Abstract
JOURNAL OF ASIAN ARCHITECTURE AND BUILDING ENGINEERING https://doi.org/10.1080/13467581.2022.2047057 URBAN PLANNING AND DESIGN Purchase decision process and information acquisition of zero-energy houses in Japan a b Hitomu Kotani and Kazuyoshi Nakano a b Department of Urban Management, Graduate School of Engineering, Kyoto University, Kyoto, Japan; Socio-Economic Research Center, Central Research Institute of Electric Power Industry, Chiyoda, Tokyo, Japan ABSTRACT ARTICLE HISTORY Received 14 July 2021 Compared with traditional houses, zero-energy houses (ZEHs) offer efficient and preferable Accepted 21 February 2022 living environments, e.g., reduced greenhouse gas emissions and lower health risks. Currently in Japan, such houses are not as popular as anticipated and sales do not meet the national KEYWORDS government target. Accordingly, household buying process should be investigated to develop Zero-energy house (ZEH); policies to encourage the spreading of ZEHs. Therefore, we investigated which factors influ - technology adoption; enced purchasers’ intentions and behaviors. We based our purchase process modeling on the received information; unified theory of acceptance and use of technology, which includes six constructs, i.e., use Bayesian structural equation modeling (Bayesian SEM); behavior, behavioral intention, performance expectancy, effort expectancy, social influence, unified theory of acceptance and facilitating conditions. Our model also considered the effects of the information content and use of technology buyers obtained and the channels they used on performance expectancy. In our estimation, we (UTAUT) used Bayesian structural equation modeling and response from 297 Japanese households. It was found that certain information content and channel combinations, e.g., health aspect information obtained from salespersons effectively enhanced performance expectancy. Although performance expectancy did not significantly facilitate the use intention, social influence and facilitating conditions effectively promoted intention leading to purchase. Our findings contribute to more appropriate information provision strategies and supporting policies to promote the spread of these houses. 1. Introduction of the envelope, and installing highly energy- efficient household equipment, all of which bring After the adoption of the Paris Agreement in 2015 about superior indoor environmental quality. More (Rogelj et al. 2016; United Nations 2015), it has been specifically, a ZEH is defined as a house satisfying expected of global society, including national govern- the following three criteria (Ministry of Economy, ments and all other stakeholders, to institute further Trade and Industry 2015; Oki et al. 2019): global warming mitigation measures. Global energy- related CO emissions amounted to approximately 32 (1) It meets the criteria for strengthened outer walls gigaton in 2016, which are predicted to increase to 36 and U value (the amount of heat that escapes gigaton in 2040 (International Energy Agency 2018). from the inside of the house to the outside Constructing zero-energy houses (ZEHs) is one of the through the floors, outer walls, roof, and win- measures to promote decarbonization. These residential dows, divided by the area of the outer skin); buildings greatly reduce energy needs through effi - (2) It reduces the primary energy consumption, ciency gains, such that the balance of energy needs excluding renewable energy, by 20% or more can be supplied with renewable technologies from the standard primary energy consumption (Stefanović, Bojić, and Gordić 2014). The adoption of (energy consumption for heating, cooling, ven- such environment-friendly living conditions is expected tilation, hot water supply, and lighting); to spread worldwide (Farhar and Coburn 2008). (3) It reduces primary energy consumption, includ- ing renewable energy, by 100% or more from the standard primary energy consumption. 1.1. Zero-energy houses in Japan In Japan, a ZEH is defined broadly as a house of The necessary equipment to meet such criteria includes which the annual energy consumption balance is heat-insulation walls, photovoltaic (PV) panels, and light- zero. This is achieved by installing renewable emitting diode-based (LED) lamps, with the option to energy-generation devices, improving the insulation include high-efficiency air conditioners, floor heating, CONTACT Hitomu Kotani kotani.hitomu.5c@kyoto-u.ac.jp Department of Urban Management, Graduate School of Engineering, Kyoto University, Kyoto, Japan Supplemental data for this article can be accessed here © 2022 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group on behalf of the Architectural Institute of Japan, Architectural Institute of Korea and Architectural Society of China. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 2 H. KOTANI AND K. NAKANO energy-efficient water heaters, home energy- ZEHs by 2020 and all newly constructed houses management systems (HEMS), vehicle-to-home (V2H) would be ZEHs by 2030 (Ministry of Economy, communication, and the like. Trade and Industry 2018b). However, on average, ZEHs represented only 10% of detached houses Japanese society also needs to deploy ZEHs from built in Japan in 2018 (Sustainable open both a societal and an individual viewpoint. From Innovation Initiative 2018). Accordingly, in an effort a social perspective, the household sector needs to to promote the sale of ZEHs to reach the govern- urgently reduce its energy consumption, which has ment goal, the process and decision of households been increasing substantially and at a more rapid to purchase such houses needed to be researched rate than that of other sectors. In this regard, an (See Appendix A for additional details of ZEHs in advantage of ZEHs is their high thermal-insulation Japan). performance that realizes energy savings for the household sector. ZEHs also potentially contribute to the balanced energy mix of society as they 1.2. Literature review increase the proportion of renewable energy. From an individual perspective, ZEHs can reduce Although several previous studies have focused on energy costs. After the Fukushima Daiichi nuclear the supplier aspect of ZEHs, such as construction disaster induced by the 2011 Tohoku earthquake companies and designers (Attia et al., 2013; Farhar and tsunami, the electricity price in Japan has fluc - and Coburn 2008; Persson and Grönkvist 2015; Shi tuated (Ministry of Economy, Trade and Industry et al. 2020; Zhao, Pan, and Chen 2018), research on 2018a) owing to nuclear power plants not being the consumer aspect is scarce. Analyses have operational and the increasing cost of generating already been conducted on consumer choices rele- thermal power because of the soaring cost of fuel. vant to various products commonly installed in Therefore, both the energy saving and generation ZEHs, such as energy-efficient water heaters (Goto, realized by ZEHs could bring about economic ben- Goto, and Sueyoshi 2011; Ma, Yu, and Urban 2018; efits to individual households. In addition, ZEHs Michelsen and Madlener 2012; Ofuji and Nishio provide several non-energy benefits. First, they 2013), PV panels (Bollinger and Gillingham 2012; function effectively during emergencies (e.g., nat- Graziano and Gillingham 2014; Noll, Dawes, and ural disasters) because of electricity self- Rai 2014; Yamaguchi et al. 2010), LED lamps consumption (Miller 2015; Young Jr, 2009). A case (Khorasanizadeh et al. 2016), and HEMS (Park et al. in point is the blackout after the occurrence of the 2017). The detailed purchase process of such 2018 Hokkaido Eastern Iburi earthquake that devices has also been analyzed (Khorasanizadeh affected three million households. All power outage et al. 2016; Park et al. 2017), namely, how house- was resolved two days after the earthquake; how- holds perceive the performance and ease of using ever, 85% of households having PV panels, which the products, how such perception affected the are necessary for ZEHs, used their self-generated intention to purchase, and how the intention pro- electricity effectively (Japan Photovoltaic Energy moted the actual purchasing behavior. However, Association (JPEA) 2018). Furthermore, some house- the purchase process of ZEHs, not energy-saving holds making use of rechargeable batteries were or energy-generating products, has not been unaffected by the outage and were even able to investigated. provide electricity to their neighbors (Monthly Information generally plays a significant role in Smart House 2018a). Second, they likely provide the process of adopting new technologies, as, dur- health benefits to residents (Monthly Smart House ing the initial stages, prospective buyers become 2018b; Oki et al. 2019). The high heat-insulation familiar with the products and their functions performance reduces the temperature gap among (Rogers 2003). In studies conducted on energy- rooms, including toilets and bathrooms and, there- related technologies, the researchers examined to fore, the risk of heat shock (i.e., a sudden change in what extent the views of neighbors or other peo- blood pressure owing to rapid temperature change, ple influenced the choice of consumers of energy- which can cause a stroke or myocardial infarction) is efficient water heaters (Goto, Goto, and Sueyoshi potentially reduced (Monthly Smart House 2018b; 2011) and PV panels (Bollinger and Gillingham Oki et al. 2019). 2012; Graziano and Gillingham 2014; Noll, Dawes, However, a gap exists between the current pre- and Rai 2014). This implies that the research focus valence of ZEHs and the target set by the national was mainly on the effect of the channels of infor- government. The goal is that the majority of mation on the adoption. For instance, recommen- detached houses built by contractors would be dations from home builders and/or sales In 2016, for instance, the final energy consumption (crude oil equivalent) of the household sector was 1.9 times that in 1973 (Ministry of Economy, Trade and Industry 2018a). JOURNAL OF ASIAN ARCHITECTURE AND BUILDING ENGINEERING 3 representatives of appliances (Goto, Goto, and 2. Conceptual model Sueyoshi 2011), as well as suggestions from neigh- 2.1. Unified theory of acceptance and use of bors (Bollinger and Gillingham 2012; Graziano and technology Gillingham 2014), were found to affect the buyers’ choices. However, the element lacking here is that We used the unified theory of acceptance and use of the role of channels depends on the content of technology (UTAUT) (Curtale, Liao, and van der Waerden the information (Kotani and Honda 2019). In other 2021; Hartwich et al. 2019; Rajapakse 2011; Sovacool words, how the recipients interpret the informa- 2017; Venkatesh et al. 2003) to model the household tion can depend on the type of information con- purchase process of ZEHs. UTAUT, which helps to under- tent they obtain and from whom it is obtained. As stand the drivers of acceptance by users of new tech- regards ZEHs, in particular, the information would nologies, has been developed by integrating the have to be associated with the various types of elements of eight prominent models, including (1) the advantages (e.g., economic, environmental, and theory of rational action (TRA), (2) technology accep- health benefits (Monthly Smart House 2018b; Oki tance model (TAM/TAM2), (3) motivational model et al. 2019), disaster risk reduction (Miller 2015; (MM), (4) theory of planned behavior (TPB), (5) model Young Jr, 2009)), and the different energy-saving agreement between TAM and TPB (combined TAM-TPB) and energy-generating equipment. Therefore, in , (6) model of personal computer utilization (MPCU), (7) analyzing the effect of information on promoting innovation diffusion theory (IDT), and (8) social cognitive the sale of ZEHs, combining various information theory (SCT). The UTAUT is believed to be more robust in contents and channels should be considered. evaluating and predicting technology acceptance than the other technology acceptance models (Taiwo and Downe 2013). Whereas UTAUT was developed originally to model the adoption process of information technol- 1.3. Objectives ogy, it has been applied to this process in a wide variety The aim of this study was to model the household of technologies, including energy-saving and energy- purchase process of ZEHs and investigate how the creating technologies (e.g., LED lamps (Khorasanizadeh intention to purchase these houses as well as the et al. 2016) and renewable energy sources (Rezaei and actual purchasing behavior were promoted, con- Ghofranfarid 2018)). sidering the effects of various combinations of The UTAUT model considers “use behavior,” which information content and channels. We targeted represents the acceptance and use of technology, and households that purchased ZEHs in Japan and we “behavioral intention,” which represents the degree of conducted statistical analysis based on Bayesian intention to adopt the technology. Behavioral inten- structural equation modeling and the dataset con- tion was initially described in the TRA, which claims taining the response received from 297 house- that the effectiveness of a certain behavior is holds. Our analysis revealed the effective a consequence of whether or not the individual information contents according to the information intends to perform the behavior. The UTAUT uses channels and the factors that could positively four core determinants of use behavior and behavioral influence the intention and actions to purchase intention, namely, (1) performance expectancy, (2) ZEHs. Our findings could aid in assessing and effort expectancy, (3) social influence, and (4) facilitat- improving the current information provision stra- ing conditions. Details on these four determinants fol- tegies and supporting policies aimed at promoting low (Rajapakse 2011; Sovacool 2017; Venkatesh et al. the spread of ZEHs. Since concepts similar to 2003). Japanese ZEHs are found in other countries, our findings could also be relevant to such countries. (1) Performance expectancy: performance expectancy The remainder of this paper is organized as follows: is defined as the degree to which the user expects Section 2 describes the framework of our analysis; that using the system will help him or her attain Section 3 shows the statistical method and data used gains in performance. This construct finds its roots in the analysis; Section 4 reports the estimation results; in perceived usefulness from TAM/TAM2 and com- Section 5 discusses the results and limitations; and bined TAM-TPB, extrinsic motivation from MM, Section 6 concludes the study, describing policy relative advantage from IDT, and outcome expec- implications. tation from SCT. More broadly, performance For example, a zero net energy (ZNE) building is defined in California, USA, as “an energy-efficient building where, on a source energy basis, the actual annual consumed energy is less than or equal to the on-site renewable generated energy.” California has set targets for the construction of ZNE buildings, including that by 2020 all new residential housing should be ZNE (California Public Utilities Commission n.d.). However, the construction cost of ZNEs could be a limiting factor. On the other hand, as demonstrated by Petersen et al. (Petersen, Michael, and Corvidae 2019), it is possible to build ZNEs at a reasonable incremental cost. Such positive information on construction cost could be an incentive for households to choose ZNEs. However, no literature is available that considers the decision-making process and the effects of acquiring information. 4 H. KOTANI AND K. NAKANO expectancy has come to mean the degree to which (2003) and Khorasanizadeh et al. (2016). More specifi - users expect that technology will be beneficial in cally, we assumed that the behavioral intention to performing particular tasks. purchase ZEHs was facilitated by performance expec- (2) Effort expectancy: effort expectancy is the tancy, effort expectancy, social influence, and facilitat- degree of ease associated with consumers’ use ing conditions, and the use behavior was promoted by of the technology. This construct finds its roots the behavioral intention and facilitating conditions. in concepts such as perceived ease of use from TAM/TAM2, complexity from MPCU, and ease of use from IDT. 2.2. Information acquisition (3) Social influence: social influence refers to the We extended the UTAUT model, adding the effect of degree to which an individual perceives that acquiring information about ZEHs (rectangle in important others (e.g., family and friends) Figure 1). Although each construct may be affected believe that he or she should use the new sys- to a greater or lesser degree by the content of informa- tem. It finds its roots in concepts such as sub- tion obtained and from whom it was obtained, we jective norms (i.e., the individual’s perception focused on their effect on the performance expec- that most people who are important to him/ tancy. The reason was that the content of information her think he/she should use the system) from that we targeted was mainly related to the benefits of TRA, TAM2, TPB, and C-TAM-TPB, social factors (i. ZEHs, which will be shown below, and we avoided e., the individual’s internalization of the refer- complicating the model. ence group’s culture) from MPCU, and image (i. We considered the following four content types, e., the degree to which the use of an innovation including the multiple benefits of ZEHs: is perceived to enhance one’s social image or status) from IDT. (1) Economic information: information on eco- (4) Facilitating conditions: facilitating conditions nomic issues, such as the price of energy- are defined as the degree to which an individual saving and energy-generating equipment of believes that an organizational and technical ZEHs and the utility cost reduction realized by infrastructure exists to perform a task or adopt ZEHs; a new system. This construct embodies per- (2) Technical information: information on the func- ceived behavioral control from TPB, facilitating tions and mechanism of energy-saving and conditions from MPCU, and compatibility from energy-generating equipment installed in ZEHs; IDT. (3) Environmental information: information on As regards the relationships between the above six environmental issues, such as greenhouse gas constructs, namely, (1) use behavior, (2) behavioral reduction and risk reduction during emergen- intention, (3) performance expectancy, (4) effort expec- cies (e.g., disasters); tancy, (5) social influence, and (6) facilitating condi- (4) Health information: information on health issues, tions, we hypothesized the connections between six such as the positive effect of the high heat- ovals, as shown in Figure 1, based on Venkatesh et al. insulation performance on the residents’ health. Figure 1. Proposed research model. JOURNAL OF ASIAN ARCHITECTURE AND BUILDING ENGINEERING 5 As regards the channels of information, the follow- the information contents and the channels and (2) how ing four types were considered: the behavioral intention and actual behavior to pur- chase ZEHs were facilitated by the constructs of UTAUT. (1) Salespersons: channels such as face-to-face communication with the salespersons of home 3. Estimation method and data builders; (2) Friends: channels such as word-of-mouth com- 3.1. Bayesian structural equation modeling munication with friends, colleagues, and family To estimate the structure mentioned in the previous members; section, we employed structural equation modeling (3) Advertisement: channels such as advertise- (SEM), which is statistical modeling technique that ments provided by home builders and the can estimate the relationships between variables, government; including not only observed but also latent variables. (4) Strangers: channels such as online word-of- Each of the six constructs of UTAUT in our structure (i. mouth and reviews posted by people other e., the ovals in Figure 1, namely, (1) use behavior, (2) than friends. behavioral intention, (3) performance expectancy, (4) effort expectancy, (5) social influence, and (6) facilitat- Rogers (2003) categorizes communication channels ing conditions) were considered difficult to be repre- as (i) interpersonal versus mass media and (ii) localite sented by a single variable (Khorasanizadeh et al. 2016) versus cosmopolite. Interpersonal channels involve and, therefore, we treated these six constructs as latent the face-to-face exchange between two or more indi- variables. We denoted each latent variable of house- viduals. Mass media channels transmit messages hold i 2 f1; . . . ; ng as ω , ω , ω , ω , ω , and B;i I;i PE;i EE;i SI;i through media such as the radio, television, and ω , respectively, and we denoted the vectors of FC;i newspapers, which facilitate a source reaching an measurement variables for each latent variable as z , B;i audience of millions. Cosmopolite communication z , z , z , z , and z , respectively. SEM is gener- I;i PE;i EE;i SI;i FC;i channels are those linking an individual to sources ally composed of (1) measurement models that exam- outside the social system under study. Interpersonal ine the relationships between observed variables and channels can be either local or cosmopolite, while latent variables and (2) a structural model that analyzes mass media channels are almost entirely cosmopolite. the interrelationships among latent variables. The salespersons considered in this study are inter- preted as interpersonal and cosmopolite channels; 3.1.1. Measurement models friends are interpersonal and localite; advertisement In our measurement models, as the observed variable is considered mass media channels. The online word- vector z was measured in the yes/no format and z , B;i I;i of-mouth and review by strangers, which are categor- z , z , z , and z were measured on a five-point PE;i EE;i SI;i FC;i ized neither into mass media nor as interpersonal, Likert scale using the survey as shown in have recently been found to be effective (Chevalier Subsection 3.2, we incorporated the probit link func- and Mayzlin 2006; Duan, Gu, and Whinston 2008; Ye tions suitable to model discrete outcome variables. Let et al. 2011) and, therefore, we also considered the channels of strangers. ω express the latent variable vectors as ω ¼ i i ω ; ω ; ω ; ω ; ω ; ω (the superscript T B;i I;i PE;i EE;i SI;i FC;i means transpose of vector). 