The trigger matters: The decision-making process for heating systems in the residential building sector

Energy Policy 102 (2017) 288–306

Contents lists available at ScienceDirect

Energy Policy journal homepage: www.elsevier.com/locate/enpol

The trigger matters: The decision-making process for heating systems in the residential building sector

MARK

⁎

Maria Hechera, , Stefanie Hatzla, Christof Knoerib, Alfred Poscha a b

Institute of Systems Sciences, Innovation and Sustainability Research, University of Graz, Austria Group for Sustainability and Technology, ETH Zurich, Switzerland

A R T I C L E I N F O

A BS T RAC T

Keywords: Adoption decision Trigger type Private homeowner Residential building Fossil heating system Alternative heating system

As heat demand of buildings accounts for a significant amount of final energy use and related carbon emissions, it’s important to gain insights into the homeowners’ decision-making processes and to identify factors determining the choice of heating systems. In this study, data was collected in an online survey carried out in 2015, from private homeowners of existing and newly built single and double-family houses in Austria who had invested in a new heating system within the last ten years (N=484). In contrast to previous studies, this study specifically investigates the triggers behind homeowner decisions to invest in a new heating system (e.g. problem, opportunity, or new building situation). Results of binary logistic regression analysis show that subsidies for heating system tabinvestments and infrastructural adjustments reveal to be most effective for homeowners in problem situations to foster alternative heating systems. For homeowners in opportunity situations (e.g. building refurbishment), in addition operational convenience appears to be important. For new buildings, the main barriers for alternative heating system adoption were found in the positive perception of fuel supply security and feasibility of fossil systems. Thus, the use of trigger-specific policy measures is proposed to foster alternative heating systems in the residential building sector.

1. Introduction Building energy demand, three quarters of which is used for thermal purposes (GEA, 2012), accounts for 34% of global final energy demand. The long lifetimes of buildings and building technologies imply that immediate action needs to be taken in order to reduce energy demand and to avoid lock-in into inefficient building technologies. According to the Global Energy Assessment Report, energy demand for heating and cooling could be reduced by about 46% by 2050 compared to the 2005 levels by applying today’s best practice technologies while still more than doubling the usable floor area. In particular, end-use technologies such as heating systems hold a large potential for efficiency improvements but more so for climate mitigation (Grubler et al., 2012; Wilson et al., 2012). In the European Union, the heating sector has thus been targeted by the European Directive for Renewable Energy Directive, (2009/28/EC) and the related National Renewable Energy Actions Plan of each member state (NREAP, 2010). In Austria, almost a quarter of the final energy demand is from the residential building sector, of which more than two thirds are used for space heating (Statistics Austria, 2016, 2015). In 2014, the Austrian

energy demand for space heating accounted for 165 PJ. While 48% of the residential heat supply is still based on fossil fuels (i.e. oil, gas, electricity, and coal), 37% of the heat demand is met by biomass, heat pumps and solar-thermal systems, and 15% through district heating systems (Statistics Austria, 2016). Compared to other European countries, Austria has a relatively high penetration of such renewables (Biermayr et al., 2016; Kranzl et al., 2013). Still, one issue regarding the replacement of heating systems is the homeowners’ preference for the incumbent, and thus more familiar, type of heating system. This is especially true for those systems based on oil and gas (Kranzl et al., 2013). Recently, a growing interest of the social dimension of energy transition and the role of users emerged, putting human needs, values, preferences and behaviour at the center of system change (Rotmann, 2016; Brauch, 2013). Within this context, also research on energy prosumers emerged where consumers begin to be more proactive in areas traditionally thought of as production, while Ellsworth-Krebs and Reid (2016) suggest to broaden this concept from electricity (i.e. photovoltaic panels) to heat prosumption (e.g. wood-based stoves). To design interventions aimed at promoting behavioural change, it is

Abbreviations: P, Problem; O, Opportunity; NB, New building ⁎ Corresponding author. E-mail address: [email protected] (M. Hecher). http://dx.doi.org/10.1016/j.enpol.2016.12.004 Received 21 August 2016; Received in revised form 30 November 2016; Accepted 1 December 2016 Available online 22 December 2016 0301-4215/ © 2016 Elsevier Ltd. All rights reserved.

Energy Policy 102 (2017) 288–306

M. Hecher et al.

stage in the innovation decision process (e.g. Bjørnstad, 2012; GarcíaMaroto et al., 2015; Mahapatra and Gustavsson, 2008; Maya Sopha et al., 2011; Nyrud et al., 2008; Sopha and Klöckner, 2011), research on the earlier triggers behind homeowner decisions to invest in a new heating system, is still lacking. Finally, most studies use survey data based on hypothetical questions rather than ex-post data on real experiences. To the present authors’ knowledge, no study based on the latter approach currently exists with respect to Austria. To summarise, our study on heating system decisions contributes to research in a threefold manner: (i) by combining innovation diffusion theory and behavioural models in order to conceptualize the homeowner adoption decision; (ii) by considering the decision-making process of homeowners with respect to the triggers behind the adoption of a new heating system; and (iii) by analysing data on real adoption decisions of private homeowners in Austria. The main objective of this study is to show how the triggers underlying installation of a new heating system affect the factors determining the adoption of alternative (i.e. biomass boilers and heat pumps) and fossil heating systems (i.e. oil and gas boilers). The paper is structured as follows. In Section 2, we describe the theoretical background of this study and present the conceptual framework we apply to empirically investigate homeowner heating system adoption decisions. In Section 3, we present the methodological procedure of the survey and the empirical data analysis. In Section 4, we illustrate the results of our empirical study, elaborate on the implications, and derive possible policy measures which may be used to foster alternative heating systems in the residential building sector. In Section 5, we summarise the key results and policy implications.

thus key to gain a better understanding of the homeowner decisionmaking process regarding heating system replacements and new installations, and of those factors which foster or hinder the adoption of alternative heating systems (e.g. Braun, 2010; Lillemo et al., 2013; Mahapatra and Gustavsson, 2010; Michelsen and Madlener, 2013; Sopha et al., 2010). In recent years, an increasing number of studies have investigated factors which influence heating system adoption decisions. According to Karytsas and Theodoropoulous (2014), these studies can be classified in terms of (i) data used; (ii) variables examined; and (iii) theoretical concepts on which authors base their research. Regarding the data basis, most studies use survey data based on hypothetical questions (e.g. asking about potential future behaviour), whereas a smaller number of studies use data based on stated preferences in past decisions. In the case of the latter, two approaches are used to examine adoption decisions. One approach is to use regression analysis employing contextual variables such as socio-demographic variables (e.g. income, educational level, household size, number of children, gender, age), spatial variables (e.g. urban versus rural area, and climate), residential variables (e.g. building type, building size, construction period, ownership), and heating system characteristics (e.g. investment costs, operating cost, and physical work). The other approach additionally considers personal variables such as the influence of consumers’ attitudes, intentions, norms, and preferences for a specific type of heating technology (for a detailed list of studies, please see Balcombe et al., 2013; Karytsas and Theodoropoulou, 2014; Michelsen and Madlener, 2013). Studies using the latter approach are usually based on either theories of innovation and technology diffusion (e.g. Diffusion of innovation model; Rogers, 2003) or theories of consumer behaviour (e.g. Theory of planned behaviour; Ajzen, 1991). However, despite the number of previous studies on the subject, three outstanding issues remain. First, theories of innovation and technology diffusion need to be grounded in and combined with theories of human behaviour (Feola and Binder, 2010). To date, only relatively little research has managed to combine innovation and technology diffusion theories with behavioural models and to consider the impact of ‘personal-sphere’ elements (Michelsen and Madlener, 2013). Second, research on technology adoption has to consider the selection of heating systems as a process rather than as a fixed choice at a certain point (Rogers, 2003). Most studies investigate the factors determining adoption decisions, but do not analyse the underlying decision processes (Friege and Chappin, 2014). Although a lot of work is based on Rogers’ perceived characteristics of innovations as one

2. Conceptual model for homeowners’ decision-making processes for heating systems To operationalize the homeowner decision-making process with respect to central heating systems, we adapted a conceptual model which is based on three theoretical approaches: (i) the model of strategic decision processes established by Mintzberg et al. (1976); (ii) the five stages model in the innovation-decision process by Rogers (2003); and (iii) the integrated theoretical framework with respect to the homeowners’ decision for residential heating systems by Michelsen and Madlener (2010). The first two approaches provided a complementary basis and facilitated consideration of adoption decisions as a process, the third approach is a conceptual framework which aids the integration of innovation diffusion theory and behavioural models and

Fig. 1. Conceptual framework of homeowners' decision-making process for central heating systems (based on Michelsen and Madlener (2010), Rogers (2003), and Mintzberg et al. (1976)).

289

Energy Policy 102 (2017) 288–306

M. Hecher et al.

field of heating engineering in April 2015. Together with the insights gained from the relevant literature, the results from the interviews served as a valuable basis for designing the items used in the quantitative questionnaire. The survey was designed for both homeowners who had actually invested in a new heating system in the last ten years, and for homeowners who had not invested in a new heating system. A draft questionnaire was reviewed by the energy experts interviewed and pre-tested by private homeowners. The main survey was conducted using an online questionnaire between September and October 2015. Data was collected from private homeowners of existing and newly built single and double-family houses in Austria through Qualtrics, a professional panel survey software company (Jang et al., 2015). The final sample consists of 1006 homeowners. Of these, 560 had invested in a new heating system in the last ten years (adopters), and 446 had not invested in a new heating system (non-adopters).

