH technology more available and affordable for consumers. With an ever-growing aging population and rising incidence in conditions such as Alzheimer’s disease and related dementia, the scope of commercially available mainstream SH devices is becoming more and more evident in the context of aging-in-place. In contrast, the opposing forces of “information age” and the “digital divide” have intrigued technology proponents and gerontologists alike. A multi-faceted approach is required to examine the diffusion of SH technology among the aging. This research study explores key determinants and the causal links among a wide array of demographic, health and functioning attributes leading to the adoption of SH technology for aging in place.

Background

SH technology is a growing consumer segment within the information communication technology (ICT) industry. The global home automation market, which was valued at $39.6bn in 2016 is now projected to rise to $81bn by 2023 at 11.2% annual growth rate (Allied Market Research, 2019). Popularity of SH devices has been spurred by their off-the-shelf characteristics including wireless connectivity with the internet and ICT devices on the same network, termed as the internet of things (IoT). In combination, these characteristics allow consumers of SH devices to:

• form sensor networks to monitor and gather information about the state of the home and its residents;

• to communicate between devices to enable automation and remote access; and

• link with user interfaces such as home displays, personal computers, tablets and smart phones to set preferences/goals, as well as to receive information and feedback (Cook, 2012; Hargreaves and Wilson, 2013).

The demand of SH automation among mainstream consumers has been driven by their potential to offer comfort, convenience and energy efficiency with home living. In the case of aging consumers, however, SH technology may not just be matter of luxury, but the difference between being able to age-in-place versus premature relocation to an institutional facility out of concern for safety, security and functional capacity. On this note, there are financial implications. The current annual cost for a resident at an assisted living facility in the USA is $48,000 (Genworth, Inc, 2018), which had the highest rise past year among all institutional care segments at 6.67% (Senior Housing News, 2018).

In a survey conducted in the USA in 2018, 3 out of 4 people older than 50 preferred to age-in-place in their home and community (American Association of Retired Persons [AARP], 2018). The trend remains the same globally with several countries committing to promote and prolong home living of the elderly as opposed to paying for long term institutional care (European Parliamentary Research Service, 2014; Nakanishi et al., 2015). SH technologies can monitor older adults’ health status, detect hazards and falls, seek emergency assistance, reduce the burden on house-hold tasks and help manage personal routines through programmed calendar and reminders (Cheek et al., 2005; Liu et al., 2016). Though much of the research on SH and aging is still preliminary, a review of the literature indicates that older adults readily accept the technology once they realize the benefits with physical activity, independence and function (Morris et al., 2013). In a study involving family caregivers of individuals with Alzheimer’s disease, the caregivers expressed appreciation for SH technologies for their potential to increase their care recipient’s safety and prevent wandering, as well as facilitate their own social connectedness and freedom to leave home (Rialle et al., 2008). Although the potential of SH technology for aging-in-place is evident here, barriers to its adoption is equally prominent.

A fundamental barrier may have to do with limited domain knowledge and overall experience of older adults with ICT, a precursor to the successful implementation of SH. In a PEW research study, it was found that older adults had far lesser ownership and exposure to broadband internet, computer, tablets and social media when compared to the younger and middle-age consumers (Pew Research Center, 2017). Specific to SH, studies also confirm the prevailing digital divide and the notion of older adults being the late adopters in the scheme of diffusion. In an earlier analysis from this survey research (N = 446), a very small proportion of older adults owned the most innovative and popular SH devices including voice activated assistants (19.3%), emergency alerting system (17.2%), remote monitoring (5%), motion activated camera (7.7%) and smart hubs (5.4%) in comparison to other conventional house hold devices (Arthanat et al., 2018).

There are several notable challenges in adapting SH technology to the home living needs of the elderly especially when dealing with neurocognitive and functional declines associated with conditions such as Alzheimer’s disease. Recent reviews suggest practical drawbacks with SH including issues with usability, robustness, portability, response to alerts, ethics and privacy (Gagnon-Roy et al., 2017), as well as low readiness of SH for implementation in homes of older adults (Liu et al., 2016). Research on the development and consumer adoption of SH technologies indicate that the technology is designed with home energy efficiency, leisure and security in mind, while primarily targeting young to middle-aged consumers and their children (Hargreaves and Wilson, 2013; Wilson et al., 2017). In a study in which caregivers expressed appreciation for SH technology (Rialle et al., 2008), a clear dichotomy in the findings was evident with a cluster of caregivers indicating aversion to SH devices that offered tracking, surveillance and activity assistance.