2.3. Proposed model As regards observed variable vectors z , z , z , B;i I;i PE;i Overall, we considered the structure shown in Figure 1. z , z , and z for each latent variable, z ¼ EE;i SI;i FC;i B;i That is, the study assumed, as presented in T z ; . . . ; z ; . . . ; z comprises n observed vari- B;i;1 B;i;j B;i;n B Subsection 2.2, that the combination of the four infor- ables; z , z , z , z , and z comprise n , n , n , I;i PE;i EE;i SI;i FC;i I PE EE mation contents and the four channels (i.e., 16 combi- n , and n observed variables, respectively. SI FC nations) affected the performance expectancy of ZEHs. The observed variable vectors are determined by y , B;i Moreover, as illustrated in Subsection 2.1, not only per- y , y , y , y , and y , which comprise unobserved I;i PE;i EE;i SI;i FC;i formance expectancy but also effort expectancy, social continuous variables. y ¼ y ; . . . ; y ; . . . ; y ; influence, and facilitating conditions were assumed to B;i;1 B;i;j B;i;n B;i B facilitate behavioral intention to purchase ZEHs. It was y , y , y , y , and y comprise n , n , n , n , and I PE EE SI I;i PE;i EE;i SI;i FC;i also assumed that behavioral intention and facilitating n variables, respectively. FC conditions promoted use behavior. Based on the pro- More specifically, ∀j 2 f1; . . . ; n g, z is a dummy B B;i;j posed structure, we examined (1) how the performance variable having either 0 or 1 and determined by y , as B;i;j expectancy of ZEHs was affected by the combination of follows: The subscripts B, I, PE, EE, SI, and FC express use behavior, behavioral intention, performance expectancy, effort expectancy, social influence, and facilitating conditions, respectively. 6 H. KOTANI AND K. NAKANO z ¼ 0 if y � 0; (1) F f (Lee, Song, and Cai 2010; Lee and Song B;i;j B;i;j k;j;4 2012). Fð�Þ is the cumulative distribution function of z ¼ 1 otherwise: (2) B;i;j standardized normal distribution Nð0; 1Þ, f is the k;j;1 frequency of the first category (i.e., Furthermore, ∀j 2 1; . . . ; n , an element of z , is f g, z � � I I;i;j I;i P f ¼ I z ¼ 1 =n ), and f is the cumulative k;j;1 k;i;j k;j;4 an ordered categorical variable with a five-point Likert scale from 1 to 5 and defined by y , as follows: I;i;j frequency of the categories that are less than five (i. P � � e., f ¼ I z < 5 =n ), where the function z ¼ 1 if y � α ; (3) k;j;4 k;i;j I;i;j I;i;j I;j;1 IðAÞ ¼ 1 if A is true, 0 otherwise. In our analysis, z ¼ 2 if α < y � α ; (4) I;i;j I;j;1 I;i;j I;j;2 we focused on the coefficients of paths from latent variables to observed ones (factor loadings)—Λ , Λ , B I Λ , Λ , Λ , and Λ . PE EE SI FC 3.1.2. Structural model z ¼ 5 if α < y ; (5) I;i;j I;j;4 I;i;j In our structural model, we considered not only the inter- where α , . . . , α are unknown threshold parameters. I;j;1 I;j;4 relationships among latent variables (i.e., ω , ω , ω , B;i I;i PE;i As elements of z , z , z , and z are also measured PE;i EE;i SI;i FC;i ω , ω , and ω ) but also the effects of information EE;i SI;i FC;i on a five-point Likert scale, they are defined in the acquisition on the latent variable, i.e., performance expec- same way. tancy. In other words, we examined how d , denoted as The measurement models associate (1) y , y , y , B;i I;i PE;i the vector regarding information acquisition for house- y , y , and y with (2) latent variable vector ω . For EE;i SI;i FC;i hold i, affected ω . d is composed of vectors of d , PE;i i Econ;i use behavior, the model is defined as follows: d , d , and d , which relate to economic, tech- Tech;i Envi;i Health;i nical, environmental, and health information acquisition, y ¼ μ þ Λ ω þ ε ; B B;i B;i (6) B;i B � � T T T T respectively, and d ¼ d ; d ; d ; d . Econ;i Tech;i Envi;i Health;i where a vector of intercepts Whether household i obtained economic information μ ¼ μ ; . . . ; μ ; . . . ; μ , an unknown vector of B B;1 B;j B;n � from (1) salespersons, (2) friends, (3) advertisements, and factor loadings Λ ¼ λ ; . . . ; λ ; . . . ; λ , and B B;1 B;j B;n (4) strangers are expressed as dummy variables d , Econ;Sales;i a random vector of measurement ε ¼ ε ; . . . ; ε ; B;i B;i;1 B;i;j d , d , and d , respectively, Econ;Friends;i Econ;Ad;i Econ;Strangers;i . . . ; ε Þ . ∀j; ε is independent and identically dis- B;i;n B;i;j B where d ¼ d ; d ; d ; Econ;i Econ;Sales;i Econ;Friends;i Econ;Ad;i tributed (i.i.d.) according to normal distribution with T d Þ : Similarly, d ¼ d ; d ; Econ;Strangers;i Tech;i Tech;Sales;i Tech;Friends;i a mean of 0 and variance of 1, i.e., Nð0; 1Þ. Similarly, y , I;i T d ; d Þ , d ¼ d ; d ; Tech;Ad;i Tech;Strangers;i Envi;i Envi;Sales;i Envi;Friends;i y , y , y , and y are given by PE;i EE;i SI;i FC;i d ; d Þ , and d ¼ ðd ; Envi;Ad;i Envi;Strangers;i Health;i Health;Sales;i y ¼ μ þ Λ ω þ ε ; (7) T I I;i I;i I;i I d ; d ; d Þ . Health;Friends;i Health;Ad;i Health;Stranders;i To formulate the model, the latent variable vector ω y ¼ μ þ Λ ω þ ε ; (8) PE PE;i PE;i PE;i PE is divided into η and �. η is an outcome latent variable i i i vector affected by other latent variables and/or d ; � is y ¼ μ þ Λ ω þ ε ; (9) EE EE EE;i EE;i EE;i an explanatory latent variable vector. In this study, � � T T η ¼ ω ; ω ; ω , � ¼ ω ; ω ; ω , and B;i I;i PE;i EE;i SI;i FC;i i i y ¼ μ þ Λ ω þ ε ; (10) � SI SI;i SI;i SI;i SI T ω ¼ η ; � . The interrelationships among latent vari- i i ables are formulated by the following structural model: y ¼ μ þ Λ ω þ ε ; FC FC;i FC;i (11) FC;i FC η ¼ Cd þ �η þ Γ� þ δ ; (12) i i i i i where the random variables ε , ε , ε , ε , and I;i;j PE;i;j EE;i;j SI;i;j where C, �, and Γ are matrices of unknown coeffi - ε are i.i.d. according to normal distributions 0 1 FC;i;j � � � � 0 ��� 0 N 0; ψ , N 0; ψ , N 0; ψ , N 0; ψ , and εI;j εPE;j εEE;j εSI;j @ A 0 ��� 0 cients. More specifically, C ¼ , N 0; ψ , respectively. The vectors of error terms ε , B;i c ��� c εFC;j 3;1 3;16 0 1 0 1 ε , ε , ε , ε , and ε do not correlate with each I;i PE;i EE;i SI;i FC;i 0 π 0 0 0 γ @ A @ A other, and each vector of error terms is independent of ω . i � ¼ 0 0 π , and Γ ¼ γ γ γ . δ 23 i 21 22 23 For model identification, some conditions are 0 0 0 0 0 0 imposed on the model. One of the factor loadings is a random variable vector, expressed by for each latent variable is fixed as 1: λ ¼ λ ¼ B;1 I;1 δ ¼ δ ; δ ; δ . The errors δ , δ , and δ are i. i B;i I;i PE;i B;i I;i PE;i λ ¼ λ ¼ λ ¼ λ ¼ 1 (Lee and Song 2012). PE;1 EE;1 SI;1 FC;1 i.d. according to Nð0; ψ Þ, Nð0; ψ Þ, and Nð0; ψ Þ, δB δI δPE Threshold parameters are fixed as ∀k 2 respectively. δ , δ , and δ do not correlate with � B;i I;i PE;i fI; PE; EE; SI; FCg; α ¼ F f and α ¼ k;j;1 k;j;1 k;j;4 each other. The vectors of explanatory latent variables JOURNAL OF ASIAN ARCHITECTURE AND BUILDING ENGINEERING 7 � are i.i.d. according to Nð0; ΦÞ, where Φ is a general ∀k; jð� 1Þ; λ , Nð1; 1Þ; (14) 0 1 k;j ϕ ϕ ϕ 11 12 13 @ A covariance matrix ϕ ϕ ϕ . δ is indepen- 21 22 23 ∀j; ψ , Gammað0:1; 0:1Þ; (15) ϕ ϕ ϕ εI;j 31 32 33 dent of � and ε . In our analysis, we focused on the coefficients of (1) paths from d to ω and (2) paths i PE;i ψ , Gammað0:1; 0:1Þ; (16) εPE;j among latent variables: C, �, and Γ . ψ , Gammað0:1; 0:1Þ; (17) εEE;j 3.1.3. Bayesian estimation We estimated path coefficients, i.e., parameters, based on ψ , Gammað0:1; 0:1Þ; (18) εSI;j the household data of the observed variable vectors, z , B;i z , z , z , z , and z , and the variable vector on I;i PE;i SI;i EE;i FC;i ψ , Gammað0:1; 0:1Þ; (19) information acquisition, d . Whereas many studies follow εFC;j a classical frequentist approach, namely, the maximum likelihood method (Hox 1998; Kotani, Honda, and Sugitani ∀kð� BÞ; j; α , Uniform α ; α ; (20) k;j;2 k;j;1 k;j;4 2019) for the estimation, we employed a Bayesian approach (Assaf, Tsionas, and Oh 2018; Lee and Song � α , Uniform α ; α ; (21) k;j;3 k;j;1 k;j;4 2012; Levy and Mislevy 2017; Lu et al. 2020; Merkle and Wang 2018; Song and Lee 2012). To the best of our where Gammaða ; b Þ is a gamma distribution with 1 1 knowledge, this study is the first to apply Bayesian SEM shape a and rate b , and Uniformða ; b Þ is 1 1 2 2 to the model based on the UTAUT in building and energy a continuous uniform distribution on the inter- research. The Bayesian estimation method treats para- val ½a ; b �. 2 2 meters as random variables. Drawing on the Bayes’ theo- Next, the prior distributions of the parameters in the rem, the prior probability distribution of unknown structural model, C, �, Γ , ψ , ψ , ψ , and Φ, are δB δI δPE parameters, i.e., prior distribution, is updated, given the specified as follows: data obtained, to posterior distribution (Gelman et al. ∀l; c , Nð0; 10Þ; (22) 3;l 2013; Lee and Wagenmakers 2014; Levy and Mislevy 2017). That is, pðθjDÞ / pðDjθÞpðθÞ, where θ is an unknown parameter vector, D is data, pðθÞ is a prior π , Nð0; 10Þ; (23) distribution of the parameters, pðDjθÞ is a likelihood, and pðθjDÞ is a posterior distribution. In most instances, π , Nð0; 10Þ; (24) obtaining the posterior distribution is done by simulation, using the so-called Markov chain Monte Carlo (MCMC) γ , Nð0; 10Þ; (25) methods. The posterior distribution simulated via MCMC expresses the uncertainty of the parameters. The sam- γ , Nð0; 10Þ; (26) pling-based Bayesian methods depend less on asympto- tic theory and, therefore, have the potential to produce γ , Nð0; 10Þ; (27) results that are more reliable, even with small samples, compared with those obtained by the maximum likeli- hood method (de Schoot et al. 2017, 2014; Lee and Song γ , Nð0; 10Þ; (28) 2012, 2004). Generally, in the initial stages of the accep- tance of technology, as was our target, it is difficult to ψ , Gammað0:1; 0:1Þ; (29) δB obtain large samples and the Bayesian method is there- fore considered suitable. Furthermore, the Bayesian ψ , Gammað0:1; 0:1Þ; (30) δI method is more flexible with complex datasets, as it treats the raw data of dummy and five-point Likert scale vari- ψ , Gammað0:1; 0:1Þ; (31) ables more easily without approximating them to contin- δPE uous variables (Lee, Song, and Cai 2010; Lee and Song 2012). 1 (32) Φ , W ðR ; ρ Þ; 3 0 0 0 1 3.1.4. Prior distribution 1 0:5 0:5 This study used non- and weakly informative priors. First, @ A R ¼ 0:5 1 0:5 ; ρ ¼ 4; (33) the prior distributions of the parameters in the measure- 0:5 0:5 1 ment model, namely, ∀k 2 fB; I; PE; EE; SI; FCg, μ , Λ , ψ , ψ , ψ , ψ , and ψ , are defined as follows: where W ð�Þ is a three-dimensional Wishart distribu- εI;j εPE;j εEE;j εSI;j εFC;j tion. ∀k; j; μ , Nð0; 10Þ; (13) k;j 8 H. KOTANI AND K. NAKANO social influence, we asked about the opinions of 3.2. Data collection family members, acquaintances, and neighbors on To collect data, we conducted a questionnaire ZEHs, and the impression of respondents of these survey, as shown in Table 1. We focused on houses, before the respondents started considering households that purchased custom-built detached such purchase. As regards the facilitating condi- ZEHs and received the subsidies related to pur- tions, we asked about the environmental surround- chasing ZEHs. The survey was answered by one of ings and personal circumstances of the respondent the family members in each household who pri- before the purchase. marily negotiated the designs and prices with For the elements of vectors d , d , d , and Econ;i Tech;i Envi;i home builders or who understood the purchase d regarding information acquisition, the ques- Health;i process well. Through the survey, we collected n ¼ tions, answer format, and the result of answers are 297 sample data. The sample characteristics are presented in Table 4. presented in Appendix B. For the observed variable vectors z , z , z , B;i I;i PE;i z , z , and z , we asked n ¼ 7, n ¼ 2, n ¼ 7, SI;i EE;i FC;i B I PE 4. Estimation results n ¼ 4, n ¼ 3, and n ¼ 8 questions, respectively. EE SI FC The questions, answer format, and results of the The posterior distribution for the unknown para- answers are presented in Table 2 and Table 3. The meters in the Bayesian SEM demonstrated in questions were produced based on previous stu- Subsection 3.1 was obtained via MCMC simulation, dies, e.g., Venkatesh et al. (2003) and given the data introduced in Subsection 3.2. The Khorasanizadeh et al. (2016). Bayesian results