focuses on heating technology decisions. All three approaches are briefly outlined in the Appendix A. Based on these two theories and the framework by Michelsen and Madlener (2010), we derived a conceptual model with three stages in order to operationalize the private homeowner decision-making process for central heating systems (see Fig. 1). Stage 1: The first stage is the need for a new system. This can either be perceived as a problem or as an opportunity by the homeowner seeking to replace an existing heating system, or it may occur in the context of new building construction. According to Mintzberg et al. (1976), problem decisions are evoked by multiple stimuli initiating a certain degree of pressure for action. Opportunity decisions tend to be also triggered by a stimulus, but are characterized by an idea rather than an unintended pressure. In both cases, the stimuli is expected to occur rather in the end of the previous heating systems’ lifetime. In our conceptual framework, we understand problem situations to be the result of a technical defect or some sort of dissatisfaction regarding the heating system. An opportunity situation is characterized as a chance or by the operation of free will and is voluntary (e.g. heating system replacement in the course of the refurbishment of a building, or in response to specific subsidies). The third trigger for starting the decision process is the need for a heating system in case of a new building. Stage 2: The need for a new system is followed by a selection phase. According to the theoretical approaches mentioned above, adoption decisions are influenced by information channels homeowners use to inform themselves about different heating technologies. The additional factors (i.e. attitude, subjective norm, and perceived behavioural control) considered, were derived from Michelsen and Madleners’ framework (2010) and are based on the theory of planned behaviour (Ajzen, 1991). (i) Attitude comprises the relative advantage of a heating system compared to other systems, including economic and environmental considerations, as well as fuel supply security; ease of use, encompassing operation requirements and fuel acquisition; compatibility with habits and norms regarding the previous heating system; and trialability, which refers to the possibility to learn from the experiences of others. (ii) Subjective norms are operationalized by their respective image, i.e. the expectations of significant others and the individual desire to use an innovative technology. (iii) Perceived behavioural control incorporates the following variables; compatibility with existing infrastructure, and voluntariness. The former takes the existing heating and building infrastructure into account (e.g. in-house heat distribution), whereas the latter measures the degree to which a homeowner perceives the adoption as a voluntary decision under the given financial circumstances. Stage 3: The homeowner decision-making process ends with a decision, namely the adoption and implementation of a certain heating system, or the rejection and delay of the investment. In addition to the above, contextual factors such as building characteristics and socio-demographic factors of homeowners are analysed as control variables that might also influence adoption decisions.

3.1.2. Operationalization In accordance with the conceptual framework, we asked the respondents to assign themselves to one of the trigger types (i.e. problem, opportunity, and new building situation). The information gathering in the selection phase was operationalized in terms of (i) information source employed and agents addressed; (ii) degree of influence of the addressed information sources; and (iii) number of recommendations received from the agents. The behavioural factors determining adoption decisions were measured in two ways: First, we asked the respondents to assess the perceived influence of 24 factors on their adoption decision (for the items please see Table B1 in the Appendix B). In addition, we asked the respondents to assess the performance of each heating technology available on the market. The performance was assessed in terms of investment costs, operational costs, environmental impact, fuel supply security, operational convenience, image, and individual feasibility. Finally, we included questions about contextual factors such as building and socio-demographic characteristics. 3.2. Empirical data analysis For our empirical data analysis, we concentrated on real preference data and considered cases of homeowners who had adopted a new heating system between 2005 and 2015, and had installed one of the four most frequent heating systems, i.e. oil boiler, gas boiler, biomass boiler, or heat pump (adopters, N=484). These four types of technologies were selected as they cover around 80% of all heating systems currently installed in Austria (Statistics Austria, 2016). The dependent variable was the type of heating system adopted, and was set dichotomous, i.e. adoption of fossil heating systems (i.e. oil or gas boilers) versus adoption of alternative heating systems (i.e. biomass boilers or heat pumps). Table 1 presents the definition of the dependent and all independent variables included in the analysis. We analysed the data for the full sample (i.e. adopters) as well as for three subsamples differentiated according to the triggers homeowners have to invest in a new heating system (i.e. problem, opportunity, and new building situation). In the course of the data analysis, we (i) assessed the data descriptively; (ii) conducted a bivariate analysis with respect to the dependent variable; and (iii) took this analysis as a basis for binary logistic regression analysis. In the first step, we explored the characteristics of the samples with respect to the dependent and independent variables in the survey. To test possible differences in the relevance of certain variables across the different subsamples, we applied Chi2 tests for categorical, and Kruskal Wallis tests for continuous independent variables. Secondly, we tested, whether the independent variables differ significantly across homeowners who adopted fossil from those who adopted alternative heating systems. For this, we applied Chi2 tests for categorical, and Mann Whitney U tests for continuous independent variables. In a last step, binary logistic regression analysis was conducted including those

3. Materials and methods 3.1. Survey procedure and operationalization In the following we briefly outline the survey and operationalization procedure. Please see the Appendix B for a comprehensive description and details. 3.1.1. Survey procedure In order to test the validity of the conceptual framework and to explore potential deficiencies in the investigated adoption decision, four guided qualitative interviews were conducted with experts in the 290

Energy Policy 102 (2017) 288–306

M. Hecher et al.

Table 1 Definition of dependent and independent variables in the analysis. Variables

Definition

Dependent variable

1 = Oil or gas boiler (fossil heating system) 2 = Biomass boiler or heat pump (alternative heating system)

Independent variables Socio-demographic and building characteristics Gender Education

Household income

Age Household size Construction period

Energy standard

Building type Previous heating system

Subsidies

Perceived performance of heating system attributes Investment costs Operational costs Environmental impact Fuel supply security Operational convenience Image Feasibility Information Information channels Influence of information channels (in cases where the information channel is not used by the respondent, the item is coded as 1 = no influence)

Recommendations

a

1 = Male 2 = Female Highest level of respondents' education 1 = Elementary school or high school 2 = University Monthly net income of household (mean value of the following categories 1 = € 1–1200, 2 = € 1201–1800, 3 = € 1801–2600, 4 = € 2601–3500, 5 = € 3501–4500, 5 = € 4501–6000, 7 = € 6001–8000, 8 = € > 8000) Respondent’s age Number of persons living in a household Construction period of buildings 1 = Before 1990 2 = From 1990 to the present Energy standard of buildings 1 = Old / unrenovated building (D, E, F, G) 2 = Building with conventional standard (C) 3 = (Ultra) low energy or passive building (A++, A+, A, B) 1 = Single-family house 2 = Double-family house 1 = Fossil fuel-based system (oil boiler, gas boiler, coal stove, electric heating) 2 = Renewable-based system (biomass boiler, heat pump, solar-thermal system, district heating) 3 = No previous heating system Use of subsidies from the municipality, federal state or national level 0 = No 1 = Yes Performance of oil and gas boiler very cheap, 4 = very expensive) Performance of oil and gas boiler very cheap, 4 = very expensive) Performance of oil and gas boiler very low, 4 = very high) Performance of oil and gas boiler very unsecure, 4 = very secure) Performance of oil and gas boiler very low, 4 = very high) Performance of oil and gas boiler very good, 4 = very bad) Performance of oil and gas boiler very good, 4 = very bad)

/ biomass boiler and heat pump (4-point Likert scale: 1 = / biomass boiler and heat pump (4-point Likert scale: 1 = / biomass boiler and heat pump (4-point Likert scale: 1 = / biomass boiler and heat pump (4-point Likert scale: 1 = / biomass boiler and heat pump (4-point Likert scale: 1 = / biomass boiler and heat pump (4-point Likert scale: 1 = / biomass boiler and heat pump (4-point Likert scale: 1 =

Number of information channels used by respondent to inform himself/herself about the different heating technologies Social network (1 = no influence, 5 = high influence) Internet (1 = no influence, 5 = high influence) Specialized information (1 = no influence, 5 = high influence) Installer (1 = no influence, 5 = high influence) Chimney sweep (1 = no influence, 5 = high influence) Energy consultant (1 = no influence, 5 = high influence) Number of recommendations received from information channelse for oil and gas boilers / biomass boilers and heat pumps Number of recommendations received from information channelsa for biomass boilers and heat pumps

sum of recommendations from social network, installer, chimney sweep, and energy consultant (multiple answer).

independent variables that were - according to the bivariate analysis significantly related to one of the adoption options.1 However, all variables related to information gathering were excluded from the

analysis due to the limited number of predictor variables in logistic regression and the relatively low number of cases in the subsamples. The results of the logistic regression models were then compared to each other in order to show that the homeowner adoption decision might be determined by different factors depending on the triggers underlying investment in a new heating system. The statistical analysis was performed using the software IBM SPSS 22 (SPSS software, version 22).

1 In the logistic regression analysis, the responses for the perceived performance of heating system attributes were used as independent variables in preference to the perceived influence of certain factors for adoption decisions. This was done due to their higher explanatory power with respect to the choice of alternative heating systems in the logistic regression models.

291

Energy Policy 102 (2017) 288–306

M. Hecher et al.

Table 2 Adopted heating system (dependent variable) for full sample and subsamples. Full sample

Adopted heating system Oil and gas Biomass and heat pump Total (N)

218 (45.0%) 266 (55.0%) 484 (100%)

Chi2 test

Subsample Problem

Opportunity

New Building

88 (62.9%) 52 (37.1%) 140 (100%)

93 (43.9%)

37 (28.0%)

119 (56.1%)

95 (72.0%)

212 (100%)

132 (100%)

Chi2 = 33.500, df=2, p < 0.01***

NB sample. To figure out whether our sample is representative of all private homeowners of single and double-family houses in Austria, we compared specific building characteristics of the full sample with Austrian census data in terms of (i) the number of single and doublefamily houses for each federal state; (ii) the building construction period; and (iii) the energy carriers of heating technologies (please see Table C3 in Appendix C).