Problem statement

While it is clear from preliminary research that SH technologies have clear potential for aging-in-place, there are also several concerns regarding their implementation and adoption among the older population. Although there has been extensive research in the past decade on ICT adoption among the aging, in depth examination of how that translates to SH technology is still forthcoming. What is evident from literature is that factors contributing to SH adoption are multi-faceted (Lee and Coughlin, 2015; Mitzner et al., 2010; Pal et al., 2018). Studies involving technology acceptance models show that the older adults’ intention to adopt SH technology is vastly explained by perceived effort and perceived usefulness, two constructs with intricate factors embedded in them (Pal et al., 2018; Peek et al., 2014; Venkatesh et al., 2003). Although inherent technology acceptance behavior has been found to be the latent precursor to SH adoption, it is important to also examine some of the pre-dispositional characteristics within this population that pave way to ownership of SH devices. The purpose of this study is to use structural equation modeling (SEM) to identify the causal pathways that exist among key demographic, health, safety and functional variables that influence information communication and SH technology adoption among the aging.

Methods

This study was based on a survey of older adults residing in the New England region of USA. The overarching goal was to examine a comprehensive set of factors associated with aging in place with major emphasis on ICT, SH automation and home modifications. Additional data segments pertaining to demographics, health, function and safety were important constructs in the study. The research protocol was reviewed and approved by the institutional review board for human subject protection at the University of New Hampshire.

Analytical framework and variables

The hypothesized framework for the SEM is presented in Figure 1. The main hypothesis was that demographics, home safety, health and functioning and ICT use, as the constructs in the data, will predict SH automation technology ownership. To account for mediating factors, the following sub-hypotheses were tested:

• H1a.Based on established literature, the researchers hypothesized the link between demographic variables (age, gender, income and education) and ICT usability (Pew Research Center, 2017; Vroman et al., 2015) and from their past research that found ICT ownership to be a significant predictor of SH automation technology ownership (blinded for review), ICT usability was considered to have a mediating effect on SH automation ownership.
• H1b.For home safety, the hypothesis was that the presence of a physical, sensory or cognitive impairment, as well as incidents of falls and accidents, will have a negative effect on home safety, which will then influence ICT usability and SH automation. Also, self-reported independence in daily routines will be related to home safety as a mediating factor.
• H1c.Home safety and social network was hypothesized to have an inverse relation to SH automation ownership. In other words, adequate social support from family, friends and the community will reduce the perceived need for SH automation leading to lesser ownership.

Figure 1:

Analytical Framework

For demographics, age, gender, marital status, education and income were the predictor variables. For the construct of health and functioning, participants reported the presence of a physical, cognitive and/or sensory impairment and their rating of independence in daily routines on a five-point Likert scale (very independent, independent, somewhat independent, dependent or very dependent). In terms of home safety, the variables were any history of falls or accidents, perception of home security in response to the question “please rate how you feel about the security of your home,” (rated as very secure, secure, somewhat secure, insecure or very insecure) and a latent variable on home safety derived from items on safety with nine home activities. The activities included entering and exiting the house; moving around the house; climbing stairs; carrying and lifting; bathing; toileting; cooking; cleaning and laundry (all rated as very safe, safe, somewhat safe, unsafe or very unsafe). Social network was derived as a latent variable from five items pertaining to:

[…] support from close family and relatives living close by, feeling connected with close family and relatives living far away, feeling connected to neighbors, having friends in the community, and activities to engage with people in the community.

For the dependent variables, the researchers accounted the ownership and use of computer, internet, smartphone, tablet and smart watch as an indicator of ICT usability. The survey question asked respondents if they owned these technologies and how effectively they can use them on a six-point scale (0-Do not own or use, 1-Very ineffective, 2-Ineffective, 3-Somewhat effective, 4-Effective and 5-Very effective). A latent variable on ICT usability was derived from these items. Finally, SH automation ownership was computed as the manifest variable based on the ownership of up to seven most popular SH devices in the market-home security system, emergency alerting system, motion activated camera, auto-set thermostat, voice-activated assistant, smart hub to control lights or appliances via phone and remote home monitoring applications on smart phone.