were sampled for 150,000 iterations For the use behavior (Table 2), we asked whether following burn-in 5000 iterations for each of three the houses had energy-saving and energy-generating chains by the Just Another Gibbs Sampler (JAGS) equipment (e.g., high-efficiency air conditioners, floor program (Plummer and others, 2003) using R2jags heating, energy-efficient water heater, and the like; (Su and Yajima 2015). Every fifth iteration was saved however, we excluded the equipment necessary for for each chain. That is, this study drew 87,000 (= ZEHs such as heat-insulation walls, PV panels, and (150,000–5000) � 3 � 5) samples for each para- LED lamps). The answers were evaluated as yes or no meter, based on which estimation results were and transformed to dummy variables taking either 1 or obtained. Model convergence was assessed via the 0. Therefore, the use behavior in this study means the Gelman-Rubin statistic (Gelman et al., 1992). All degree of installation of additional energy-saving and - parameters achieved statistical values of less creating equipment that improves the performance of than 1.1. ZEHs. The estimation results for the factor loadings of As regards other constructs (Table 3), answers the measurement models (i.e., Λ , Λ , Λ , Λ , Λ , and B I PE EE SI were evaluated on a five-point Likert scale. As Λ ) are shown in Subsection 4.1, whereas those for FC regards the behavioral intention, we asked about coefficients of (1) paths from d to ω and (2) paths i PE;i the intention to purchase ZEHs at the time the among latent variables (i.e., C, �, and Γ ) are shown in respondents started to consider the purchase. As Subsection 4.2. In these subsections, the results are regards performance expectancy and effort expec- presented in tables that include posterior mean tancy, we asked about the opinions on the perfor- mance, functions, and maintenance of ZEHs and Table 2. Questions, answer format, and results of the response related equipment at the time when the respon- to use behavior. dents started to become aware of ZEHs. As regards Variable Definition Percentage of answers (%) Yes Table 1. Outline of the questionnaire survey. Use behavior Date Nov 28–30, 2018 z Is the following equipment installed in 58.6 B;i;1 Target Households that purchased custom-built detached your house? ZEHs and that received subsidies from national and/ High-efficiency air conditioner or local governments. 1 ¼ yes; 0 ¼ no. Style of Web survey conducted by a marketing research firm All questions below were also Survey (Rakuten Insight, Inc., Japan). answered in the yes/no format. During the above period, the firm published the z Floor heating 45.1 B;i;2 questions on an online survey portal and identified z Energy-efficient water heater 78.5 B;i;3 the respondent samples who met the target z Induction heating (IH) stove 74.7 B;i;4 conditions and had answered all questions. z HEMS 51.2 B;i;5 The samples were from almost the entire country. z Rechargeable battery 14.8 B;i;6 Sample Size 297 z V2H communication 8.1 B;i;7 We treated all five categories shown in Table A.1 as ZEHs. JOURNAL OF ASIAN ARCHITECTURE AND BUILDING ENGINEERING 9 Table 3. Questions, answer format, and results of the response to constructs other than use behavior. Variable Definition Percentage of answers (%) Strongly disagree Disagree Neutral Agree Strongly agree Behavioral intention z I intended to purchase a ZEH. 4.7 5.1 32.3 37.0 20.9 I;i;1 1 ¼ strongly disagree ; 2 ¼ disagree; 3 ¼ neutral; 4 ¼ agree; 5 ¼ strongly agree. All questions below were also answered in the five-point Likert scale. z I was willing to purchase a ZEH even if it was more expensive than other types of 3.7 6.7 35.4 36.4 17.8 I;i;2 houses. Performance expectancy z I believed that ZEHs would save more utility costs. 1.3 1.7 14.5 46.8 35.7 PE;i;1 z I believed that ZEHs would pay for its initial cost soon if I sold the surplus PV 5.7 8.8 27.6 38.0 19.9 PE;i;2 energy. z I believed that ZEHs would be environmentally friendly because of reduced CO 2.4 6.4 27.3 45.1 18.9 PE;i;3 2 emissions. z Functions of equipment other than energy-saving or -generating were 3.0 6.4 32.7 38.7 19.2 PE;i;4 interesting. z Layout and exterior design of ZEHs were interesting. 4.7 10.1 37.0 34.3 13.8 PE;i;5 z I believed ZEHs would function effectively during disasters and blackouts. 1.7 5.1 25.3 46.1 21.9 PE;i;6 z I believed ZEHs would function effectively for our health because of high thermal- 2.0 0.3 16.5 50.5 30.6 PE;i;7 insulation performance and improved ventilation systems. Effort expectancy z I believed it would be easy to learn how to operate the installed equipment. 0.3 4.7 34.3 45.5 15.2 EE;i;1 z I believed the daily maintenance of the equipment would be easy. 1.0 6.4 31.3 48.5 12.8 EE;i;2 z I believed the maintenance cost of equipment would be reasonable. 0.7 13.1 38.7 38.4 9.1 EE;i;3 z I believed it would be easy to learn the unique functions of the equipment. 1.0 12.1 39.7 36.7 10.4 EE;i;4 Social influence z My family members thought I should purchase a ZEH. 7.1 10.4 37.0 30.0 15.5 SI;i;1 z I believed that to live in a ZEH would be socially preferable. 3.4 5.7 31.3 44.4 15.2 SI;i;2 z I believed that to live in a ZEH would be an advanced lifestyle. 3.0 2.7 29.3 42.8 22.2 SI;i;3 Facilitating conditions z I just had an appropriate opportunity to reconstruct and extend my house. 10.4 7.1 17.5 32.3 32.7 FC;i;1 z I was aware of the subsidies for ZEHs provided by the national and/or the local 10.8 17.2 22.6 32.3 17.2 FC;i;2 government. z I had enough money to purchase a ZEH. 7.4 12.1 41.8 26.9 11.8 FC;i;3 z I was aware of the after-sales service or guarantees of a ZEH or its equipment 8.8 18.9 30.6 28.6 13.1 FC;i;4 should problems arise. z I was aware of the campaigns launched by home builders, such as gift vouchers 10.1 19.2 33.0 24.2 13.5 FC;i;5 and zero interest rate on loans. z I thought the utility costs (e.g., gas and electricity) where I live were high. 5.7 10.8 37.0 30.6 15.8 FC;i;6 z I believed the climatic conditions (e.g., solar radiation and snow) where I live 5.4 11.1 41.4 31.0 11.1 FC;i;7 would be suitable to introduce ZEHs. z Where I live, the local municipality and community implemented activities aimed 13.1 23.9 34.0 20.2 8.8 FC;i;8 at zero energy consumption. (sometimes referred to as “Bayesian estimate” (Lee also list the CS solution for the factor loadings and 2007; Song and Lee 2012)), standard deviation (SD), coefficients of the paths among latent variables. The and highest density interval (HDI). The α% HDI sum- estimation results for all parameters are listed in marizes the distribution by specifying an interval Appendix C. For model checking, we conducted spanning most of the distribution (i.e., α% of it) posterior predictive checking (Gelman et al. 2013) such that every point inside the interval has higher (verifying whether the estimated model fitted the credibility than any point outside it (Kruschke 2014; obtained data appropriately) and, consequently, the Meredith and Kruschke 2016). In the tables for the simulated data from the estimated model were structural model, we also included the probability close to the obtained data from the survey (the details are presented in Appendix D). We also con- that a parameter exceeds 0 (i.e., ∫ pðθjDÞdθ, where ducted sensitivity analysis of the prior distribution θ is a parameter and D is data), denoted by Pr . for parameters of paths between latent variables Sometimes, it is also desirable to transform (details are presented in Appendix E) and, accord- Bayesian estimates to a completely standardized ingly, the estimation results were almost insensitive (CS) solution such that both observed and latent to the prior inputs. variables are standardized (Lee 2007). Thus, we 10 H. KOTANI AND K. NAKANO Table 4. Questions, answer format, and results of the response to information acquisition. Variable Definition Percentage of answers (%) Yes Economic information d Did you obtain economic information through the following channels? 86.9 Econ;Sales;i Salespersons, e.g., face-to-face communication with salespersons of home builders 1 ¼ yes; 0 ¼ no. The values of all variables below were assigned in the same way. d Friends, e.g., word-of-mouth communication (including online) with friends, colleagues, and family members 21.9 Econ;Friends;i d Advertisement, e.g., advertisement by mass media (television, radio, newspapers, magazines, web pages, and 40.4 Econ;Ad;i social networking service [SNS]) provided by home builders and the government d Strangers, e.g., online word-of-mouth and reviews posted by people other than friends 19.5 Econ;Strangers;i Technical information d Did you obtain technical information through the following channels? 60.9 Tech;Sales;i Salespersons d Friends 8.1 Tech;Friends;i d Advertisements 18.5 Tech;Ad;i d Strangers 6.1 Tech;Strangers;i Environmental information d Did you obtain environmental information through the following channels? 55.9 Envi;Sales;i Salespersons d Friends 7.1 Envi;Friends;i d Advertisements 18.9 Envi;Ad;i d Strangers 6.1 Envi;Strangers;i Health information d Did you obtain health information through the following channels? 36.7 Health;Sales;i Salespersons d Friends 4.0 Health;Friends;i d Advertisements 11.1 Health;Ad;i d Strangers 1.7 Health;Strangers;i 4.1. Measurement model estimation parameters λ , λ , λ , λ , and λ ) were FC;2 FC;4 FC;5 FC;7 FC;8 larger than the other observed variables. The estimation results of the measurement models Accordingly, it is implied that z , z , z , FC;i;2 FC;i;4 FC;i;5 —Λ , Λ , Λ , Λ , Λ , and Λ —are presented in B I PE EE SI FC z , and z are connected more strongly with FC;i;7 FC;i;8 Table 5. Regarding use behavior, the row described facilitating conditions. as “z ω ,” which presents the result of the B;i;2 B;i coefficient of the path from the latent variable ω B;i to the observed one z (i.e, factor loading λ ), B;i;2 B;2 4.2. Structural model estimation demonstrated a positive posterior mean (i.e., 0.480) and 95% HDI excluding 0 (i.e., [0.023, 1.001]). The estimation results of the structural model are Similarly, all posterior means of other factor load- discussed in this subsection. First, the results of the ings regarding use behavior (i.e., λ to λ ) were coefficients of paths from information acquisition (i. B;3 B;7 positive and all the 95% HDIs excluded 0. Therefore, e., d ) to performance expectancy (i.e., ω )—C— i PE;i every observed variable was likely positively corre- are presented in Table 6. Some variables of d were lated with the use behavior. The other latent found to strongly influence ω . The coefficient of PE;i variables (behavioral intention, performance expec- d (i.e., health information obtained from Health;Sales;i tancy, effort expectancy, social influence, and facil- salespersons) (i.e., parameter c ) had a larger 3;13 itating conditions) indicated a similar tendency (i.e., mean (i.e., 0.327) and the 95% HDI (i.e., [0.087, significant positive correlations with the corre- 0.571]) excluded 0, implying that households that sponding observed variables) and showed relatively received such information have a higher value of large CS solutions. The CS solution illustrated the ω . On the other hand, the coefficients of PE;i observed variables that were closely linked with the d and d (i.e., c and c ) had Tech;Friends;i Envi;Sales;i 3;6 3;9 corresponding latent variables; for example, as a larger mean (i.e., 0.449 and 0.245, respectively) regards the facilitating conditions, the CS solutions and their 90% HDIs (i.e., [0.064, 0.824] and [0.021, for z , z , z , z , and z (i.e., 0.467], respectively) excluded 0 despite the 95% FC;i;2 FC;i;4 FC;i;5 FC;i;7 FC;i;8 As indicated in Subsection 3.1.1, λ , λ , λ , λ , λ , and λ are fixed as 1 for model identification, which results in posterior mean and SD of B;1 I;1 PE;1 EE;1 SI;1 FC;1 corresponding parameters being 1 and 0, respectively. JOURNAL OF ASIAN ARCHITECTURE AND BUILDING ENGINEERING 11 Table 5. Bayesian estimation results of factor loadings in measurement models. Path Parameter Mean SD 95% HDI CS solution Use behavior z ω λ 1 0 0.501 B;i;1 B;i B;1 z ω λ 0.480 0.250 [0.023, 1.001] 0.267 B;i;2 B;i B;2 z ω λ 1.141 0.410 [0.419, 1.970] 0.551 B;i;3 B;i B;3 z ω λ 0.697 0.315 [0.144, 1.353] 0.374 B;i;4 B;i B;4 z ω λ 1.896 0.562 [0.879, 3.015] 0.739 B;i;5 B;i B;5 z ω λ 0.704 0.322 [0.110, 1.356] 0.377 B;i;6 B;i B;6 z ω λ 1.221 0.498 [0.331, 2.229] 0.577 B;i;7 B;i B;7 Behavioral intention z ω λ 1 0 0.893 I;i;1 I;i I;1 z ω λ 0.977 0.087 [0.811, 1.152] 0.880 I;i;2 I;i I;2 Performance expectancy z ω λ 1 0 0.683 PE;i;1 PE;i PE;1 z ω λ 0.811 0.123 [0.575, 1.050] 0.589 PE;i;2 PE;i PE;2 z ω λ 0.934 0.134 [0.680, 1.199] 0.668 PE;i;3 PE;i PE;3 z ω λ 1.147 0.154 [0.860, 1.456] 0.805 PE;i;4 PE;i PE;4 z ω λ 0.799 0.125 [0.565, 1.049] 0.583 PE;i;5 PE;i PE;5 z ω λ 1.050 0.143 [0.780, 1.332] 0.737 PE;i;6 PE;i PE;6 z ω λ 1.056 0.143 [0.788, 1.340] 0.759 PE;i;7 PE;i PE;7 Effort expectancy z ω λ 1 0 0.892 EE;i;1 EE;i EE;1 z ω λ 0.926 0.138 [0.667, 1.200] 0.726 EE;i;2 EE;i EE;2 z ω λ 0.924 0.139 [0.670, 1.205] 0.748 EE;i;3 EE;i EE;3 z ω λ 1.073 0.151 [0.793, 1.373] 0.840 EE;i;4 EE;i EE;4 Social influence z ω λ 1 0 0.631 SI;i;1 SI;i SI;1 z ω λ 1.310 0.153 [1.023, 1.617] 0.802 SI;i;2 SI;i SI;2 z ω λ 1.376 0.166 [1.059, 1.704] 0.841 SI;i;3 SI;i SI;3 Facilitating conditions z ω λ 1 0 0.578 FC;i;1 FC;i FC;1 z ω λ 1.354 0.177 [1.021, 1.707] 0.736 FC;i;2 FC;i FC;2 z ω λ 1.210 0.165 [0.902, 1.543] 0.671 FC;i;3 FC;i FC;3 z ω λ 1.621 0.197 [1.243, 2.007] 0.854 FC;i;4 FC;i FC;4 z ω λ 1.402 0.182 [1.050, 1.759] 0.761 FC;i;5 FC;i FC;5 z ω λ 1.088 0.158 [0.794, 1.406] 0.611 FC;i;6 FC;i FC;6 z ω λ 1.415 0.177 [1.075, 1.763] 0.777 FC;i;7 FC;i FC;7 z ω λ 1.461 0.180 [1.120, 1.818] 0.803 FC;i;8 FC;i FC;8 Table 6. Bayesian estimation results of coefficients of the paths from d to performance expectancy ω . i PE;i Explanatory variable Parameter Mean SD 95% HDI 90% HDI Pr d c 0.195 0.165 [−0.126, 0.521] [−0.072, 0.467] 0.884 Econ;Sales;i 3;1 d c −0.062 0.159 [−0.382, 0.245] [−0.321, 0.202] 0.345 Econ;Friends;i 3;2 d c 0.193 0.133 [−0.077, 0.446] [−0.030, 0.407] 0.928 Econ;Ad;i 3;3 d c 0.177 0.177 [−0.179, 0.517] [−0.116, 0.465] 0.844 Econ;Strangers;i 3;4 d c −0.010 0.146 [−0.305, 0.272] [−0.257, 0.225] 0.473 Tech;Sales;i 3;5 d c 0.449 0.231 [−0.010, 0.903] [0.064, 0.824] 0.976 Tech;Friends;i 3;6 d c 0.032 0.163 [−0.290, 0.350] [−0.241, 0.294] 0.577 Tech;Ad;i 3;7 d c −0.326 0.249 [−0.821, 0.158] [−0.731, 0.084] 0.091 Tech;Strangers;i 3;8 d c 0.245 0.136 [−0.021, 0.515] [0.021, 0.467] 0.968 Envi;Sales;i 3;9 d c 0.179 0.290 [−0.384, 0.757] [−0.299, 0.653] 0.734 Envi;Friends;i 3;10 d c 0.175 0.160 [−0.132, 0.499] [−0.093, 0.433] 0.867 Envi;Ad;i 3;11 d c −0.219 0.278 [−0.761, 0.335] [−0.680, 0.234] 0.211 Envi;Strangers;i 3;12 d c 0.327 0.123 [0.087, 0.571] 0.997 Health;Sales;i 3;13 d c −0.041 0.302 [−0.635, 0.550] [−0.540, 0.451] 0.447 Health;Friends;i 3;14 d c −0.014 0.193 [−0.392, 0.366] [−0.320, 0.312] 0.473 Health;Ad;i 3;15 d c 0.194 0.446 [−0.671, 1.085] [−0.525, 0.933] 0.669 Health;Strangers;i 3;16 Note: HDI values in bold indicate the corresponding interval does not include zero. HDIs including 0. Accordingly, d and Second, the results of the coefficients of paths Tech;Friends;i between latent variables—� and Γ —are presented in d are likely to positively affect ω , although Envi;Sales;i PE;i they are unsure compared with d . Table 7. Among the paths to behavioral intention, the Health;Sales;i 12 H. KOTANI AND K. NAKANO Table 7. Bayesian estimation results of coefficients of paths between latent variables. Path Parameter Mean SD 95% HDI 90% HDI Pr CS solution ω ! ω π 0.124 0.081 [−0.038, 0.281] [−0.008, 0.257] 0.941 0.097 PE;i I;i 23 ω ! ω γ −0.198 0.144 [−0.480, 0.077] [−0.423, 0.038] 0.069 −0.171 EE;i I;i 21 ω ! ω γ 1.409 0.246 [0.941, 1.887] 1.000 0.957 SI;i I;i 22 ω ! ω γ 0.239 0.145 [−0.052, 0.522] [0.008, 0.484] 0.951 0.144 FC;i I;i ω ! ω π 0.254 0.088 [0.090, 0.429] 1.000 0.427 I;i B;i 12 ω ! ω γ −0.365 0.151 [−0.674, −0.088] 0.002 −0.370 FC;i B;i 13 Note: HDI values in bold indicate the corresponding interval does not include zero. path from social influence (i.e., γ ) had the largest poster- 5.1. Effective combination of information content ior mean (i.e., 1.409) and CS solution (i.e., 0.957), 95% HDI and channels (i.e., [0.941, 1.887]) non-overlapping 0, and Pr indicating The results of Table 6 showed that the following three 100.0%. Accordingly, social influence is considered to combinations of information content and channels likely have a large positive effect on behavioral intention. As had positive effects on the performance expectancy, regards the facilitating conditions (i.e., γ ), the posterior namely, (1) health information obtained from salesper- mean (i.e., 0.239) was positive and the 90% HDI (i.e., sons, (2) environmental information obtained from sales- [0.008, 0.484]) did not include 0. Therefore, the facilitating persons, and (3) technical information obtained from conditions are likely to positively influence the behavioral friends (Figure 2). intention. In addition, each path coefficient of the perfor- The associated health and environmental issues are mance expectancy and effort expectancy (i.e., π and γ ) 23 21 science-based and/or long-term effects, and households had a small posterior mean and the 90% HDI overlapped could have difficulty in evaluating such issues, whereas 0. This means that the two latent variables are unlikely to salespersons would interpret and convey them objec- affect the behavioral intention. tively and accurately. Accordingly, we consider that The paths from the behavioral intention and facilitat- health and environmental information on ZEHs trans- ing conditions to the use behavior (i.e., π and γ ) had 12 13 mitted by salespersons had effectively enhanced the per- positive (i.e., 0.254) and negative (i.e., −0.365) posterior formance expectancy. On the other hand, technical means, respectively, and the 95% HDIs ([0.090, 0.429] and information, e.g., performance, mechanism, and mainte- [−0.674, −0.088], respectively) excluded 0. Accordingly, nance of equipment installed in ZEHs, was likely under- the behavioral intention and facilitating conditions are stood better after the users started using it. Friends can considered to definitively have positive and negative give useful advice (e.g., what equipment is suitable) effects, respectively, on the use behavior. according to the context, leading to the receivers under- standing the technical issues better. Accordingly, techni- 5. Discussion cal information from friends is considered to have raised the performance expectancy. We modeled the purchasing process of ZEHs (Figure 1) by The above findings imply that scientific information, households, including the effects of combining content including on health and the environment, should be and channels of information. We conducted statistical transmitted through objective channels, whereas techni- analysis targeting Japanese households that purchased cal information should be spread through flexible chan- custom-built detached ZEHs. As ZEHs are not common in nels. This implication is consistent with the finding of Japan (Sustainable open Innovation Initiative 2018), the another study (Kotani and Honda 2019) focusing on process needed to be investigated to consider and such combination in a different context (not adoption of develop strategies that would be more effective in pro- technology, but reconstruction of houses); consequently, moting the spreading of such houses. our findings are considered more convincing. Our estimation results, presented in Table 6– 7, are summarized in Figure 2. In this figure, the paths between UTAUT constructs are listed with the CS solutions demon- 5.2. Effective factors in promoting intention strated in Table 7; as regards information acquisition, the formation and behavior combinations of information content and channels hav- As illustrated in Table 7, we found that the intention to ing significant effects on the performance expectancy are purchase a ZEH (i.e., behavioral intention) was promoted listed with their posterior means in Table 6. primarily by the social influence and facilitating conditions Based on our results, we discuss (1) the effective and not by the performance expectancy or effort expec- combining of information content and channels in tancy (Figure 2). This result is consistent with the results of Subsection 5.1 and (2) factors that effectively promote previous studies that examined the adoption process of an the behavioral intention and use behavior in energy-saving product based on the UTAUT Subsection 5.2. Several limitations of this study and (Khorasanizadeh et al. 2016). We found that the effect of future avenues for study are described in social influence was the most significant. The Subsection 5.3. JOURNAL OF ASIAN ARCHITECTURE AND BUILDING ENGINEERING 13 Figure 2. Summary of estimation results: the number next to each arrow indicates the CS solution (as regards information acquisition, posterior mean is presented); ** indicates that 95% Bayesian credible interval does not include 0; * indicates that 90% credible interval does not include 0. measurement model's estimation results (CS solution in Overall, the implication was that the intention to pur- Table 5) indicated that the social influence (i.e., ω ) was chase a ZEH was enhanced primarily by the perception of SI;i linked strongly with “I believed that to live in a ZEH would the normative and institutional aspects of ZEHs (i.e., social be socially preferable” (i.e., z ) and “I believed that to live influence and facilitating conditions), rather than the per- SI;i;2 in a ZEH would be an advanced lifestyle” (i.e., z ). This ception of the technical functions and detailed benefits of SI;i;3 implies that the intention was facilitated considerably by ZEHs (i.e., performance expectancy and effort expec- the subjective norm regarding ZEHs (e.g., to live in ZEHs is tancy). One of the reasons why social influence was socially preferable) and a positive image (e.g., ZEHs are more influential than performance expectancy or effort equipped with advanced technologies). This finding is expectancy could be the characteristics of the adopters consistent with a previous study that stresses the impor- (buyers). As mentioned in Section 1, Japanese society is tance of subjective norms (Schepers and Wetzels 2007). still in the initial stages of widespread acceptance of such Our survey included an open-ended question on concerns houses. According to Rogers (2003), adopters in the initial before the purchase and the deciding factors for the stages—early adopters—tend to be respected by their purchase. One response was, “I wanted to live in an envir- peers, and to continue to earn this esteem, they must onmentally friendly house even if it was expensive.” make judicious decisions to adopt the innovation; there- Another reported, “ . . . despite the high initial investment fore, they are probably quite susceptible to social influ - cost, I decided to purchase a ZEH as I did not want to live in ence. Another reason for