4. Results and discussion 4.1. Sample description Table 2 shows the dependent variable (i.e. adoption of heating system) for the full sample and subsamples. Overall 45% of the homeowners adopted oil and gas boilers, whereas 55% adopted biomass boilers and heat pumps. 29% of the homeowners considered themselves to have acted in a problem situation (P), 44% in an opportunity situation (O), and 27% in a new building situation (NB). Problem-triggered homeowners installed significantly more fossilbased heating systems (63%) than homeowners who experienced an opportunity situation (44%) or homeowners with a new building did (28%). The highest frequency of alternative heating systems was observed in new buildings. In the following we briefly describe the main independent variables. In the Appendix C, the descriptive statistics for all samples are provided in Tables C1 and C2. Regarding socio-demographics, the full sample is dominated by male homeowners (60%) and homeowners with an elementary or high school education (74%). The respondents have a mean net monthly household income of €3,180, on average about four persons live in a household, and the respondents have an average age of 42 years, while the age significantly decreases from the P sample to the O and NB sample. Considering the building characteristics, about half of the buildings were constructed after 1990. Most new buildings (90%) are considered as (ultra) low energy or passive buildings, while for renovated buildings a conventional energy standard (about 43%) was most frequent. The majority of all buildings (66%) are single-family houses and between 66% in the O sample and 75% in the P sample had a fossil-based heating system as their previous heating system. As expected, construction period and energy standards was significantly lower for the NB sample compared to the P and O sample. With regard to the behavioural factors, the use of subsidies when investing in a new heating system significantly decreased from 66% in the P sample to 41% in the NB sample. Factors such as investment costs, fuel supply security, image, and feasibility were generally evaluated as being higher for alternative than for fossil heating systems. Conversely, operational costs, environmental impact, and operational convenience received a higher rating for fossil than for alternative heating systems. This difference was more pronounced for the P and O sample than for homeowners of new buildings. Regarding information, on average around two information channels were consulted. The relative levels of influence are found to be highest for the installer, followed by the social network, internet, and specialized information, while the influence of chimney sweeps and energy consultants is lowest. Finally, results show that homeowners received on average less than one recommendation to install a fossilbased system, and more than one recommendation to install an alternative system. The number of recommendations for fossil systems however was significantly decreasing from the P sample to the O and

4.2. Bivariate analysis Bivariate analysis did provide initial insights into the impact specific factors had on heating system adoption decisions. In this subsection however, we solely discuss the results regarding the information variables, as these could not be included in the logistic regression analysis. The result tables of the entire bivariate analysis are illustrated in the Appendix D (please see Tables D1 and D2 for the full sample as well as Tables D3 and D4 for the subsamples). Overall (i.e. for the full sample), the most frequently addressed information sources were found to be installers, which is in line with comparable studies that identify these as playing the most important role as change agents (Mahapatra and Gustavsson, 2008; Sopha et al., 2011). Homeowners who chose alternative heating systems consult a significantly higher number of information channels and more actively searched for information than homeowners who adopted fossil systems. Consequently, homeowners with alternative heating systems are more influenced by specialized information, the internet, and their social networks, whereas homeowners who chose fossil systems are more influenced by installers. When asking homeowners about the reasons why specialized information was important for their decision, the most frequently stated were (i) the possibility to compare relevant attributes across specific technologies; and (ii) the chance to see and touch heating systems at trade fairs. With regard to the internet and social network, they mentioned the possibility of gaining access to a wealth of experience and objective opinion (in contrast, for example, to gaining access to the opinion of installers who clearly want to sell the system). In addition, different information sources seem to be more or less relevant for adoption decisions, depending on homeowners’ heating investment triggers (i.e. for the subsamples). Thus, in order to foster alternative heating systems group-specific information strategies might be more successful than general information strategies. For homeowners in a P or O situation, it could be most effective to provide specialized information which enables homeowners to efficiently compare the relevant attributes of all possible technology options. For homeowners in an O situation, in addition, internet platforms and discussion forums, where experiences and opinions about heating systems can constructively be discussed, could be useful. As also suggested by Mahapatra and Gustavsson (2008), installers could be trained specifically to inform homeowners who wish to replace their heating system. Other studies also demonstrate that different informa292

Energy Policy 102 (2017) 288–306

M. Hecher et al.

choice of alternative heating systems for the full sample (the same results but a different form of presentation to that given in Table 3). The previous heating system has a significant influence on deciding in favour of alternative over fossil heating systems. For homeowners installing their first heating system (i.e. new building situation) and homeowners replacing a heating system which is already based on renewable energy carriers (i.e. biomass boiler, heat pump, solarthermal system, and district heating), the likelihood of choosing alternative systems is higher compared to those who are replacing fossil heating systems (i.e. oil and gas boiler, coal stove, and electric heating). A reason that homeowners of new buildings install alternative heating systems - especially heat pumps - is that as in other European countries (e.g. Bleicher and Gross, 2016; Sopha et al., 2010; Bjørnstad, 2012) Austria financially supports the diffusion of renewable heating systems due to energy policy targets (Biermayr et al., 2016). Thus, the installation rate of heat pumps have increased during the last years and makes it common mainly in the new building sector (Decker and Menrad, 2015; Forsén, 2008; Biermayr et al., 2016). In consequence, homeowners of new buildings are influenced by their social network and tend to choose a social comparison decision strategy due to subjective norms (Sopha et al., 2010). For homeowners who replace their old heating system, the previous heating infrastructure has the strongest impact on the probability of choosing alternative heating systems. This is likely due to homeowners’ satisfaction with the old type of system as well as to the influence of daily habits and routines (Claudy et al., 2011; Rouvinen and Matero, 2013). This is in line with the results of Michelsen and Madlener (2012) who show that homeowners with oil and gas boilers are more likely to choose the same heating system, and that preferences for familiar technologies decrease the probability of switching to a new heating technology (Michelsen and Madlener, 2016). The perceived operational costs are found to be the most important heating system attribute influencing the adoption decision. Homeowners who view the operational costs of oil and gas boilers as being higher are more likely to choose alternative heating systems. This is also found in other studies, e.g. Sopha and colleagues (2011) found that operational costs and expected increases in fuel prices determine the adoption of wood pellet heating. Achtnicht and Madlener (2014) identified present and expected future costs as the most frequently stated driver for heating system replacements, while Lillemo et al. (2013) conclude that heating system investments are mainly driven by considerations regarding heating cost reductions. Bleicher and Gross (2016) show that the intention to lower heating costs can trigger the installation of a heat pump even if the heating costs are higher in the end. The probability of deciding in favour of alternative heating systems increases significantly for households who use subsidies from the municipality, federal state, or national level. Other studies also indicate that due to the economic utility of investment subsidies, it is rational for households to make use of financial support (Bjørnstad, 2012) and thus facilitate the diffusion of heating technologies (e.g. Mahapatra and Gustavsson, 2009; Sardianou and Genoudi, 2013). However, in contrast to the above, Sopha et al. (2011) found that in Norway, only a small share of adopters indicate the presence of subsidies as being a driver behind wood pellet heating, although they do indicate that this may be due to the relatively low funding amount offered compared to the total investment costs incurred. This shows that willingness to pay must be taken into account when designing financial support for renewable heating systems (Scarpa and Willis, 2010). Homeowners’ perceived feasibility of biomass boilers and heat pumps is also significant in increasing the probability of choosing such technologies. Homeowners who evaluate their infrastructural preconditions as being more suitable for alternative heating systems are more likely to install such systems. The existing infrastructure of a house can be more or less compatible with a specific technology and thus may be perceived as a functional barrier (Claudy et al., 2015, 2013; Michelsen

tion sources influence heating system adoption decisions (Michelsen and Madlener, 2013; Nyrud et al., 2008; Scarpa and Willis, 2010). However, in addition to the findings of previous studies, it now appears worthwhile to tailor possible homeowner information strategies so as to match the different triggers underlying the decision to invest in a new heating system. Regarding the recommendations for heating systems from various agents, the results show that the chance of being recommended to invest in an alternative heating system is higher in cases where professional energy consultants or people from the social network are contacted for advice than in cases where installers and chimney sweeps are consulted. Considering the fact that energy consultants are one of the least frequently contacted information sources by all groups of homeowners, but negatively affect the adoption of fossil systems (Michelsen and Madlener, 2012), measures to facilitate the use of their services might be a promising approach. 4.3. Logistic regression Based on the bivariate analysis, we then selected the variables for inclusion in the logistic regression models, while checking that the following prerequisites were satisfied. First, we checked the assumption of chi-square for each independent categorical variable. The expected frequencies in each cell of the cross tables are greater than 1 and no more than 20% are less than 5. This assures that sufficient information is available from these predictors (Field, 2009). Second, multicollinearity was tested for all continuous independent variables. By quite some margin, the variance inflation factor never exceeded the value of 10, thus indicating that there need be no cause for concern (Backhaus et al., 2011; Schendera, 2008; Field, 2009). And finally, for all continuous predictors, we checked the linearity of the logit. For each variable, we created a new variable that is the log of the original variable and included its interaction term in the regression model. No single interaction term was significant, indicating that the assumption of linearity of the logit was met for all variables (Schendera, 2008; Field, 2009). The relevant parameters and tests needed to assess the model fit and the contribution of the predictors included in the model are shown and explained in the Appendix E in Table E1. The results for each of the regression models are presented in Table 3, together with the respective logistic regression coefficients (B), standard deviations (SD), Wald test statistics with degrees of freedom (df), significance levels (Sig.), odds ratios, and confidence intervals for each of the included predictor variables. In the full sample, the significant factors are (i) the previous heating system and subsidies; (ii) perceived operational costs for oil and gas; and (iii) perceived feasibility and operational convenience of biomass boilers and heat pumps. These factors increase the chance of adopting alternative heating systems. The likelihood of adopting such systems decreases with (i) age of homeowners; (ii) perceived feasibility, image, and fuel supply security of oil and gas boilers; and (iii) perceived environmental impact of biomass boilers and heat pumps. For the subsamples, the determining factors are characterized differently, especially in terms of the importance of factors, and also because certain factors are not significant or only significant in certain subsamples. In the next two subsections, we discuss these results and elaborate on the implications with respect to the main objective of our study. We first focus on the contextual and behavioural factors which tend to push homeowners towards alternative over fossil systems for the full sample, and compare these with results from previous studies (subsection 4.3.1). Only in a second step, we discuss the influencing factors for each of the respective homeowner trigger groups (i.e. the problem, opportunity, and new building situations) (subsection 4.3.2). 4.3.1. Influencing factors for the choice of alternative heating systems (full sample) Table 4 shows a ranked overview of the significant predictors for the 293

Energy Policy 102 (2017) 288–306

M. Hecher et al.

Table 3 Logistic regression for the choice of alternative over fossil heating systems for full sample and subsamples. B

SD

Wald

df

Sig.