Participants

The study involved convenience sampling of individuals 65 and older. They were eligible to participate if they lived in the community including independent homes, apartments or senior housing facilities. Conceivably, older adults living in long term care facilities such as assisted living and nursing homes were not considered. The target sample size was 400, which is the recommended number for any population above 10,000 people and for surveys that involve categorical data and estimated margin of error of 0.05 (Bartlett, Kotrlik and Higgins, 2001). The sample size was also deemed adequate for SEM analysis. For SEM, the sample size requirements for stability have varied from a minimum of 30 to a maximum of 460 subjects depending on the number of indicators, factor loading, statistical power, the bias in the parameter estimates and overall solution propriety (Wolf et al., 2013).

Survey questionnaire

A rigorous multi-step process was adopted to develop the survey questionnaire. Initially, the researchers identified key domains attributed to aging-in-place through a literature review. These domains included health, home activities, home safety, technology, community access and social support network. Interviews were then conducted with 10 older adults who were purposively sampled based on chosen characteristics of income, health history and living situation. The sample included individuals in high and low income brackets (below the regional median), living alone or with family, and those with and without a chronic disability. The interview questions centered on identified domains and perspectives on SH technology. Three members in the team coded the data independently and met for discussion and triangulation at three scheduled intervals after analysis of the first three, second three and last four interview data, respectively. Although the perspectives varied, data saturation (i.e. the absence of distinct codes and themes) was observed by the 10th interview indicating sampling adequacy. The researchers derived measurable indicators from the content analysis as questions on the survey and organized them into sections corresponding to the aging-in-place domains.

The survey draft was pilot tested with 12 older adults who first completed the questionnaire and then provided feedback and suggestions through a 2 h focus group. Note that these older adults were not part of the initial interviews. The participants were asked to discuss their views on each survey section. The final survey was created following the analysis of the audio-recorded data. Revisions to the draft included addition of items of a few items in each section and changing the wording on some items for clarity. None of the participants who contributed to the survey development were in the survey sample.

Data collection

The survey was uploaded online on Qualtrics® survey platform. Two sampling sources and administration methods were used for survey completion- Qualtrics® panel and one-on-one interviews with older adults by students in the occupational therapy program. In general, online survey panels are now being used as a popular sampling source for surveys and recent studies that compared the two methods found the quality of the data sets to be comparable (Heen et al., 2014; Weinberg et al., 2014). Using the Qualtrics® panel, participants were recruited through a large sample pool organized by the company in various demographic and customer profiles. Potential respondents are pre-registered in the panels and provide their personal and demographic information. Qualtrics sends out invitations to participate in surveys that match their profile and interests.

For the panel, older adults in the New England region of USA were sampled. Respondents were screened out automatically if they did not meet the inclusion criteria of age and living situation. To ensure response quality, the first 50 responses were reviewed by a coordinator at Qualtrics® for reliability and any missing sections of data. Once reliability of the panel was established, the survey was launched. A student research assistant monitored all individual responses thereafter. The survey was limited to one individual per household and verified through geographical location and IP address. Response validity was also examined through comments at the end of each survey section. To ensure reliability, the survey was set up with quality filters, which are dummy questions at random points in the survey, in which participants were expected to respond to a question using a certain choice on the given Likert scale. This feature indicated that participants were paying attention to the questions and any response that deviated from the assigned choice were screened out for review and possible exclusion.

The online survey was active from March to May of 2018. The average time for response completion was about 15 min. The survey administered by interviews took nearly 25min. Respondents were not offered any direct incentives through the research project and were invited only based on their interest with the research topic. Please note though that respondents in Qualtrics panel may have received a small incentive based on the length of the survey, their specific panelist profile and survey completion difficulty. Incentives that the company offers include cash, airline miles, gift cards, redeemable points, sweepstake entries and vouchers.