performance expectancy or an out-of-date house.” These answers also likely validate effort expectancy not being influential was probably our findings on the significant effect of social influence. that the government and private business operators The estimation results of the measurement models (CS had just started their promotional activities and therefore solution in Table 5) indicated that the facilitating condi- had insufficient time to inform prospective buyers of the tions were linked strongly with the following: “I was aware benefits of ZEHs so that the intention to purchase could of the subsidies for ZEHs provided by the national and/or be activated. the local government” (i.e., z ); “I was aware of the As presented in Table 7, we also found that the use FC;i;2 after-sales service or guarantees of a ZEH or its equipment behavior and behavioral intention were correlated posi- should problems arise” (i.e., z ); “I was aware of the tively (Figure 2). In other words, the behavior intention FC;i;4 likely led households to install equipment to improve the campaigns launched by home builders, such as gift vou- performance of ZEHs. On the other hand, against our chers and zero-interest rate on loans” (i.e., z ); and FC;i;5 expectations, the facilitating conditions were found to “Where I live, the local municipality and community have a negative effect on the use behavior (Figure 2). As implemented activities aimed at zero energy consump- mentioned in Subsection 3.2, the use behavior represents tion” (i.e., z ). Accordingly, the support systems pro- FC;i;8 the degree of installation of additional energy-saving and vided by various stakeholders, such as national and local energy-creating equipment, which improves the perfor- governments, home builders, and local communities are mance of ZEHs. The subsidies from the national and local also considered triggering factors in the intention to pur- governments, one of the elements of the facilitating con- chase. This inference is also evident from the following ditions, could have induced behavior other than the answers to the open-ended question, “ . . . campaigns and behavior to install the additional equipment. This beha- subsidies for ZEHs were useful” and “ . . . the subsidies let vior could be to upgrade the necessary equipment (e.g., me decide to purchase a ZEH.” 14 H. KOTANI AND K. NAKANO PV panels, heat insulation walls, and LED lamps) for ZEHs. 6. Conclusion and policy implications One household also responded to the open-ended ques- Targeting Japanese households that purchased custom- tion, “I upgraded the PV panels because of the subsidies.” built detached ZEHs, this study aimed to explore how Overall, possibly, the facilitating conditions and use beha- intention and behavior to purchase were facilitated by vior were negatively correlated owing to the following a range of factors, including both the information content hypothesis, “the performance expectancy promoted the and the channels buyers used. Our conceptual model was behavior to upgrade the necessary equipment for ZEHs constructed based on the UTAUT (Figure 1) and estimated but offset it, terminating the installation of optional by means of Bayesian SEM, which is suitable to treat small equipment.” This is one of the possible hypotheses that samples and discrete variables, with data collected from requires further analysis in the future. Future work should approximately 300 households. As already mentioned in also include: (1) We should improve the observed vari- Section 5, our main results are shown in Figure 2. The first ables to measure the use behavior since some of the CS prominent result is that performance expectancy, i.e., the solutions of use behavior, such as those for z , z , and B;i;2 B;i;4 perception of the usefulness and benefits of ZEHs was z , were small (Table 5) (e.g., it would be important to B;i;6 enhanced significantly by certain combinations of infor- consider the performance or quality of each equipment, mation content and channels. These are (1) health infor- as well as include the necessary equipment). (2) We mation obtained from salespersons, (2) environmental should explore the multifaceted effects of the facilitating information obtained from salespersons, and (3) technical conditions on the purchase behavior of households. information obtained from friends. The second, and the most prominent, result is that behavioral intention was facilitated more significantly by the social influence and 5.3. Limitations and future directions facilitating conditions (i.e., perception of social image of Including topics that we have already described, the ZEHs and support provided by various stakeholders, e.g., current study has some limitations, and future work national and local governments, home builders, and local should investigate various aspects. (1) We focused communities) than by the performance expectancy or on the effect of information acquisition exclusively effort expectancy (i.e., the perception of performance on the performance expectancy. As described in and benefits of ZEHs and installed equipment). The third Subsection 2.2, this was because the content of result is that the behavioral intention positively affected information we targeted was mainly related to the the use behavior, i.e., the degree of installation of addi- benefits of ZEHs, and the parsimonious model tional equipment to improve the performance of ZEHs. In eased the estimation and interpretation. However, sum, our study revealed a series of possible processes for example, information through SNS or other conducted by households to purchase ZEHs: channels might enhance the ZEHs’ positive image. To explore further details, the effects on other con- The above-mentioned three combinations of structs (e.g., social influence) should be investigated information content and channels substantially in the future. (2) The data used in this study were promoted the performance expectancy of ZEHs. collected after purchase and from households that The performance expectancy, however, was voluntarily answered all questions, which may uncertain to have a significant effect on the inten- include cognitive biases to justify their purchase as tion to purchase. well as sampling bias; therefore, analysis based on The intention was significantly facilitated by the random sampling data collected since before pur- normative and institutional aspects of ZEHs, chase will add further evidence. (3) We limited the which, consequently, led to further installation samples to the households that purchased ZEHs. of equipment to improve the quality of ZEHs. Future work should collect samples of households that purchased non-ZEHs and compare the differ - Based on the above results, we propose several policy ences between the two. (4) Our study focused only implications. The above-mentioned three combinations on the initial stages of adoption, whereas the char- should be exploited for effective information provision acteristics of adopters generally differ according to strategies. For example, it could be effective for sales- the stages (Rogers 2003); therefore, our discussion persons of home builders to provide consumers with are considered valid at most for the present and health information (e.g., ZEHs can contribute to improv- near future. To discuss longer-term policies, follow- ing blood pressure and reducing the risk of heat shock) up studies are needed. (5) The present study and environmental information (e.g., ZEHs can contribute focused exclusively on custom-built detached to greenhouse gas reduction, as well as disaster risk houses, but it is also important to target renovated reduction because of using PV panels and rechargeable houses (Oki et al. 2019) and zero-energy apartments batteries). Economic information was not clearly effective, (ZEH-M) (Sustainable open Innovation Initiative regardless of channels, but it showed positive effects with 2018). relatively high probability if provided by salespersons and JOURNAL OF ASIAN ARCHITECTURE AND BUILDING ENGINEERING 15 advertisements (Table 6). Therefore, to enhance the effect References of economic information, it would be appropriate to Assaf, A. G., M. Tsionas, and H. Oh. 2018. “The Time Has Come: focus on these channels. Through these strategies, the Toward Bayesian SEM Estimation in Tourism Research.” performance expectancy of ZEHs could be enhanced, so Tourism Management 64: 98–109. doi:10.1016/j. that the intention to purchase a ZEH is stimulated, which, tourman.2017.07.018. Attia, S., A. De Herde, E. Gratia, and J. L. M. Hensen. 2013. probably, would affect the purchase decision. On the “Achieving Informed Decision-making for Net Zero Energy other hand, as mentioned previously, the intention was Buildings Design Using Building Performance Simulation facilitated significantly by the normative and institutional Tools.“ Building Simulation 6: 3–21. doi:10.1007/s12273-013- aspects (i.e., the social influence and facilitating condi- 0105-z. tions) and, therefore, it could be concluded that society as Bollinger, B., and K. Gillingham. 2012. “Peer Effects in the a whole should take a positive position to improve accep- Diffusion of Solar Photovoltaic Panels.” Marketing Science 31 (6): 900–912. doi:10.1287/mksc.1120.0727. tance. For example, the promotion of public discourse California Public Utilities Commission, n.d. Zero Net Energy. saying that ZEHs are up-to-date and socially preferable Chevalier, J. A., and D. Mayzlin. 2006. “The Effect of Word of houses, the branding strategies implemented by the Mouth on Sales: Online Book Reviews.” Journal of Marketing government and private sectors, and the support from Research 43 (3): 345–354. doi:10.1509/jmkr.43.3.345. various stakeholders (e.g., after-sales service from private Curtale, R., F. Liao, and P. van der Waerden. 2021. “User Acceptance of Electric Car-sharing Services: The Case of business operators and activities aimed at zero energy the Netherlands.” Transportation Research Part A: Policy consumption implemented by local municipalities and and Practice 149: 266–282. doi:10.1016/j.tra.2021.05.006. communities) were effective and should be continued. de Schoot, R., D. Kaplan, J. Denissen, J. B. Asendorpf, Although our analysis has some limitations, the afore- F. J. Neyer, and M. A. G. Van Aken. 2014. “A Gentle mentioned introductory findings and implications could Introduction to Bayesian Analysis: Applications to contribute to assessing and improving the policies and Developmental Research.” Child Development 85 (3): 842–860. doi:10.1111/cdev.12169. activities to ensure widespread acceptance of ZEHs. de Schoot, R., S. D. Winter, O. Ryan, M. Zondervan-Zwijnenburg, and S. Depaoli. 2017. “A Systematic Review of Bayesian Articles in Psychology: The Last 25 Years.” Psychological Disclosure statement Methods 22 (2): 217. doi:10.1037/met0000100. Duan, W., B. Gu, and A. B. Whinston. 2008. “Do Online No potential conflict of interest was reported by the Reviews matter?—An Empirical Investigation of Panel author(s). Data.” Decision Support Systems 45 (4): 1007–1016. doi:10.1016/j.dss.2008.04.001. Farhar, B., and T. Coburn. 2008. “A New Market Paradigm for Funding Zero-energy Homes: A Comparative Case Study.” Environment: Science and Policy for Sustainable Development This study was supported partly by the Japan Society for the 50 (1): 18–32. doi:10.3200/ENVT.50.1.18-32. Promotion of Science (JSPS) Grants-in-Aid for Scientific Gelman, A., J. B. Carlin, H. S. Stern, D. B. Dunson, A. Vehtari, Research (KAKENHI) (Grant No. 18K13845) and the Maeda and D. B. Rubin. 