Odds ratio

95% Confidence interval

Alternative heating system (biomass boiler and heat pump) vs. fossil heating system (oil and gas boiler) Full sample Construction period (0= < 1990, 1=≥1990) Previous heating system Fossil-based (0) vs. renewable-based (1) Fossil-based (0) vs. no heating system (1) Subsidies (0=no, 1=yes) Age Operational costs: oil and gas Operational costs: biomass and heat pump Environmental impact: oil and gas Environmental impact: biomass and heat pump Fuel supply security: oil and gas Fuel supply security: biomass and heat pump Operational convenience: biomass and heat pump Image: oil and gas Image: biomass and heat pump Feasibility: oil and gas Feasibility: biomass and heat pump Subsample: Problem Previous heating system (0=fossil, 1= renewable) Subsidies (0=no, 1=yes) Age Operational costs: oil and gas Operational costs: biomass and heat pump Environmental impact: oil and gas Environmental impact: biomass and heat pump Fuel supply security: oil and gas Image: oil and gas Image: biomass and heat pump Feasibility: oil and gas Feasibility: biomass and heat pump Subsample: Opportunity Previous heating system (0=fossil, 1=renewable) Subsidies (0=no, 1=yes) Age Investment costs: oil and gas Investment costs: biomass and heat pump Operational costs: oil and gas Environmental impact: oil and gas Fuel supply security: oil and gas Operational convenience: biomass and heat pump Image: oil and gas Feasibility: oil and gas Feasibility: biomass and heat pump Subsample: New Building Subsidies (0 = no, 1 = yes) Investment costs: oil and gas Operational costs: oil and gas Operational costs: biomass and heat pump Environmental impact: oil and gas Fuel supply security: oil and gas Fuel supply security: biomass and heat pump Image: oil and gas Image: biomass and heat pump Feasibility: oil and gas Feasibility: biomass and heat pump

−0.088

0.404

1.569 1.766 1.410 −0.026 1.574 −0.394 0.129 −0.548 −0.467 −0.285 1.319 −0.727 −0.560 −1.453 1.353

1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

p=0.828 p < 0.01*** p < 0.01*** p < 0.01*** p < 0.01*** p < 0.05** p < 0.01*** p=0.181 p=0.688 p < 0.1* p < 0.1* p=0.470 p < 0.01*** p < 0.05** p=0.143 p < 0.01*** p < 0.01***

0.916

0.415

2.021

0.440 0.492 0.333 0.013 0.319 0.295 0.323 0.325 0.271 0.394 0.312 0.298 0.383 0.293 0.296

0.047 21.047 12.709 12.896 17.960 4.066 24.342 1.793 0.161 2.841 2.974 0.523 17.868 5.944 2.143 24.517 20.896

4.802 5.847 4.095 0.974 4.827 0.674 1.138 0.578 0.627 0.752 3.738 0.483 0.571 0.234 3.870

2.027 2.230 2.133 0.949 2.583 0.378 0.605 0.306 0.368 0.347 2.028 0.270 0.270 0.132 2.166

11.377 15.330 7.859 0.999 9.021 1.201 2.143 1.093 1.066 1.628 6.890 0.867 1.209 0.416 6.912

2.277 2.559 −0.022 1.354 0.704 −0.427 −0.596 −0.437 −3.042 0.622 −2.006 2.438

0.864 0.944 0.030 0.643 0.645 0.842 0.824 0.593 1.072 0.865 0.737 0.767

6.944 7.349 0.568 4.426 1.189 0.257 0.523 0.543 8.058 0.517 7.399 10.111

1 1 1 1 1 1 1 1 1 1 1 1

p < 0.01*** p < 0.01*** p=0.451 p < 0.05** p=0.275 p=0.612 p=0.470 p=0.461 p < 0.01*** p=0.472 p < 0.01*** p < 0.01***

9.746 12.919 0.978 3.871 2.021 0.653 0.551 0.646 0.048 1.863 0.135 11.449

1.792 2.032 0.923 1.097 0.571 0.125 0.110 0.202 0.006 0.342 0.032 2.548

53.002 82.153 1.036 13.663 7.160 3.401 2.772 2.064 0.390 10.161 0.571 51.452

1.359 1.504 −0.031 −0.104 −0.712 1.629 0.329 −0.250 1.497 −0.742 −1.230 1.050

0.547 0.488 0.019 0.410 0.394 0.475 0.453 0.345 0.425 0.411 0.439 0.437

6.177 9.501 2.696 0.064 3.272 11.764 0.528 0.525 12.417 3.269 7.858 5.780

1 1 1 1 1 1 1 1 1 1 1 1

p < 0.05** p < 0.01*** p=0.101 p=0.800 p < 0.1* p < 0.01*** p=0.467 p=0.469 p < 0.01*** p < 0.1* p < 0.01*** p < 0.05**

3.894 4.501 0.970 0.902 0.491 5.097 1.390 0.779 4.469 0.476 0.292 2.858

1.333 1.729 0.935 0.404 0.227 2.010 0.572 0.396 1.943 0.213 0.124 1.214

11.374 11.714 1.006 2.013 1.061 12.925 3.380 1.532 10.277 1.064 0.691 6.728

0.177 −2.083 1.807 −0.433 0.388 −1.173 −0.011 0.939 0.795 −2.159 1.892

0.627 0.667 0.734 0.720 0.586 0.564 0.765 0.629 0.791 0.660 0.750

0.080 9.758 6.056 0.361 0.437 4.329 0.000 2.225 1.010 10.707 6.368

1 1 1 1 1 1 1 1 1 1 1

p= 0.777 p < 0.01*** p < 0.05** p=0.548 p=0.509 p < 0.05** p=0.989 p=0.136 p=0.315 p < 0.01*** p < 0.05**

1.194 0.125 6.093 0.649 1.473 0.310 0.989 2.556 2.214 0.115 6.633

0.349 0.034 1.445 0.158 0.467 0.103 0.221 0.745 0.470 0.032 1.526

4.083 0.460 25.701 2.659 4.650 0.934 4.431 8.774 10.437 0.421 28.836

(2013), user convenience (e.g. reduction in time and effort), is a significant factor in choosing pellet boilers and heat pumps. In contrast, the findings of Michelsen and Madlener (2012) are more mixed. They show that while a preference for greater comfort increases the probability of adopting heat pumps and gas boilers, it decreases the adoption of pellet and oil boilers. Another study by Michelsen and Madlener (2016) also found that a preference for high comfort levels increases the chance of sticking with oil and gas boilers rather than switching to pellet boilers or heat pumps. Our analysis shows that alternative heating systems receive a lower ranking in terms of operational convenience than do fossil systems. This assessment is also in line with Lillemo et al. (2013) and Nyrud et al. (2008) who

and Madlener, 2016). Also Sopha et al. (2011) found that technical barriers or difficulties in the course of refitting the house (e.g. fuel storage space or chimney) are one of the most important barriers in the installation of wood pellet stoves. Referring to heat pumps, Bleicher and Gross (2016, 2015) show that although there are still uncertainties in their technical feasibility and environmental impacts, homeowners find strategies to cope with these uncertainties, e.g. by considering oneself as a ‘pioneer’ but also due to socio-cultural values regarding the acceptance of technologies and social norms. A higher estimation of the operational convenience with regard to biomass boilers and heat pumps increases the probability of choosing alternative technologies. As indicated by Lillemo and colleagues 294

Energy Policy 102 (2017) 288–306

M. Hecher et al.

4.3.2. Importance of the specific factors for the choice of alternative heating systems (subsamples) Table 5 illustrates that for problem-triggered homeowners, subsidies are important as they are most likely to increase the probability of adopting alternative heating systems, followed by the technical feasibility of alternative systems, the previous heating system, and the operational costs of fossil heating systems. The barriers of choosing alternative systems are found to be the image and feasibility of fossil systems. For homeowners who replace their heating system in a problem situation, financial support for heating systems, as well as the necessary infrastructural adjustments which go along with such investments (e.g. fuel storage room, façade insulation), seem to be crucial increasing the uptake of biomass boilers and heat pumps. Within this group, only one third of the homeowners make use of subsidies (Table C1). Thus, a special focus in this respect should be placed on homeowners who suffer from heating system failure or functional unreliability. It is important to raise awareness and facilitate access to financial incentives, as these are the most stated reasons for not using existing subsidies. Bureaucratic obstacles have to be removed considering the fact that homeowners within this group are significantly older than those with new buildings (Table C2) and may have less time to inform themselves about subsidy availability and application procedures. The results also show that providing problem-triggered homeowners with specific information with regard to the following three aspects may raise policy effectiveness. First, providing information concerning the infrastructural requirements and necessary modifications for alternative heating systems in order to raise perceived feasibility. Second, providing information concerning expected operational costs over the whole life cycle of all technology options (Stolarski et al., 2016). This is likely to be particularly effective for homeowners with fossil systems as these tend to focus more on the economic aspects of heating systems (Decker et al., 2009). Third, providing information on the advantages of renewable over fossil energy carriers (e.g. price stability, fuel supply security, environmental impacts) might prove successful in lowering the relatively favourable image of fossil systems in this group compared to that of the O and NB group (Table C2). Last, but no less important, it would seem crucial to reach homeowners who intend to replace their heating system as early as possible in the decision process, since it was found that those who consider themselves to be triggered by an opportunity are more likely to install alternative heating systems. For homeowners in an opportunity situation, the factors raising the probability of installing alternative systems are found to be the same as those for the problem-triggered group above, however the importance of these factors is ranked differently (Table 5). In addition, only for this group of homeowners, does a higher assessment of the operational convenience of biomass boilers and heat pumps increase the probability of adopting alternative systems. Besides the barriers found for the P group, higher estimates of the investment costs of alternative heating systems are also related to a decrease in the likelihood of adopting such systems (10% significance level). Based on these results, we conclude that for homeowners in an opportunity situation, the economic aspects of the heating investment also play the most important role. However, the perceived operational costs for oil and gas have a higher impact on the likelihood of adopting alternative heating systems than subsidies lowering investment costs. Nevertheless, besides (i) offering financial support for renewable heating systems and infrastructural adjustments; and (ii) providing specific information on heating system operational costs, feasibility of alternative systems, and advantages of renewable energy carriers, a further promising policy approach in Austria would seem to be (iii) promoting the operating convenience of heat pumps and modern

Table 4 Ranked overview of significant predictors from logistic regression analysis for the choice of alternative over fossil systems for full sample. Full sample

Ranked odds ratio

Previous HS (0 = fossil, 1= no HS) Previous HS (0 = fossil, 1 = renewable) Operational costs: oil and gas Subsidies (0 = no, 1 = yes) Feasibility: biomass and heat pump Operational convenience: biomass and heat pump Feasibility: oil and gas Image: oil and gas Environmental impact: biomass and heat pump Fuel supply security: oil and gas Age

5.847*** 4.802*** 4.827*** 4.095*** 3.870*** 3.738*** 0.234*** 0.483** 0.578* 0.627* 0.974**

HS: heating system; *** p < 0.01, ** p < 0.05, * p < 0.1.