Data analysis

Demographic data and descriptive analyzes of variables were conducted using SPSS, version 25.0 (IBM). SEM was used to examine the relationship among latent constructs. STATA (StataCorp LLC, version 15) was used to fit the SEM model. The model tested the study hypotheses that demographics, health functions, safety and ICT will influence SH automation ownership and also identified the mediating effects in the sub-hypotheses (Figure 1). To simplify the model fitting procedure, the SEM model was fitted in two steps. First, a measurement model was fitted using exploratory factor analysis to derive the latent factor structures on home safety, social support and ICT usability. The factor scores to quantify these latent constructs were then created and used in the SEM to examine the relationship among them. The regression-based analysis for SEM modeling produced both direct and indirect effects on SH automation and other endogenous variables through hypothesized pathways. The model was also adjusted by potential moderating variables including social-demographic characteristics (age, gender, education and income etc.). The model fitting indices for comparative fit index (CFI) and non-normed fit index (NNFI or TLI) were also obtained to assess the model fitting adequacy with satisfactory value of 0.95 or above. The researchers specified the robust estimation for the variance-covariance matrix in a SEM model fitting process so that the model could be estimated under more relaxed assumptions (i.e. no requirement for normal distribution and identically distributed from the current observation to the next).

Results

The survey was completed by 454 respondents. Three responses were identified as questionable based on the quality filters and excluded from the data. Also, four individuals who were administered the survey by interview were excluded as their reported age was below 65. Therefore, the final sample size for the study was 447. In total, 416 responses were gathered online through Qualtrics and the remaining 31 were completed in-person by students. While the survey used two distinct data sources with the vast portion of data collected through the online panel, a previous study conducted on this data showed no noticeable differences found in key measures of SH ownership between the two sample cohorts (Authors blinded for review).

Demographics

Table 1 displays the key demographics for the 447 respondents. The average age of the sample was 70.9 (SD = 5.33). The major cohorts in the sample were women (68.8%), those who were married (54.9%) and those that obtained a college education and higher (58.6%). The household income was consistently distributed in all socio-economic categories with about 42.9% reporting income more than $60,000, around the national median household income. The majority (55%) lived with a spouse or partner, while 35% resided alone. There was an even distribution of respondents living in cities (20.4%), towns (23.7%), suburbs (28.4%) and rural small towns (27.5%).

Table 1:

Participant Demographics (N=447)

Characteristics		Mean (S.D)	Range
Age (N=447)		70.9 (5.33)	65–95
		Percentage	Frequency
Gender (N=445)	Males	31.2%	139
	Females	68.8%	306
Marital Status (N=443)	Single	18.7%	83
	Married	54.9%	243
	Divorced	20.8%	92
	Separated	1.6%	7
	With partner	4.1%	18
Education (N=443)	Below high school	0.5%	2
	Completed high school	18.6%	83
	Some college	22.1%	98
	Associate degree / Diploma	10.7%	48
	Bachelor’s degree	25.3%	112
	Master’s degree or higher	22.6%	100
Income (N=439)	Below $15,000	5.2%	23
	$15,000 to $30,000	18.0%	79
	$30,000 to $45,000	16.9%	74
	$45,000 to $60,000	17.1%	75
	$60,000 to $75,000	10.3%	45
	$75,000 to $90,000	10.3%	45
	Above $90,000	22.3%	98
Employment (N=440)	Full time	7.6%	34
	Part time	8.1%	36
	Self-employed	4.7%	21
	Unemployed & seeking job	1.6%	7
	Retired	76.9%	342
Ethnicity (N=444)	White (Caucasian)	95.7%	425
	African American	1.1%	5
	Hispanic or Latino	0.7%	3
	Asian	0.9%	4
	Native Indian	0.4%	2
	Other	1.1%	5
Place of living (N=447)	City	20.4%	91
	Town	23.7%	106
	Suburb	28.4%	127
	Rural or Small town	27.5%	123
Living situation (N=447)	With spouse or partner	55.3%	247
	Alone	35.1%	157
	With family	9.6%	43

Health and functioning

The majority of participants (73%) reported having a medical condition many of which were chronic yet manageable such as hypertension and diabetes. Nearly (23%) indicated having a physical impairment impacting their mobility or balance and about 34.2% reported a sensory impairment with vision or hearing and 10.1% had a cognitive impairment. About, (14%) of older adults reported at least one injurious fall or accident at home. A great proportion of the sample felt independent (15.9%) to very independent (77.6%) when asked to rate their independence in managing daily personal routines on a five-point Likert scale.

Information communication technology ownership

With respect to ownership of ICT, a very high proportion of older adults had a computer (98.7%) and internet connection (99.6%) and 63.4% and 66.3% indicated they were very effective in using them, respectively. In terms of the devices that were applicable to SH automation, majority of respondents used a smart phone (80.8%) with about half of them (46.3%) indicating they could use it effectively. A tablet was owned by 74.3% of older adults and 42.7% reported they were very effective in using it, while a smart watch with some form of activity tracking was being used by 52.8% and only 17.4% felt they used it very effectively.