2013. Bayesian Data Analysis. New York: Engineering Foundation. Chapman and Hall/CRC. Gelman, A., and D. B. Rubin.others. 1992. “Inference from Iterative Simulation Using Multiple Sequences.” Statistical Science 7 (4): 457–472. doi:10.1214/ss/1177011136. Notes on contributors Goto, H., M. Goto, and T. Sueyoshi. 2011. “Consumer Choice on Hitomu Kotani is an assistant professor at Kyoto University, Ecologically Efficient Water Heaters: Marketing Strategy and Japan. He obtained his Ph.D. in Engineering in 2016 from Policy Implications in Japan.” Energy Economics 33 (2): Kyoto University. His research interests are urban and regio- 195–208. doi:10.1016/j.eneco.2010.09.004. nal planning as well as social network analysis. Graziano, M., and K. Gillingham. 2014. “Spatial Patterns of Solar Photovoltaic System Adoption: The Influence of Neighbors Kazuyoshi Nakano is a researcher at Central Research and the Built Environment.” Journal of Economic Geography Institute of Electric Power Industry, Japan. He received his 15 (4): 815–839. doi:10.1093/jeg/lbu036. Ph.D. in Informatics in 2011 from Kyoto University, and his Hartwich, F., C. Witzlack, M. Beggiato, and J. F. Krems. 2019. “The specialized fields are consumer behavior and energy demand First Impression Counts – A Combined Driving Simulator and and supply in Japan. Test Track Study on the Development of Trust and Acceptance of Highly Automated Driving.” Transportation Research Part F: Traffic Psychology and Behaviour 65: ORCID 522–535. doi:10.1016/j.trf.2018.05.012. Hox, J. J. 1998. “An Introduction to Structural Equation Hitomu Kotani http://orcid.org/0000-0002-5033-0855 Modeling.” Family Science Review 11: 354–373. International Energy Agency, 2018. World Energy Outlook Japan Photovoltaic Energy Association (JPEA), 2018. Report on Availability of data autonomous operation of photovoltaic power system in Data will be made available on request. disaster. 16 H. KOTANI AND K. NAKANO Khorasanizadeh, H., A. Honarpour, M. S.-A. Park, J. Parkkinen, and Ministry of Economy, Trade and Industry, 2018b. Basic Energy R. Parthiban. 2016. “Adoption Factors of Cleaner Production Plan. Technology in a Developing Country: Energy Efficient Monthly Smart House. 2018a. Safe and Secure. Japan: Lighting in Malaysia.” Journal of Cleaner Production 131: ASKLAST CO., LTD. 97–106. doi:10.1016/j.jclepro.2016.05.070. Monthly Smart House. 2018b. ZEH Master 2018. Japan: Kotani, H., and R. Honda. 2019. “Effective Combinations of ASKLAST CO., LTD. Information Content and Channels for the Post-disaster Noll, D., C. Dawes, and V. Rai. 2014. “Solar Community Reconstruction of Rural Housing: A Case Study of the Organizations and Active Peer Effects in the Adoption 2015 Gorkha Nepal Earthquake.“ International Journal of of Residential PV.” Energy Policy 67: 330–343. Disaster Risk Reduction 39: 101118. doi:10.1016/j. doi:10.1016/j.enpol.2013.12.050. ijdrr.2019.101118. Ofuji, K., and K. Nishio. 2013. “Analysis of Residential Water Kotani, H., R. Honda, and R. Sugitani. 2019. “Transition in Heater Choice Using Multinomial Choice Models.” Journal of Subjective Factors to Trigger Recovery Action of Rural Environmental Engineering (Transactions of AIJ) 78 (683): Residents Affected by the 2015 Gorkha Nepal 89–95. doi:10.3130/aije.78.89. Earthquake.” International Journal of Disaster Risk Oki, R., Y. Tsuneoka, S. Yamaguchi, S. Sugano, Reduction 39: 101239. doi:10.1016/j.ijdrr.2019.101239. N. Watanabe, T. Akimoto, Y. Hayashi, S. Wakao, and Kruschke, J. 2014. Doing Bayesian Data Analysis: A Tutorial S. Tanabe. 2019. “Renovating a House to Aim for Net- with R, JAGS, and Stan. Academic Press. zero Energy, Thermal Comfort, Energy Self-consump- tion and Behavioural Adaptation: A Method Proposed Lee, S.-Y. 2007. Structural Equation Modeling: A Bayesian for ENEMANE HOUSE 2017.“ Energy and Buildings 201: Approach. John Wiley & Sons. 183–193. doi:10.1016/j.enbuild.2019.06.041. Lee, S.-Y., and X.-Y. Song. 2004. “Evaluation of the Bayesian and Maximum Likelihood Approaches in Analyzing Park, E.-S., B. Hwang, K. Ko, and D. Kim. 2017. “Consumer Structural Equation Models with Small Sample Sizes.” Acceptance Analysis of the Home Energy Management Multivariate Behavioral Research 39 (4): 653–686. System.” Sustainability 9 (12): 2351. doi:10.3390/su9122351. doi:10.1207/s15327906mbr3904_4. Persson, J., and S. Grönkvist. 2015. “Drivers for and Barriers to Lee, S.-Y., and X.-Y. Song. 2012. Basic and Advanced Bayesian Low-energy Buildings in Sweden.” Journal of Cleaner Structural Equation Modeling: With Applications in the Medical Production 109: 296–304. doi:10.1016/j.jclepro.2014.09.094. and Behavioral Sciences. John Wiley & Sons. Petersen, A., G. Michael, and J. Corvidae. 2019. The Economics Lee, S.-Y., X.-Y. Song, and J.-H. Cai. 2010. “A Bayesian of Zero-Energy Homes: Single-Family Insights. Rocky Approach for Nonlinear Structural Equation Models Mountain Institute. with Dichotomous Variables Using Logit and Probit Plummer, M.,others, 2003. “JAGS: A Program for Analysis of Links.” Structural Equation Modeling: A Multidisciplinary Bayesian Graphical Models Using Gibbs Sampling.” In: Journal 17 (2): 280–302. doi:10.1080/ Proceedings of the 3rd International Workshop on 10705511003659425. Distributed Statistical Computing Vienna, Austria, 1–10. Lee, M. D., and E.-J. Wagenmakers. 2014. Bayesian Cognitive Rajapakse, J., 2011 “Extending the Unified Theory of Modeling: A Practical Course. Cambridge University Press. Acceptance and Use of Technology (UTAUT) Model.” In: Levy, R., and R. J. Mislevy. 2017. Bayesian Psychometric The 4th International Conference on Interaction Sciences, Modeling. Chapman and Hall/CRC. Busan, Korea (South), 47–52. Lu, Q., J. Chai, S. Wang, Z. G. Zhang, and X. C. Sun. 2020. Rezaei, R., and M. Ghofranfarid. 2018. “Rural Households’ “Potential Energy Conservation and CO2 Emissions Renewable Energy Usage Intention in Iran: Extending the Reduction Related to China’s Road Transportation.” Journal Unified Theory of Acceptance and Use of Technology.” of Cleaner Production 245: 118892. doi:10.1016/j. Renewable Energy 122: 382–391. doi:10.1016/j. jclepro.2019.118892. renene.2018.02.011. Ma, B., Y. Yu, and F. Urban. 2018. “Green Transition of Energy Rogelj, J., M. Den Elzen, N. Höhne, T. Fransen, H. Fekete, Systems in Rural China: National Survey Evidence of H. Winkler, R. Schaeffer, F. Sha, K. Riahi, and Households’ Discrete Choices on Water Heaters.” Energy M. Meinshausen. 2016. “Paris Agreement Climate Policy 113: 559–570. doi:10.1016/j.enpol.2017.11.046. Proposals Need a Boost to Keep Warming Well below 2° Meredith, M., and J. Kruschke, 2016. “HDInterval: Highest C.” Nature 534 (7609): 631. doi:10.1038/nature18307. (Posterior) Density Intervals.”. R Packag. version 0.1 3. Rogers, E. M. 2003. Diffusion of Innovations. Free Press. Merkle, E. C., and T. Wang. 2018. “Bayesian Latent Variable Schepers, J., and M. Wetzels. 2007. “A Meta-analysis of the Models for the Analysis of Experimental Psychology Data.” Technology Acceptance Model: Investigating Subjective Psychonomic Bulletin & Review 25 (1): 256–270. Norm and Moderation Effects.” Information & Management doi:10.3758/s13423-016-1016-7. 44 (1): 90–103. doi:10.1016/j.im.2006.10.007. Michelsen, C. C., and R. Madlener. 2012. “Homeowners’ Shi, F., S. Wang, J. Huang, and X. Hong. 2020. “Design Preferences for Adopting Innovative Residential Heating Strategies and Energy Performance of a Net-zero Energy Systems: A Discrete Choice Analysis for Germany.” Energy House Based on Natural Philosophy.” Journal of Asian Economics 34 (5): 1271–1283. doi:10.1016/j. Architecture and Building Engineering 19 (1): 1–15. eneco.2012.06.009. doi:10.1080/13467581.2019.1696206. Miller, W. 2015. “What Does Built Environment Research Have Song, X.-Y., and S.-Y. Lee. 2012. “A Tutorial on the Bayesian to Do with Risk Mitigation, Resilience and Disaster Approach for Analyzing Structural Equation Models.” Recovery? Sustain.” Sustainable Cities and Society 19: Journal of Mathematical Psychology 56 (3): 135–148. 91–97. doi:10.1016/j.scs.2015.07.004. doi:10.1016/j.jmp.2012.02.001. Ministry of Economy, Trade and Industry, 2015. Sovacool, B. K. 2017. “Experts, Theories, and Electric Definition of ZEH and future measures proposed by Mobility Transitions: Toward an Integrated Conceptual the ZEH Roadmap Examination Committee. Framework for the Adoption of Electric Vehicles.” Ministry of Economy, Trade and Industry, 2018a. Japan’s Energy Research & Social Science 27: 78–95. Energy White Paper 2018. doi:10.1016/j.erss.2017.02.014. JOURNAL OF ASIAN ARCHITECTURE AND BUILDING ENGINEERING 17 Stefanović, A., M. Bojić, and D. Gordić. 2014. “Achieving Net Yamaguchi, Y., K. Akai, J. Shen, and N. Fujimura. 2010. Zero Energy Cost House from Old Thermally Non-insulated “Prediction of Photovoltaics and Solar Thermal System House Using Photovoltaic Panels.” Energy and Buildings 76: Diffusion and Evaluation of Promotion Policy Based on 57–63. doi:10.1016/j.enbuild.2014.02.057. Consumer’s Choice.” Journal of Japan Society of Energy Su, Y.-S., and M. Yajima, 2015. “R2jags: Using R to Run and Resources 31: 38–44. “JAGS”.” R package version 0.5-7Available: CRAN. Ye, Q., R. Law, B. Gu, and W. Chen. 2011. “The Influence of R-project. org/package= R2jags. (September 2015). User-generated Content on Traveler Behavior: An Empirical Sustainable open Innovation Initiative, 2017. “Report of Net Investigation on the Effects of E-word-of-mouth to Hotel Zero-energy House Support Projects in 2017.” Online Bookings.” Computers in Human Behavior 27 (2): Sustainable open Innovation Initiative, 2018. “Report of Net 634–639. doi:10.1016/j.chb.2010.04.014. Zero-energy House Support Projects.” Young, W. R., Jr, 2009. “Emergency Power for Critical Items.” Taiwo, A. A., and A. G. Downe. 2013. “The Theory of User In: AIP Conference Proceedings, vol.1157, 201–206. Acceptance and Use of Technology (UTAUT): doi:10.1063/1.3208023. A Meta-analytic Review of Empirical Findings.” Journal ZEH Roadmap Follow-up Committee, 2018. Summary of Theoretical and Applied Information Technology 49: proposed by the ZEH Roadmap Follow-up 48–58. Committee. United Nations, 2015. Adoption of the Paris Agreement. Zhao, X., W. Pan, and L. Chen. 2018. “Disentangling the Venkatesh, V., M. G. Morris, G. B. Davis, and F. D. Davis. 2003. Relationships between Business Model Innovation for Low “User Acceptance of Information Technology: Toward a or Zero Carbon Buildings and Its Influencing Factors Using Unified View.“ MIS Quarterly 27: 425–478. doi:10.2307/ Structural Equation Modelling.” Journal of Cleaner 30036540. Production 178: 154–165. doi:10.1016/j.jclepro.2018.01.010. 18 H. KOTANI AND K. NAKANO Appendix A History of ZEHs in Japan Table A.1 (Continued). Category Definition To contextualize the study, this section explains the history of ZEH+ ● Reduces the primary energy consumption by 25% or ZEHs and the policy and activities taken to ensure widespread more, excluding renewable energy, and by 100% or acceptance in Japan. As briefly mentioned in Subsection 1.1, more, including renewable energy, compared with increasing the number of ZEH constructions is one of the main the standard primary energy consumption. strategies for decarbonization in Japan. The Japanese govern- ● Meets two or more of the following conditions: (1) ment target is that by 2020 more than half of the new residen- strengthened envelope quality, (2) installation of HEMS, and (3) installation of charging facilities for EV. tial detached houses constructed should be ZEHs. To meet this Nearly ZEH+ Meets the requirements of ZEH+, but the renewable target, the construction of ZEHs has been subsidized since 2012 energy offset is only more than 75% of the energy by the Ministry of Economy, Trade and Industry. The initiative consumption. has also been updated, and the “ZEH Roadmap,” launched in ZEH Oriented ● Achieves the same level as ZEH in energy efficiency. 