identified wood stoves as being particularly poor in terms of their perceived operational convenience. In contrast, heat pumps have been rated as being rather good in terms of their operational convenience (Decker et al., 2009; Lillemo et al., 2013). In contrast to the above mentioned results, the feasibility of oil and gas boilers was found to be a significant and the most important barrier in the adoption of alternative systems. Those who assessed the feasibility of oil and gas boilers as being high, were less likely to choose alternative systems. This is in line with the other studies discussed above. The image of oil and gas boilers is another heating system attribute found to negatively affect the adoption of alternative systems. Biomass boilers and heat pumps are less likely to be chosen by respondents who rank oil and gas boilers rather highly. In our study, ‘image’ is understood to be the desire to use an innovative technology and to meet the expectations of others (i.e. the desire to conform to a social norm). Claudy et al. (2015) also found that social norms regarding heating systems can be considered as a psychological barrier in adoption decisions, while, in contrast, Michelsen and Madlener (2012) were not able to show that image affects the adoption decision. Our results further indicate that the perceived environmental impact of biomass boilers and heat pumps and fuel supply security of oil and gas boilers both have a significant effect on the adoption decision (10% significance level). The probability of choosing alternative systems decreases for homeowners who assess the environmental impact of these systems more highly, and for homeowners who view the fuel supply security of fossil systems as being more secure. In other studies, while both factors are found to be a significant predictor, they tend to play a minor role compared to other factors (Mahapatra and Gustavsson, 2010, 2008; Sopha et al., 2010). For example, Mahapatra and Gustavvson (2008) show that besides investment costs, environmental benignity and security of fuel supply are weak predictors motivating homeowner choice with respect to district heating, heat pumps, and wood pellet boilers. Finally, the results of the full sample show that the likelihood of choosing alternative technologies decreases with an increase in respondents’ age. This finding corresponds with results from Sopha et al. (2010) as well as with those from Mahapatra and Gustavsson (2008) who both show that older homeowners are less likely to choose renewable heating technologies. In contrast to this, Karytsas and Theodoropoulou (2014) found that heat pumps tend to be installed by homeowners in higher age groups. This is explained by the fact that older people often have more money to invest.

295

Energy Policy 102 (2017) 288–306

M. Hecher et al.

Table 5 Ranked overview of significant predictors from logistic regression analysis for the choice of alternative over fossil systems for subsamples. Ranked odds ratios for subsamples Problem Subsidies (0 = no, 1 = yes) Feasibility: biomass and heat pump Previous HS (0 = fossil, 1 = renewable) Operational costs: oil and gas Image: oil and gas Feasibility: oil and gas

Opportunity 12.919*** 11.449*** 9.746*** 3.871** 0.048*** 0.135***

New Building

Operational costs: oil and gas Subsidies (0 = no, 1 = yes) Operational convenience: biomass and heat pump Previous HS (0 = fossil, 1 = renewable) Feasibility: biomass and heat pump Feasibility: oil and gas Investment costs: biomass and heat pump Image: oil and gas

5.097*** 4.501*** 4.469*** 3.894** 2.858** 0.292*** 0.491* 0.476*

Feasibility: biomass and heat pump Operational costs: oil and gas Fuel supply security: oil and gas Feasibility: oil and gas Investment costs: oil and gas

6.633** 6.093** 0.310** 0.115*** 0.125***

HS: heating system; *** p < 0.01, ** p < 0.05, * p < 0.1.

reduce the validity of results (Sopha and Klöckner, 2011). As a rationale for avoiding potential regret, homeowners’ retrospective assessment of heating system characteristics may tend to produce an evaluation in favour of the adopted heating technology (Rogers, 2003). In addition, homeowner choice of heating technology can be up to ten years old, and this may introduce some element of distortion with regard to their original decision making. However, since investing in a new heating system is a cost-intensive decision, and one which is only rarely taken, one can assume that to a large extent homeowners are able to adequately reconstruct their decision process. The second limitation is related to the conceptual framework. This only focuses on the decision for heating technologies and does not reflect possible related decisions such as building retrofits or the construction of new buildings (Michelsen and Madlener, 2013). Third, although our sample was shown to be well-balanced in terms of selected building characteristics based on Austrian census data, it has to be noted that the concrete data for our population (i.e. private homeowners of single and double-family houses who actually decided for a new heating system) is not available. A statement on the representativity of our sample is thus limited which has to be taken into consideration by interpreting the results. Finally, the results of this study are limited to Austria and have to be interpreted in the light of the prevailing market situation (e.g. fuel prices, investment costs, available technologies, etc).

biomass boilers (Büchner et al., 2015). With regard to the barriers in the heating decision, governmental grants to lower the perceived investment costs for renewable heating systems and infrastructural adjustments might have a positive impact. This is also in line with the finding that subsidies are a supporting factor in this group. Concerning information, (i) the access to internet platforms and discussion forums enabling homeowners to describe and share their experiences, (ii) specific training for installers wishing to support homeowners within this group, as well as (iii) raising awareness of and support for energy consultancy services, would seem to be useful tools in fostering alternative heating systems within the old building stock. For homeowners in a new building situation, it is striking that the technical feasibility of alternative heating systems is more important than economic aspects, and that subsidies are not found to be significant. This is also the only group where higher evaluation with respect to the fuel supply security of oil and gas boilers decreases the probability of alternative systems. Another influencing factor (which could not be shown to be significant for the other samples) is the perceived investment costs for oil and gas boilers. Surprisingly, for those homeowners who evaluate the investment costs of fossil systems more highly than others, the likelihood of adopting alternative systems decreases. Compared to homeowners in a problem or opportunity situation, those with new buildings have the highest adoption rate of alternative heating systems (see Table 2). The focus should thus be set on the remaining one third of homeowners who still decide to install technologies based on fossil fuels. Our results reveal that a policy of informing them about the equal technical feasibility and the beneficial operational costs of renewable energy technologies is likely to be most effective. Information strategies also need to confront the barriers such homeowners face in installing alternative technologies. These are found to reside in the high estimation homeowners place on fuel supply security and on the feasibility of fossil fuels. Higher subsidies for biomass boilers and heat pumps, which would improve their relative cost advantages over fossil systems, do not seem to be a promising approach in the new building sector.

5. Conclusion and policy implications The purpose of this study was to investigate the factors determining private homeowner adoption decisions with respect to alternative heating systems (i.e. biomass boilers and heat pumps) in Austria. The main contribution of this study lies in its analysis of this decision from a process perspective and in its incorporation of homeowner heating investment triggers. The results indicate that not only the choice of heating systems, but also the information channels and determining factors all vary depending on whether the homeowner’s need for a new system is perceived as a (i) problem situation; (ii) opportunity situation; or (iii) arises due to the construction of a new building. Overall, it is crucial to reach homeowners early enough in the decision process in order to avoid the emergence of problem situations. For homeowners who find themselves in such a situation, the most effective policy appears to be the provision of financial support for heating system investments and in particular, the financial support of the associated infrastructural adjustments, in order to strengthen the

4.4. Limitations of the study Although this study is the first (to our knowledge) to consider the initial situation – the trigger - as a starting point for homeowner heating technology adoption decisions, several limitations have to be highlighted. The first concerns the ex-post survey design which may

296

Energy Policy 102 (2017) 288–306

M. Hecher et al.

of the decision making process, looking at the real preferences relating to heating system attributes before and after the adoption decision would further improve the validity of results. And finally, research based on innovation and technology diffusion theories as well as on theories of human behaviour has the potential to be used for agentbased modelling for ex-ante policy analysis, and would benefit from the integration of empirical knowledge (Knoeri et al., 2011; Michelsen and Madlener, 2016).

perceived feasibility of alternative systems. For homeowners in opportunity situations, economic aspects are found to be important, but so are (i) the operational convenience of heating systems, (ii) access to internet platforms and discussion forums, and (iii) energy consultancy services. In the new building sector, about two thirds of the homeowners already install alternative heating systems. However, the remaining barriers that need to be addressed were found to lie in the high evaluation placed on both fuel supply security and on the feasibility of fossil systems. Based on the results and limitations of this study, we propose the following issues for further research. First, it is essential to identify private homeowners who intend to invest in a new heating system, as this is the prerequisite for reasonable and trigger-specific information strategies and policy measures. Second, further data analysis, which investigates the differences between adopters and non-adopters with regard to the key factors of adoption decisions, might give insights into the first issue suggested for future research. Third, a longitudinal study

Acknowledgements This work was supported by the federal government of Styria (Land Steiermark - Erneuerbare Energien und Klimaschutz) and in the form of a scholarship granted by the Faculty of Environmental, Regional and Educational Sciences at the University of Graz in Austria. The authors thank Sebastian Seebauer and Eva Fleiss for their support and constructive comments.