Home automation devices

Respondents could receive an ownership score from 0 up to 7 (if they owned all seven of the listed popular home automation devices). However, the average ownership was very low at 1.18 (SD = 1.23) and median ownership was one device for the majority (32.7%) of them. The devices that were most owned were automatic thermostats (43.2%), voice activated assistants (19.3%) and home security systems (19.8%), while the least owned devices were motion activated cameras (8%), remote activation for lights and appliances (5.5%) and smartphone remote monitoring (5.1%).

Path model

Prior to the SEM analysis, factor analysis of the latent variables of home safety, social network and ICT usability showed that the corresponding items loaded into one factor with eigenvalues of 5.71, 2.19 and 2.04, respectively. Table 2 depicts the direct, indirect and total effects of the predictor variables on SH automation ownership and its mediating factors of ICT usability, home safety and home security. The path is displayed in Figure 2. The model was adequately fitted by 0.960 on CFI and 0.964 on NNFI (TLI). For the main hypothesis 1 (H1), none of the demographic variables were significantly correlated (directly or indirectly) with SH automation ownership. However, lower than college education and income bracket (<45k) had negative direct effect on SH automation with borderline significance (p < 0.10). The total effect of college education and income was also negative with borderline significance (p = 0.07 and 0.06, respectively). Increased self-reported independence with daily routines was negatively correlated to SH automation directly (p < 0.05) but not through other hypothesized pathways (indirect effect with p = 0.15). Home safety was significantly related to home automation only through indirect reciprocal pathways for home security (p < 0.01). Both home security (p < 0.01) and ICT usability index (p < 0.5) were positively related to SH automation ownership directly and their total effects were significant (p < 0.01 and p < 0.05, respectively). The effect of home security on home automation was not mediated away by the reciprocal path through home safety.

Table 2:

Structural Equation Model Results

Home Automation Ownership (as Dependent Variable)
Predictors	Direct Effect			Indirect Effect			Total Effect
	ß	95% CI	P	ß	95% CI	P	ß	95% CI	P
Age	0.0045	−0.0190 – 0.0280	0.71	−0.005	−0.0111 – 0.0007	0.08	−0.0007	−0.0242 – 0.0228	0.95
Education<College	−0.2251	−0.4701 – 0.0199	0.07	−0.004	−0.0297 – 0.0213	0.75	−0.2293	−0.4743 – 0.0157	0.07
Income<$45,000	−0.2054	−0.4582 – 0.0474	0.11	−0.036	−0.0797 – 0.0065	0.10	−0.2420	−0.4948 – 0.0108	0.06
Daily Routines	−0.2302	−0.4301 – −0.0303	0.02	0.0614	−0.0229 – 0.1457	0.15	−0.1688	−0.3432 – 0.0056	0.06
Home Safety	−0.0334	−0.1902 – 0.1234	0.68	0.1345	0.0541 – 0.2149	<0.01	0.1011	−0.0361 – 0.2383	0.15
Home Security	0.2814	0.0854 – 0.4774	<0.01	−0.000	−0.0339 – 0.0327	0.97	0.2808	0.0828 – 0.4788	<0.01
ICT Usability	0.1575	0.0085 – 0.3065	0.04	No Path			0.1575	0.0085 – 0.3065	0.04
ICT Usability (as Dependent Variable)
Predictors	Direct Effect			Indirect Effect			Total Effect
	ß	95% CI	P	ß	95% CI	P	ß	95% CI	P
Age	−0.0330	−0.0487 – −0.0173	<0.001	No Path			−0.0330	−0.0487 – −0.0173	<0.001
Gender (Male)	−0.0503	−0.2149 – 0.1143	0.55	No Path			−0.0503	−0.2149 – 0.1143	0.55
Education<College	−0.0266	−0.1854 – 0.1322	0.74	No Path			−0.0266	−0.1854 – 0.1322	0.74
Income<$45,000	−0.2327	−0.3973 – −0.0681	<0.01	No Path			−0.2327	−0.3973 – −0.0681	<0.01
Home Safety	0.2491	0.1589 – 0.3393	<0.001	0.0107	−0.0363 – 0.0577	0.66	0.2598	0.1834 – 0.3362	<0.001
Home Security	0.0558	−0.0716 – 0.1832	0.39	−0.024	−0.0752 – 0.0268	0.35	0.0316	−0.1056 – 0.1688	0.65
Home Safety (as Dependent Variable)
Predictors	Direct Effect			Indirect Effect			Total Effect
	ß	95% CI	P	ß	95% CI	P	ß	95% CI	P
Physical Impairment	−0.8464	−1.0483 – −0.6445	<0.001	0.0259	−0.0368 – 0.0886	0.42	−0.8204	−1.0046 – −0.6362	<0.001
Cognitive Impairment	−0.1635	−0.4281 – 0.1011	0.23	0.0050	−0.0107 – 0.0207	0.53	−0.1585	−0.4133 – 0.0963	0.22
Sensory Impairment	−0.1333	−0.2940 – 0.0274	0.10	0.0041	−0.0077 – 0.0159	0.49	−0.1293	−0.2841 – 0.0255	0.10
Falls & Accidents	−0.4421	−0.6675 – −0.2167	<0.001	0.0135	−0.0198 – 0.0468	0.43	−0.4286	−0.6462 – −0.2110	<0.001
Home Security	−0.0932	−0.2951 – 0.1087	0.37	0.0029	−0.0108 – 0.0166	0.68	−0.0903	−0.2804 – 0.0998	0.35
Daily Routines	0.6071	0.4836 – 0.7306	<0.001	−0.018	−0.0637 – 0.0265	0.42	0.5885	0.4787 – 0.6983	<0.001
Home Security (as Dependent Variable)
Predictors	Direct Effect			Indirect Effect			Total Effect
	ß	95% CI	P	ß	95% CI	P	ß	95% CI	P
Social Support	0.1124	0.0438 – 0.1810	<0.01	−0.003	−0.0112 – 0.0044	0.40	0.1090	0.0404 – 0.1776	<0.01
Home Safety	0.3392	0.2451 – 0.4333	<0.001	−0.010	−0.0378 – 0.0170	0.46	0.3288	0.2543 – 0.4033	<0.001