2015, proposes measures to enhance cooperation between the However, reducing primary energy consumption, government, industry groups, and private business operators including renewable energy, is not required. (Ministry of Economy, Trade and Industry 2015). Since 2016, ● Applicable in narrow land areas in urban regions. home construction companies aiming at constructing 50% or more of their orders as ZEHs (including “Nearly ZEHs” ) have been designated officially as qualified ZEH home builders. Appendix B. Sample characteristics In 2018, a new roadmap (ZEH Roadmap Follow-up Committee 2018) was launched, revealing the new categor- Table B.1 shows the sample characteristics, namely, age of ization of ZEHs, which includes the following five categories respondents, family structure, number of family members, according to the climate and geographical conditions: “ZEH,” annual household income, construction year of house, and “Nearly ZEH,” “ZEH+,” “Nearly ZEH+,” and “ZEH Oriented.” the ZEH category. The details of each category are presented in Table A.1. Currently, government ministries allocate subsidies to Table B.1 Sample characteristics builders constructing houses according to these five cate- gories. For example, the Ministry of the Environment pro- Characteristics Choices n % vides 700 thousand yen (approximately 6500 USD) for a ZEH, Age whereas the Ministry of Economy, Trade and Industry pro- 10–19 0 0.0 vides 1.15 million yen (approximately 10,500 USD) for a ZEH 20–29 18 6.1 +. Householders are required to make use of qualified home 30–39 131 44.1 builders to receive the national government subsidies. In 40–49 94 31.6 some areas, local governments also provide subsidies for 50–59 26 8.8 the construction of ZEHs. 60–69 22 7.4 In addition to such subsidies, other measures are taken to 70 and older 6 2.0 promote ZEHs (Sustainable open Innovation Initiative 2017). Family structure The government provides information on ZEHs through the Single 9 3.0 media and encourages qualified home builders to use their Couple 55 18.5 qualification logo (e.g., display the logo on their brochures) for Couple and children (Respondent lives 197 66.3 the branding of ZEHs. Private business operators also advertise with children) ZEHs, featuring them on portal sites of real estate, and holding Couple and children (Respondent lives 17 5.7 seminars and exhibitions for consumers. Despite their efforts, with parents) the current market share of ZEHs among newly constructed Over three generations 18 6.1 detached houses is only approximately 10% of the national Others 1 0.3 average. This implies that to meet the government goal, the Number of family members spreading rate of such houses needs to be increased 1 9 3.0 substantially. 2 55 18.5 3 81 27.3 Table A.1 Categories and definitions of ZEHs 4 104 35.0 Category Definition 5 32 10.8 ZEH ● 6 13 4.4 Meets the criteria for strengthened outer walls, defined according to the regions. (This is also the case with 7 and more 3 1.0 Nearly ZEH, ZEH+, Nearly ZEH+, and ZEH Oriented.) Household annual Reduces the primary energy consumption by 20% or income more, excluding renewable energy, and by 100% or Less than 2 million yen 2 0.7 more, including renewable energy, compared with 2 million yen or more and less than 8 2.7 the standard primary energy consumption. 3 million yen Nearly ZEH Reduces the primary energy consumption by 20% or 3 million yen or more and less than 9 3.0 more, excluding renewable energy, and by 75% or more 4 million yen but less than 100%, including renewable energy, com- pared with the standard primary energy consumption. 4 million yen or more and less than 28 9.4 Applicable in three climate zones, namely, cool, low 5 million yen solar-radiation, and heavy snow. (Continued) (Continued) Until 2017, houses that satisfied the following conditions were defined as “Nearly ZEHs”: (1) houses that meet the criteria for strengthened outer walls and UA value and (2) houses that reduce primary energy consumption, excluding renewable energy, by 20% or more, and, including renewable energy, by 75% or more but less than 100% from the standard primary energy consumption. JOURNAL OF ASIAN ARCHITECTURE AND BUILDING ENGINEERING 19 Table B.1 (Continued). Table C.1 (Continued). Characteristics Choices n % Parameters Mean SD 95% HDI 5 million yen or more and less than 39 13.1 μ −0.563 0.156 [−0.865, −0.253] PE;7 6 million yen ψ 0.665 0.108 [0.467, 0.883] εPE;1 6 million yen or more and less than 78 26.3 ψ 0.719 0.085 [0.556, 0.885] εPE;2 8 million yen ψ 0.627 0.082 [0.473, 0.791] εPE;3 8 million yen or more and less than 51 17.2 ψ 0.414 0.066 [0.290, 0.545] 10 million yen εPE;4 10 million yen or more and less than 28 9.4 ψ 0.720 0.084 [0.562, 0.885] εPE;5 12.5 million yen ψ 0.538 0.078 [0.396, 0.698] εPE;6 12.5 million yen or more and less than 14 4.7 ψ 0.476 0.078 [0.330, 0.631] εPE;7 15 million yen α −1.904 0.110 [−2.102, −1.684] PE;1;2 15 million yen and more 15 5.1 α −0.993 0.105 [−1.200, −0.788] PE;1;3 Unsure 10 3.4 α −1.094 0.076 [−1.240, −0.943] PE;2;2 Refusal to disclose 15 5.1 α −0.253 0.075 [−0.396, −0.103] PE;2;3 Construction year α −1.390 0.107 [−1.591, −1.174] PE;3;2 2018 58 19.5 α −0.416 0.089 [−0.594, −0.246] PE;3;3 2017 88 29.6 α −1.338 0.097 [−1.521, −1.143] PE;4;2 2016 76 25.6 α −0.271 0.080 [−0.429, −0.115] PE;4;3 2015 65 21.9 α −1.086 0.085 [−1.251, −0.921] 2014 5 1.7 PE;5;2 α −0.026 0.080 [−0.182, 0.130] PE;5;3 2013 2 0.7 α −1.537 0.119 [−1.767, −1.304] 2012 3 1.0 PE;6;2 α −0.539 0.094 [−0.717, −0.350] Category of ZEH PE;6;3 α −1.938 0.073 [−2.047, −1.796] ZEH 265 89.2 PE;7;2 α −0.878 0.098 [−1.072, −0.689] Nearly ZEH 12 4.0 PE;7;3 Effort expectancy ZEH+, Nearly Z+, or ZEH Oriented 20 6.7 μ 0.134 0.123 [−0.107, 0.368] EE;1 μ −0.041 0.097 [−0.230, 0.149] EE;2 μ −0.040 0.121 [−0.278, 0.195] EE;3 μ −0.031 0.104 [−0.233, 0.173] EE;4 Appendix C. Details of the posterior ψ 0.182 0.057 [0.082, 0.295] εEE;1 distribution ψ 0.542 0.078 [0.397, 0.700] εEE;2 The summary statistics of the posterior distribution related to ψ 0.474 0.082 [0.323, 0.640] εEE;3 the parameters except for those presented in Table 5– 7 are ψ 0.338 0.059 [0.227, 0.454] εEE;4 presented in Table C.1 and Table C.2. α −1.273 0.275 [−1.805, −0.747] EE;1;2 α −0.146 0.149 [−0.439, 0.134] EE;1;3 α −1.493 0.149 [−1.773, −1.193] EE;2;2 Table C.1 Bayesian estimation results of the parameters, not α −0.371 0.109 [−0.587, −0.160] EE;2;3 shown in Table 5, in the measurement models α −1.094 0.179 [−1.440, −0.743] EE;3;2 α −0.033 0.123 [−0.274, 0.207] Parameters Mean SD 95% HDI EE;3;3 Use behavior α −1.142 0.149 [−1.430, −0.848] EE;4;2 α −0.009 0.104 [−0.215, 0.191] μ 0.227 0.084 [0.064, 0.395] EE;4;3 B;1 Social influence μ −0.135 0.076 [−0.286, 0.013] B;2 μ −0.025 0.075 [−0.171, 0.125] μ 0.893 0.114 [0.679, 1.122] SI;1 B;3 μ −0.010 0.082 [−0.166, 0.154] μ 0.697 0.087 [0.528, 0.872] SI;2 B;4 μ 0.038 0.079 [−0.115, 0.195] μ 0.006 0.105 [−0.203, 0.211] SI;3 B;5 ψ 0.657 0.076 [0.515, 0.809] μ −1.128 0.108 [−1.343, −0.919] εSI;1 B;6 ψ 0.414 0.060 [0.300, 0.533] εSI;2 μ −1.679 0.204 [−2.092, −1.316] B;7 ψ 0.340 0.059 [0.230, 0.459] Behavioral intention εSI;3 α −0.966 0.073 [−1.107, −0.825] μ −0.041 0.090 [−0.219, 0.135] SI;1;2 I;1 α 0.046 0.073 [−0.100, 0.187] SI;1;3 μ −0.066 0.092 [−0.249, 0.112] I;2 α −1.336 0.095 [−1.517, −1.147] SI;2;2 ψ 0.240 0.048 [0.149, 0.334] εI;1 α −0.292 0.084 [−0.455, −0.127] SI;2;3 ψ 0.262 0.047 [0.173, 0.356] εI;2 α −1.494 0.106 [−1.691, −1.281] SI;3;2 α −1.257 0.085 [−1.424, −1.092] I;1;2 α −0.378 0.083 [−0.546, −0.220] SI;3;3 α −0.220 0.072 [−0.361, −0.077] I;1;3 Facilitating conditions α −1.260 0.092 [−1.435, −1.076] I;2;2 μ 0.014 0.068 [−0.117, 0.148] FC;1 α −0.157 0.076 [−0.305, −0.009] I;2;3 μ 0.005 0.073 [−0.140, 0.147] FC;2 Performance expectancy μ −0.018 0.078 [−0.169, 0.136] FC;3 μ −0.549 0.155 [−0.858, −0.251] PE;1 μ −0.003 0.079 [−0.158, 0.151] FC;4 μ −0.469 0.131 [−0.728, −0.214] PE;2 μ 0.000 0.076 [−0.146, 0.152] FC;5 μ −0.542 0.148 [−0.834, −0.254] PE;3 μ 0.019 0.076 [−0.131, 0.168] μ −0.658 0.165 [−0.983, −0.338] FC;6 PE;4 μ −0.018 0.083 [−0.181, 0.145] μ −0.484 0.134 [−0.750, −0.224] FC;7 PE;5 μ −0.042 0.076 [−0.191, 0.106] μ −0.603 0.161 [−0.918, −0.287] FC;8 PE;6 (Continued) (Continued) 20 H. KOTANI AND K. NAKANO Table C.2 (Continued). Table C.1 (Continued). ϕ 0.435 0.089 [0.266, 0.609] Parameters Mean SD 95% HDI ϕ 0.262 0.049 [0.172, 0.361] ψ 0.685 0.090 [0.516, 0.863] 32 εFC;1 ϕ 0.343 0.077 [0.201, 0.495] ψ 0.532 0.064 [0.413, 0.660] εFC;2 ψ 0.614 0.070 [0.482, 0.752] εFC;3 ψ 0.335 0.048 [0.244, 0.430] εFC;4 ψ 0.490 0.060 [0.376, 0.608] εFC;5 Appendix D. Posterior predictive check ψ 0.681 0.078 [0.533, 0.837] εFC;6 Graphical posterior predictive check is one of the methods for ψ 0.452 0.058 [0.344, 0.570] εFC;7 posterior predictive checking (Gelman et al. 2013). We drew ψ 0.402 0.053 [0.301, 0.507] εFC;8 simulated values from the joint posterior predictive distribu- α −0.911 0.060 [−1.027, −0.793] FC;1;2 tion of replicated data (replicated 15,000 data units for each α −0.357 0.061 [−0.478, −0.238] FC;1;3 household) and compared these samples with the observed α −0.604 0.066 [−0.731, −0.471] FC;2;2 data. The supplemental files (available at https://doi.org/10. α −0.012 0.067 [−0.142, 0.119] FC;2;3 14989/269264) demonstrate how accurately the estimated α −0.877 0.075 [−1.021, −0.730] FC;3;2 model predicts the answers of the households shown in α 0.231 0.076 [0.084, 0.381] FC;3;3 Tables 2 and Tables 3. The observed data are expressed by α −0.605 0.072 [−0.745, −0.464] FC;4;2 line plots (use behavior) and histograms (behavioral intention, α 0.148 0.071 [0.011, 0.289] FC;4;3 performance expectancy, effort expectancy, social influence, α −0.550 0.070 [−0.687, −0.412] FC;5;2 and facilitating conditions) and the predicted values by box α 0.269 0.070 [0.131, 0.403] FC;5;3 plots. The posterior mean and the observed answers of the α −0.945 0.084 [−1.109, −0.779] FC;6;2 households were found almost identical, indicating the high α 0.096 0.072 [−0.042, 0.241] FC;6;3 reproducibility of our estimated model. α −0.958 0.086 [−1.123, −0.786] FC;7;2 α 0.124 0.080 [−0.031, 0.282] FC;7;3 α −0.390 0.066 [−0.521, −0.263] FC;8;2 α 0.453 0.077 [0.301, 0.603] Appendix E. Sensitivity analysis of prior FC;8;3 distribution For sensitivity analysis of prior distributions, a normal distri- bution with a mean of 1 and variance of 1 was given for each Table C.2 Bayesian estimation results of parameters, not path coefficient between latent variables: shown in Tables 6 and Tables 7, in the structural model π ,Nð1; 1Þ; (E:1) Parameters Mean SD 95% HDI ψ 0.243 0.101 [0.079, 0.441] π ,Nð1; 1Þ; (E:2) δB ψ 0.121 0.046 [0.037, 0.209] δI γ ,Nð1; 1Þ; (E:3) ψ 0.490 0.113 [0.283, 0.715] 13 δPEI ϕ 0.705 0.176 [0.385, 1.049] γ ,Nð1; 1Þ; (E:4) ϕ 0.434 0.080 [0.288, 0.596] ϕ 0.338 0.066 [0.220, 0.471] γ ,Nð1; 1Þ; (E:5) (Continued) γ ,Nð1; 1Þ: (E:6)
Journal
Journal of Asian Architecture and Building Engineering
– Taylor & Francis
Published: Mar 4, 2023
Keywords: Zero-energy house (ZEH); technology adoption; received information; Bayesian structural equation modeling (Bayesian SEM); unified theory of acceptance and use of technology (UTAUT)