Appendix A: Theoretical background Mintzberg et al. (1976) model of strategic decision processes identifies three phases. (i) The identification phase consists of a decision recognition routine including the perception of a crisis, problem or opportunity as starting points for decisions, and a diagnostic routine which includes the consideration of existing and new information channels. (ii) The development phase consists of the search and design routine. This phase comprises the solution search mechanism and encompasses different types of search behaviour ranging from the use of passive and familiar information sources to those entailing more active and professional activity. (iii) The selection phase consists of the screening routine (i.e. determination of choice criteria), evaluation-choice routine (i.e. evaluation of consequences regarding all alternatives in terms of criteria and authorization routine (i.e. making a choice of acceptance or rejection) (Mintzberg et al., 1976). Rogers’ (2003) innovation-decision process comprises five stages through which an individual decision maker passes: (i) knowledge about the existence and function of the innovation; (ii) persuasion, i.e. the formation of an attitude towards the innovation with respect to its relative advantage, compatibility, complexity, trialability, and observability (i.e. perceived characteristics of innovations); (iii) decision, encompassing individual action regarding adoption or rejection of innovation; (iv) implementation of the innovation; and (v) confirmation of the final decision. A prerequisite for the initiation of a decision-making process are prior conditions such as a perceived need or problem. Information gathering via different communication channels is used to guide progress through the various decision-making stages. These channels become more and more specific during the decision process (e.g. moving from mass media to interpersonal communication channels) (Rogers, 2003). The theoretical framework developed by Michelsen and Madlener (2010) is the result of a comprehensive review of decision-making models with a view towards extracting general lessons for adoption decisions and their application in residential heating systems. They use the theory of planned behaviour (Ajzen, 1991) which considers behaviour as a result of intentions influenced by attitudes, social norms and perceived behavioural control, and integrate this with the perceived characteristics of innovations as stated by Rogers (2003). Accordingly, attitudes towards a specific heating system are a construct of the variables relative advantage, ease of use, trialability, result demonstrability, and compatibility with habits and norms. Subjective norms, defined as the perceived social pressure concerning the given behaviour, comprise the variables image and visibility. Finally, perceived behavioural control describes the perceived degree of control over the adoption decision, and is operationalized through the construct voluntariness and compatibility with existing infrastructure. Besides these ‘personal-sphere’ determinants, Michelsen and Madlener (2010) also consider perceived economic and non-economic external factors (e.g. energy prices, technology availability, etc.) when explaining the intention to adopt a certain heating system. Appendix B: Survey procedure In order to test the validity of the conceptual framework and to explore potential deficiencies in the investigated adoption decision, four guided qualitative interviews were conducted with experts in the field of heating engineering in April 2015. The interviewees were Austrian experts who had carried out the ‘Heizungs-Check 2014′, a comprehensive investigation of private homeowners’ heat generation, distribution, and transfer systems funded by the Austrian national government from the Austrian Climate and Energy Fund. This pilot action and measure financially supports efficiency measures (e.g. replacement of heating pumps, pipe insulations, and installation of thermostatic heads) and heating system replacements towards environment and climate friendly systems. In total, the investigation entailed about 400 heating inspections carried out by four organizations in different federal states in Austria. One expert was interviewed from each organization, in total covering more than half of the inspections conducted. The interviews were recorded, transcribed, and analysed. The interview material was structured and reduced by means of a category system deductively established based on the theoretical background considered in this study and inductively extended based on the interview material (Mayring, 2010). The qualitative content analysis was conducted using the software MAXQDA (MAXQDA software, version 12). A draft questionnaire was reviewed by the energy experts interviewed and pre-tested by private homeowners. For the pre-test, we selected two homeowners for each of the following three categories (i) had recently refurbished their heating system; (ii) had recently built a new house and

297

Energy Policy 102 (2017) 288–306

M. Hecher et al.

Table B1 Items measuring perceived influence of factors for adoption decision and performance of heating system attributes for each of the addressed heating systems based on conceptual framework. Conceptual framework

Attitude

Relative advantage

Economic factors

Perceived influence of factors for adoption decision (5-point Likert scale: 1 = no influence, 5 = very high influence)

Performance of heating system attributes(4-point Likert scale plus category ‘don’t know’)

Investment costs

Investment costs (1 = very cheap, 4 = very expensive) Operational costs (1 = very cheap, 4 = very expensive)

Annual operational / fuel costs

Ecological factors

Fuel supply security

Ease of use

Subjective norm Perceived behavioural control

Compatibility with habits and norms Trialability Image Compatibility with infrastructure

Voluntariness

Future expected fuel costs Subsidies Environmental and climate protection aspects Use of renewable fuels / energy carriers Use of domestic fuels Use of own fuels (e.g. for forest owners) Independency from fossil fuels Independence from grid connection (e.g. district heating grid, gas grid, heat pump) Operational effort for heating system Effort connected with fuel procurement Experiences with previous heating system Experiences of other persons with specific heating system Expectations of others Use of innovative and modern technology Existing heating infrastructure Disposal of previous heating system Possibility of fuel storage Possibility to connect to a gas or district heating grid Constructional effort Use of building (primary or secondary residence) Future use of building Financial situation

Environmental impact (1 = very low, 4 = very high) Fuel supply security (1 = very unsecure, 4 = very secure)

Operational convenience (1 = very low, 4 = very high)

Image (1 = very good, 4 = very bad) Feasibility (1 = very good, 4 = very bad)

therefore had invested in a new heating system; and homeowners who (iii) had not invested in a new heating system within the last ten years. This was done in order to detect possible misinterpretations of the questions and items, to test filter questions, and to ascertain the duration of the survey. In a final step, we revised the questionnaire guided by the feedback received from both experts and pre-test participants. The main survey was conducted using an online questionnaire between September and October 2015. Data was collected from private homeowners of existing and newly built single and double-family houses in Austria. Homeowners of single and double-family houses were chosen as the target group because they have the authority to choose heating systems independently, and because it is assumed that a single person is mainly responsible for the adoption decision. This contrasts with the decision process in multi-family houses, and with tenants in rented accommodation who are usually not fully included in the decision process. An online survey was chosen as the appropriate survey instrument, mainly because we aimed to obtain data covering a large variety of installed technologies. Another effective means of reaching private homeowners would have been to approach them via funding agencies. However, as we also intended to reach those homeowners who had installed oil and gas boilers and who had not benefited from national subsidy schemes, this approach was ruled out. In order to obtain a well-balanced sample with 1,000 private homeowners, we commissioned Qualtrics, a professional panel survey software company (Jang et al., 2015). From their panel, they addressed private homeowners of single and double-family houses in Austria who were or would be mainly involved in the decision-making process of a heating system. Regarding homeowners’ spatial distribution, a preliminary quota was set reflecting the number of buildings in the respective federal states in Austria. The survey was sent to 40,900 persons, started by 7,560 persons (18.5%), and filled in completely by 1,006 persons (2.5%). From the 6,554 respondents who started but did not complete the questionnaire, 64% terminated the questionnaire, 26% did not belong to the target group, 6% were dismissed because the quota was already fulfilled, and 5% of the persons did not meet the quality checks. Quality checks are individual questions or screeners used to ensure that respondents are actually reading and responding thoughtfully to the questions in the survey. We included two attention filters to reduce ‘straight-liners’ and ‘speeders’, which look like the following: “This is an attention filter. Please select ‘strongly disagree’ for this statement.” Additionally, we included a survey duration check, which means that respondents who took the survey in less than 1/3 the average duration were excluded from the survey .

298

Energy Policy 102 (2017) 288–306

M. Hecher et al.

Table C1 Profiles of respondents for full sample and subsamples for categorical independent variables. Full sample

Gender (N=484) Male Female Education (N=484) Elementary and high school University Construction period (N=465) Before 1990 From 1990 to the present Energy standard (N=335) Old / unrenovated building Building with conventional (Ultra) low energy / passive Building type (N=484) Single-family house Double-family house Previous heating system Fossil fuel based system Alternative system No previous heating system Subsidies (N=454) No Yes

Subsample

Chi² test

Problem

Opportunity

New Building

288 (59.5%) 196 (40.5%)

88 (62.9%) 52 (37.1%)

133 (62.7%) 79 (37.3%)

67 (50.8%) 65 (49.2%)

Chi² = 5.763, df = 2, p = .056

359 (74.2%) 125 (25.8%)

99 (70.7%) 41 (29.3%)

158 (74.5%) 54 (25.5%)

102 (77.3%) 30 (22.7%)

Chi² = 1.550, df = 2, p = .461

254 (54.6%) 211 (45.4%)

103 (75.7%) 33 (24.3%)

151 (72.9%) 56 (27.1%)

0 (0.0%) 122 (100.0%)

Chi² = 199.357, df = 2, p < .01***

56 (16.7%) 106 (31.6%) 173 (51.6%)

27 (31.8%) 37 (43.5%) 21 (24.7%)

29 (21.5%) 57 (42.2%) 49 (36.3%)

0 (0.0%) 12 (10.4%) 103 (89.6%)

Chi² = 107.615, df = 4, p < .01***

317 (65.5%) 167 (34.5%)

91 (65.0%) 49 (35.0%)

118 (55.7%) 94 (44.3%)

108 (81.8%) 24 (18.2%)

Chi² = 24.652, df = 2, p < .01***

237 (50.2%) 103 (21.8%) 132 (28.0%)

101 (74.8%) 34 (25.2%) 0 (0.0%)

136 (66.3%) 69 (33.7%) 0 (0.0%)

0 (0.0%) 0 (0.0%) 132 (100.0%)

Chi² = 475.842, df = 4, p < .01***

240 (52.9%) 214 (47.1%)

87 (65.9%) 45 (34.1%)

103 (51.2%) 98 (48.8%)

50 (41.3%) 71 (58.7%)

Chi² = 15.695, df = 2, p < .01***

Table C2 Profiles of respondents for full sample and subsamples for continuous independent variables. Full sample

Subsample Problem

Household size (N=479) Household income (N=372)a Age (N=483) Perceived performance of …Investment costs Oil and gas (N=466) Biomass and heat pump (N=460) Operational costs Oil and gas (N=468) Biomass and heat pump (N=463) Environmental impact Oil and gas (N=465) Biomass and heat pump (N=462) Fuel supply security Oil and gas (N=470) Biomass and heat pump (N=467) Operational convenience Oil and gas (N=462) Biomass and heat pump (N=466) Image Oil and gas (N=469) Biomass and heat pump (N=465) Feasibility Oil and gas (N=460) Biomass and heat pump (N=459) Information channels (N=484) Influence of social network (N=483) Influence of internet (N=484) Influence of specialized information (N=484) Influence of installer (N=484) Influence of chimney sweep (N=484) Influence of energy consultant (N=484) Recommendations for oil and gas (N=456) Recommendations for biomass and heat pump (N=456) a