Figure 2:

Structural Equation Model

*p<0.05 and **p<0.01

For hypothesis 1a (H1a), with ICT as a mediator dependent variable, age and income below $45,000 had a negative direct effect on ICT usability (p < 0.01) with no indirect pathways. Gender and education had no significant influence on ICT usability (and no indirect path) in the model. However, home safety with activities had a significant direct and total effect on ICT usability (p < 0.01). Home security had no significant effects on ICT usability.

H1b focused on home safety as a mediator leading to SH automation ownership. Having a physical impairment had a strong negative direct effect (ß = −0.85) and total effect (ß = −0.82) on home safety (p < 0.001). Similarly, incidents of falls and accidents showed a negative effect (direct and total) with home safety (p < 0.001). Self-reported independence with daily routines had a strong positive direct (ß = 0.6) and total effect (ß = 0.59) on home safety (p < 0.001). However, sensory and cognitive impairment, as well as sense of home security had no significant direct or indirect effect on home safety. For Hypothesis 1c (H1c) pertaining to home security as a mediator to home automation, social support and increased sense of home safety both had significant direct effect and total effect on home security (p < 0.01 and p < 0.001, respectively).

Discussion

The rapid growth in SH automation-based devices and IoTs has broadened the scope of technology-mediated aging-in-place. The market demand for the technology has been driven by their ease of integration into consumers’ homes and their capability to be interlinked with other devices on a network. While availability and affordability of the technology have also benefitted mainstream consumers, acceptance and adoption remain barriers in the aging population (Arthanat et al., 2018). As older adults are not inclined to adopting SH automation and ICTs naturally, implementation of the technology in the context of aging-in-place interventions requires a deeper understanding of their consumer profile. Recent studies have used the well-established technology acceptance model (Davis, 1989; Venkatesh et al., 2003) to highlight the interactions of inherent behavioral traits with adoption of technology (Macedo, 2017; Pal et al., 2018; Li et al., 2019). This study was unique in underscoring the pre-dispositional characteristics that may predict their affinity or aversion to SH automation technology.