Kruskal Wallis test (df = 2) Opportunity

New Building

Mean

SD

Mean

SD

Mean

SD

Mean

SD

3.75 3,180 42.25

1.45 1,542 12.08

3.66 3,101 45.38

1.61 1,488 12.71

3.71 3,157 42.42

1.45 1,688 12.50

3.92 3,290 38.61

1.24 1,373 9.51

Chi² = 5.473, p = .065 Chi² = 2.149, p = .342 Chi² = 21.823, p < .01***

2.64 2.77

0.72 0.63

2.68 2.80

0.61 0.66

2.76 2.74

0.71 0.65

2.38 2.78

0.78 0.57

Chi² = 20.806, p < .01*** Chi² = 0.284, p = .868

2.99 2.13

0.60 0.59

2.95 2.11

0.61 0.66

2.98 2.14

0.62 0.59

3.07 2.12

0.54 0.52

Chi² = 2.601, p = .272 Chi² = 0.458, p = .795

3.08 2.02

0.62 0.59

3.02 2.05

0.59 0.64

3.05 1.99

0.63 0.57

3.19 2.02

0.64 0.57

Chi² = 8.308, p < .05** Chi² = 0.990, p = .610

2.41 3.38

0.71 0.52

2.41 3.30

0.65 0.51

2.46 3.45

0.75 0.49

2.31 3.36

0.69 0.55

Chi² = 2.878, p = .237 Chi² = 7.848, p < .05**

3.12 2.91

0.70 0.59

3.14 2.88

0.68 0.59

3.15 2.97

0.68 0.57

3.06 2.85

0.74 0.60

Chi² = 0.916, p = .633 Chi² = 3.639, p = .162

2.28 3.25

0.65 0.52

2.39 3.19

0.63 0.57

2.34 3.28

0.65 0.51

2.08 3.25

0.61 0.47

Chi² = 17.511, p < .01*** Chi² = 2.445, p = .295

2.64 2.79 2.35 2.39 2.28 2.18 3.18 1.43 1.50 0.68 1.22

0.74 0.66 1.32 1.60 1.46 1.57 1.58 1.05 1.17 0.92 1.38

2.64 2.69 2.32 2.41 2.14 1.93 3.37 1.61 1.39 0.98 1.10

0.70 0.79 1.23 1.59 1.44 1.52 1.47 1.17 1.04 1.06 1.37

2.67 2.81 2.39 2.33 2.37 2.27 3.06 1.43 1.57 0.67 1.23

0.74 0.63 1.28 1.62 1.49 1.57 1.59 1.06 1.24 0.90 1.44

2.60 2.85 2.30 2.46 2.27 2.30 3.18 1.25 1.51 0.38 1.34

0.76 0.55 1.49 1.58 1.44 1.61 1.67 0.87 1.18 0.66 1.30

Chi² Chi² Chi² Chi² Chi² Chi² Chi² Chi² Chi² Chi² Chi²

mean category (for categories see Table 1 with definition of variables).

299

= = = = = = = = = = =

1.243, p = .537 3.314, p = .191 0.351, p = .839 0.501, p = .778 2.265, p = .322 5.342, p < .1* 2.845, p = .241 9.939, p < .01*** 2.046, p = .360 29.349, p < .01*** 4.544, p = .103

Energy Policy 102 (2017) 288–306

M. Hecher et al.

Table C3 Comparison of full sample data regarding building characteristics (N=1006) with Austrian census data. Sample (N=1006)

Austrian census data

Number of single and double-family houses in the federal states of Austriab Burgenland 85 9%

Deviation from census data

91,692

6%

3%

Carinthia 81 9% Lower Austria 172 19% Upper Austria 163 18% Salzburg 62 7% Styria 164 18% Tyrol 70 8% Vorarlberg 35 4% Vienna 68 8% Construction period of single and double-family houses in Austriab before 1919 44 5%

132,063 439,779 319,319 93,117 262,712 123,248 74,728 54,752

8% 28% 20% 6% 17% 8% 5% 3%

1% −9% −2% 1% 1% 0% −1% 5%

192,037

12%

−7%

1919 to 1944 43 5% 1945 to 1960 85 10% 1961 to 1970 73 8% 1971 to 1980 119 14% 1981 to 1990 128 15% 1991 to 2000 155 18% 2001 and later 237 27% Energy carriers of heating technologies installed in the residential building sectorb Coal 5 1%

99,571 176,817 221,401 261,299 241,002 199,227 200,056

6% 11% 14% 16% 15% 13% 13%

−1% −1% −6% −2% 0% 5% 14%

63,931

1%

0%

1,231,919 315,005 760,520 1,094,920 1,047,857 3,745,552 451,743 267,171

14% 4% 9% 12% 12% 42% 5% 3%

1% 8% 10% 18% −6% −37% −3% 7%

Wood logs Wood pellets and wood chips Oil Gas District heating Electricity Solar-thermal systems Heat pumpsa

151 118 193 302 60 48 21 104

15% 12% 19% 30% 6% 5% 2% 10%

Chi2 test

Chi2 = 182.137, df = 8, p = 0.000***

Chi2 = 196.246, df = 7, p = 0.000***

Chi2 = 999.424, df = 8, p = 0.000***

a STATcube Statistics Austria Database: Register 2011 - Austrian census on buildings and housing (selection settings: dwelling with main residence; federal state or construction period according to building type; residential buildings with one and two dwellings, and private property owner). b Statistics Austria Energy Statistics: Energy use of households 2013/2014.

Appendix C: Sample description Table C3 shows a comparison of data from the full sample with Austrian census data in terms of (i) the number of single and double-family houses for each federal state; (ii) the building construction period; and (iii) the energy carriers of heating technologies. For the first two characteristics, the comparison shows that the deviations are not greater than 10%, although new buildings (i.e. construction period 2001 and later) are slightly overrepresented. With respect to the energy carriers, the deviations are greater for oil and gas (overrepresented) as well as electricity (underrepresented). However, this can be explained by the fact that Austrian census data is based on the whole residential building sector including multi-family houses. Table C3 shows that there is a significant difference between full sample data and Austrian census data which would lead to the assumption that our sample is not respresentative, however it has to be noted that the concrete data for our population (i.e. private homeowners of single and double-family houses who actually decided for a new heating system) is not available. A statement on the representativity of our sample is thus limited.

300

Energy Policy 102 (2017) 288–306

M. Hecher et al.

Appendix D: Bivariate analysis

Table D1 Dependent variable with respect to independent categorical variables for full sample.

Gender (N=484) Male Female Education (N=484) Elementary and high school University Construction period (N=465) Before 1990 From 1990 to the present Energy standard (N=335) Old / unrenovated building Building with conventional standard (Ultra) low energy / passive building Building type (N=484) Single-family house Double-family house Previous heating system (N=472) Fossil fuel-based system Renewable-based system No previous heating system Subsidies (N=454) No Yes

Oil and gas

Biomass and heat pump

Chi2 test

61.0% 39.0%

58.3% 41.7%

Chi2=0.373, df = 1, p= 0.541

72.0% 28.0%

75.9% 24.1%

Chi2 = 0.962, df = 1, p = 0.327

63.1% 36.9%

48.1% 51.9%

Chi2 = 10.331, df = 1, p < 0.01***

19.9% 39.0% 41.1%

14.4% 26.3% 59.3%

Chi2 = 10.817, df = 2, p < 0.01***

65.1% 34.9%

65.8% 34.2%

Chi2=0.023, df= 1, p= 0.881

71.9% 10.5% 17.6%

32.8% 30.9% 36.3%

Chi2 = 72.256, df = 2, p < 0.01***

65.9% 34.1%

42.2% 57.8%

Chi2=25.312, df=1, p < 0.01***

Table D2 Dependent variable with respect to independent continuous variables for full sample. Oil and gas

Household size (N=479) Household income (N=372)a Age (N=483) Perceived performance of … Investment costs Oil and gas (N=466) Biomass and heat pump (N=460) Operational costs Oil and gas (N=468) Biomass and heat pump (N=463) Environmental impact Oil and gas (N=465) Biomass and heat pump (N=462) Fuel supply security Oil and gas (N=470) Biomass and heat pump (N=467) Operational convenience Oil and gas (N=462) Biomass and heat pump (N=466) Image Oil and gas (N=469) Biomass and heat pump (N=465) Feasibility Oil and gas (N=460) Biomass and heat pump (N=459) Information channels (N=484) Influence of social network (N=483) Influence of internet (N=484) Influence of specialized information (N=484) Influence of installer (N=484) Influence of chimney sweep (N=484) Influence of energy consultant (N=484) Recommendations for oil and gas (N=456) Recommendations for biomass and heat pump (N=456) a

Biomass and heat pump

Mann Whitney U test

Mean

SD

Mean

SD

3.65 3184 44.75

1.49 1539 12.13

3.84 3177 40.20

1.41 1548 11.66

Z=−1.626, p=0.104 Z=−0.173, p=0.863 Z=−4.232, p < 0.01***

2.62 2.80

0.56 0.67

2.65 2.74

0.83 0.60

Z= −0.527, p=0.598 Z= −1.249, p=0.212

2.74 2.20

0.56 0.65

3.20 2.07

0.55 0.54

Z=−8.512, p < 0.01*** Z=−2.104, p < 0.05**

2.91 2.10

0.57 0.61

3.22 1.95

0.63 0.57

Z= −6.099, p < 0.01*** Z = −2.788, p < 0.01***

2.65 3.31

0.64 0.57

2.20 3.45

0.70 0.46

Z= −6.986, p < 0.01*** Z = −2.475, p < 0.05**

3.14 2.81

0.70 0.61

3.10 2.98

0.70 0.55

Z= −0.641, p=0.521 Z = −2.999, p < 0.01***

2.54 3.14

0.60 0.58

2.08 3.34

0.61 0.45

Z= −8.128, p < 0.01*** Z=−3.683, p < 0.01***

2.89 2.55 2.22 2.24 2.11 1.88 3.33 1.51 1.40 1.23 0.55

0.57 0.70 1.31 1.57 1.40 1.46 1.54 1.12 1.04 0.90 1.08

2.44 2.97 2.45 2.52 2.42 2.43 3.06 1.37 1.58 0.21 1.80

0.79 0.57 1.32 1.62 1.50 1.62 1.60 0.99 1.25 0.63 1.35

Z = −6.369, p < 0.01*** Z= −6.115, p < 0.01*** Z=−2.119, p < 0.05** Z=−1.851, p < 0.1* Z=−2.264, p < 0.05** Z=−3.903, p < 0.01*** Z=−1.886, p < 0.1* Z=−1.645, p=0.100 Z = −1.561, p=0.118 Z−13.721, p < 0.01*** Z=−11.722, p < 0.01***

mean category (for categories see Table 1 with definition of variables).

301

302

Yes

Renewable-based system No previous heating system Subsidies (N=454) No

Double-family house Previous heating system (N=472) Fossil fuel-based system

Building with conventional standard (Ultra) low energy / passive building Building type (N=484) Single-family house

From 1990 to the present Energy standard (N=335) Old / unrenovated building

University Construction period (N=465) Before 1990

Female Education (N=484) Elementary and high school

Gender (N=484) Male

Subsamples

46.9%

53.1%

73.5%

26.5%

52.0% –

9.4% –

48.0%

90.6%

61.5%

67.0% 38.5%

17.2%

28.6%

33.0%

41.4%

41.4%

26.8%

44.6%

17.6%

82.4%

71.8%

28.2%

26.9%

73.1%

69.3%

30.7%

40.4%

59.6%

Biomass and heat pump

35.2%

64.8%

Oil and gas

Problem

Triggers

Chi2 = 5.725, df = 1, p < 0.05**

Chi2 = 30.304, df = 1, p < 0.01***

Chi2 = 0.436, df = 1, p = 0.509

Chi2 = 2.320, df = 2, p = 0.313

Chi = 1.945, df = 1, p = 0.163

2

Chi2 = 0.223, df = 1, p = 0.637

Chi2 = 0.372, df = 1, p = 0.542

Chi2 test

Table D3 Dependent variable with respect to independent categorical variables for subsamples.