One of the encouraging findings in the study is the high rate of ownership of ICT devices including smart phones and tablets. In sharp contrast, however, the ownership of SH automation technology was abysmally low with much of the sample owning only one out of the seven popular devices (as median ownership). Preference appeared to be toward more stand-alone devices such as thermostats, voice-activated assistants and home security systems as opposed to those that needed wireless integration to work (such as remote home monitoring and smart phone control). Findings from the main hypothesis indicate that older adults may be motivated to adopt SH automation devices through their ongoing exposure to ICT and concern for home security. On the other hand, through an inverse relationship one can conclude that decreasing independence in daily routines will also predispose adoption of SH automation technology.

Contrary to the researchers’ expectation, demographics played no significant direct role in SH automation ownership other than education and income to some extent (β = −0.22, p = 0.07 and β = −0.20, p = 0.1, respectively). A conceivable explanation may be that the influence of demographics on SH automation may be mediated through ICT, which was shown to be the case with the H1a. Herein, increasing age and below college education had a significant negative effect (p < 0.01) on ICT usability in the sample. The finding coincides with that of several studies that show the link between demographics and ICT exposure (Friemel, 2016; Niehaves and Plattfaut, 2014; Pew Research Center, 2017; Vroman et al., 2015).

A major take-away from the study was the mediating effect of home safety on ICT usability. From the causal pathway, the researchers deduced that the incidents of falls and accidents, which was inversely related to home safety (β = −0.44, p < 0.01), influences one’s decision to adopt the technology. More importantly, from the standpoint of health and functioning, the presence of a physical impairment undermined home safety, while independence in daily routines had a positive association with sense of home safety. Therefore, decreasing health and functioning through the mediating effect of safety may influence older adults to rely on ICT, and consequently, SH technology. A similar pattern was evident in the case of home security and social network. The data revealed that the presence or absence of family or friends in the neighborhood significantly influenced sense of home security, which, in turn, had an effect on SH automation ownership.

Implications

The study contributes to the growing body of knowledge and interest in ICT and SH technology adoption among aging consumers. The results clearly demonstrate the theory of technology acceptance in practice through actual SH ownership. Perceived usefulness with SH technology as related to health and functioning, and above all, home safety and security was a key determinant with its ownership. The findings have implications for multiple stakeholders for facilitation of technology-mediated aging-in-place. It is important for home health and home modification professionals to be mindful of the interactions among the demographic and pre-dispositional factors and to customize (simplify or intensify) technology integration and training. It is crucial for technology developers, manufacturers and industry experts to not only realize the inherent usefulness that older consumers see with SH technology but also to integrate their input in product design to overcome prevailing usability barriers. Considering the above demographic attributes and perceived benefits, policymakers must incorporate SH technologies in community programs that support aging-in-place and telehealth.

Limitations

There are limitations to acknowledge in this study. The survey sample was by convenience and sampling bias cannot be ruled out due to interest or dislike for SH technology. The sample was from the New England region of the USA with unique demographic and consumer profiles. Therefore, the findings may not generalize in entirety to older adults in different parts of the world. The study used data on the most popular devices to compute scores on ICT usability and home automation. The researchers did not tease out causal pathways for specific types of devices such as those for safety, health monitoring and home comfort.

Future research

Further analysis of the survey data could determine affinity of older adults toward specific type of SH automation devices and corresponding determinants. A subsequent research could combine the effects of demographic characteristics and behavioral traits defined in the technology acceptance model (Venkatesh et al., 2003) to derive a more comprehensive representation of technology adoption. It would also be worthwhile to objectively verify the causal links of technology adoption through the predictors of knowledge and digital literacy of older adults (Yu et al., 2017) and their effects on actual usage frequency of the devices. ICT researchers have also suggested more longitudinal and experimental studies to derive more reliable conclusions on the direction of the associations between consumer characteristics and technology adoption (Elliot et al., 2013).

Conclusion

This study examined the key determinants of demographics, health and functioning, home safety and social network, as well as their causal pathways to ICT and SH automation technology ownership among older adults. While ICT adoption was relatively higher, ownership of SH technology is still lagging in this population. As a positive indication, increased exposure to ICT will directly influence ownership of SH technology, and as a result both may be projected to increase proportionally. Demographics (including income and education) had moderately lesser role to play with SH automation technology. Instead, concerns for home safety and home security mediated through falls and accidents, physical impairments, independence and social support seem to facilitate perceived usefulness and adoption of the technology.