37.4%

62.6%

15.9% –

84.1%

40.9%

59.1%

32.1%

43.4%

24.5%

24.7%

75.3%

26.9%

73.1%

39.8%

60.2%

Oil and gas

58.2%

41.8%

47.0% –

53.0%

47.1%

52.9%

39.0%

41.5%

19.5%

28.8%

71.2%

24.4%

75.6%

35.3%

64.7%

Biomass and heat pump

Opportunity

Chi2 = 8.640, df = 1, p < 0.01***

Chi2 = 21.754, df = 1, p < 0.01***

Chi2 = 0.813, df = 1, p = 0.367

Chi2 = 0.834, df = 2, p = 0.659

Chi = 0.431, df = 1, p = 0.512

2

Chi2 = 0.174, df = 1, p = 0.677

Chi2 = 0.450, df = 1, p = 0.502

Chi2 test

– 100.0%

– 100.0%

45.2%

63.3%

36.7%

–

–

54.8%

15.8%

84.2%

94.0%

24.3%

75.7%

78.1%

6.0%

–

– 21.9%

100.0%

–

– 100.0%

22.1%

77.9%

50.5%

49.5%

Biomass and heat pump

24.3%

75.7%

45.9%

54.1%

Oil and gas

New Building

Chi2 = 3.140, df = 1, p < 0.1*

–

Chi2 = 1.304, df = 1, p = 0.253

Chi2 = 6.209, df = 2, p < 0.05**

–

Chi2 = 0.075, df = 1, p = 0.785

Chi2 = 0.224, df = 1, p = 0.636

Chi2 test

M. Hecher et al.

Energy Policy 102 (2017) 288–306

2.62

Perceived performance of … Investment costs Oil and gas (N=466)

303

2.48

2.22

2.31

2.15

Information channels (N=484)

Influence of social network (N=483)

Influence of internet (N=484)

2.86

3.08

2.62

2.81

3.19

3.24

2.61

2.15

2.85

2.22

2.78

Biomass and heat pump (N=459)

Feasibility Oil and gas (N=460)

Biomass and heat pump (N=465)

Image Oil and gas (N=469)

Biomass and heat pump (N=466)

Operational convenience Oil and gas (N=462)

Biomass and heat pump (N=467)

Fuel supply security Oil and gas (N=470)

Biomass and heat pump (N=462)

Environmental impact Oil and gas (N=465)

Biomass and heat pump (N=463)

Operational costs Oil and gas (N=468)

2.83

47.35

Age (N=483)

Biomass and heat pump (N=460)

3.53 3304

1.45

1.56

1.23

0.81

0.52

0.60

0.57

0.61

0.65

0.56

0.56

0.59

0.56

0.72

0.54

0.63

0.56

11.75

1.63 1578

2.12

2.60

2.50

3.05

2.23

3.39

2.00

2.99

3.04

3.40

2.06

1.90

3.29

1.94

3.25

2.75

2.79

42.04

3.88 2734

1.44

1.64

1.23

0.63

0.81

0.46

0.54

0.55

0.71

0.41

0.66

0.68

0.53

0.53

0.62

0.71

0.68

13.67

1.57 1,245

SD

Mean

Mean

SD

Biomass and heat pump

Oil and gas

Problem

Triggers

Household size (N=479) Household income (N=372)a

Subsamples

Table D4 Dependent variable with respect to independent continuous variables for subsamples.

Z = −4.750, 0.01*** Z = −3.825, 0.01*** Z = −1.359, 0.174 Z = −0.992, 0.321 Z = −0.166, 0.868

p=

p=

p=

p <

p <

Z = −5.424, p < 0.01*** Z = −2.290, p < 0.01***

Z = −1.258, p = 0.209 Z = −1.558, p = 0.119

Z = −4.499, p < 0.01*** Z = −1.335, p =0.182

Z = −4.178, p < 0.01*** Z = −2.542, p < 0.05**

Z = −4.693, p < 0.01*** Z = −2.134, p < 0.05**

Z=−0.378, p=0.706

Z=−1.341, p=0.180

Z=−1.168, p=0.243 Z= −1.845, p < 0.1* Z=−2.466, p < 0.05**

Mann Whitney U test

2.05

2.23

2.30

2.58

2.89

3.21

2.58

2.85

3.10

3.41

2.71

2.05

2.92

2.16

2.68

2.82

2.62

44.32

3.68 2990

Mean

1.37

1.60

1.33

0.64

0.61

0.56

0.62

0.58

0.72

0.55

0.70

0.58

0.55

0.62

0.61

0.69

0.57

12.40

1.41 1435

SD

Oil and gas

Opportunity

2.62

2.41

2.46

2.98

2.50

3.34

2.15

3.06

3.18

3.49

2.26

1.94

3.16

2.13

3.20

2.68

2.88

40.94

3.73 3273

Mean

1.55

1.64

1.24

0.56

0.79

0.46

0.62

0.55

0.65

0.43

0.73

0.56

0.66

0.57

0.54

0.62

0.78

12.43

1.49 1843

SD

Biomass and heat pump

Z = −3.516, 0.01*** Z = −4.094, 0.01*** Z = −1.014, 0.310 Z = −0.826, 0.409 Z = −2.702, 0.01***

p <

p=

p=

p <

p <

Z = −5.100, p < 0.01*** Z = −1.637, p = 0.102

Z = −0.784, p = 0.433 Z = −2.671, p < 0.01***

Z = −4.393, p < 0.01*** Z = −0.941, p =0.347

Z = −3.319, p < 0.01*** Z = −1.560, p = 0.119

Z = −5.998, p < 0.01*** Z = −0.353, p = 0.724

Z=−2.770, p < 0.01*** Z=−1.744, p < 0.1*

Z=−0.121, p=0.904 Z=−0.495, p= 0.621 Z= −1.939, p < 0.1*

Mann Whitney U test

2.14

2.11

2.00

2.66

2.94

3.10

2.24

2.72

3.13

3.21

2.58

2.13

3.01

2.26

2.82

2.69

2.63

39.50

3.84 3319

Mean

1.38

1.56

1.49

0.55

0.63

0.54

0.55

0.72

0.75

0.60

0.62

0.71

0.65

0.59

0.51

0.72

0.52

10.72

1.32 1663

SD

Oil and gas

New building

2.33

2.60

2.42

2.92

2.47

3.31

2.02

2.89

3.03

3.42

2.21

1.98

3.26

2.06

3.16

2.81

2.29

38.27

3.96 3279

Mean

1.46

1.58

1.48

0.54

0.77

0.43

0.63

0.55

0.73

0.51

0.70

0.50

0.63

0.48

0.53

0.51

0.84

9.05

1.21 ,252

SD

Biomass and heat pump

Z = −3.371, p < 0.01*** Z = −1.928, p < 0.1* Z = −1.492, p = 0.136 Z = −1.580, p = 0.114 Z = −0.694, p = 0.488 (continued on next page)

Z = −2.130, p < 0.05** Z = −2.029, p < 0.05**

Z = −0.643, p = 0.520 Z = −0.971, p = 0.332

Z = −3.058, p < 0.01*** Z = −1.796, p < 0.1*

Z = −2.220, p < 0.01*** Z = −1.007, p = 0.314

Z = −3.482, p < 0.01*** Z = −1.721, p < 0.1*

Z=−2.426, p < 0.01*** Z=−0.519, p = 0.604

Z=−0.818, p=0.413

Z=−0.732, p=0.464 Z=−0.136, p=0.892

Mann Whitney U test

M. Hecher et al.

Energy Policy 102 (2017) 288–306

a

1.48

3.48

1.63

1.32

1.33

0.49

Influence of chimney sweep (N=484)

Influence of energy consultant (N=484)

Recommendations for oil and gas (N=456)

Recommendations for biomass and heat pump (N=456)

2.17

0.38

1.50

1.58

3.19

2.35

mean category (for categories see Table 1 with definition of variables).

0.85

0.89

0.98

1.21

1.36

1.68

1.46

1.06

1.15

1.11

1.44

1.69

SD

Mean

Mean

SD

Biomass and heat pump

Oil and gas

Problem

Triggers

Influence of specialized information (N=484) Influence of installer (N=484)

Subsamples

Table D4 (continued)

Z = −2.586, p < 0.05** Z = −1.431, p = 0.152 Z = −0.050, p = 0.960 Z = −1.154, p = 0.249 Z = −6.539, p < 0.01*** Z=−7.035, p < 0.01***

Mann Whitney U test

0.59

1.27

1.47

1.45

3.35

1.96

Mean

1.28

0.93

1.12

1.03

1.51

1.47

SD

Oil and gas

Opportunity

1.76

0.17

1.64

1.41

2.82

2.52

Mean

1.36

0.47

1.32

1.08

1.62

1.61

SD

Biomass and heat pump

Z = −2.564, p < 0.01*** Z = −2.311, p < 0.01*** Z = −0.789, p = 0.430 Z = −0.842, p = 0.400 Z = −9.486, p < 0.01*** Z=−7.701, p < 0.01***

Mann Whitney U test

0.57

0.89

1.41

1.38

2.95

2.14

Mean

1.01

0.81

1.01

1.11

1.72

1.64

SD

Oil and gas

New building

1.65

0.17

1.55

1.20

3.27

2.37

Mean

1.27

0.46

1.24

0.75

1.65

1.60

SD

Biomass and heat pump

Z = −0.729, p = 0.466 Z = −0.964, p = 0.335 Z = −0.714, p = 0.475 Z = −0.353, p = 0.724 Z = −5.786, p < 0.01*** Z=−5.020, p < 0.01***

Mann Whitney U test

M. Hecher et al.

Energy Policy 102 (2017) 288–306

304

Energy Policy 102 (2017) 288–306

M. Hecher et al.

Appendix E: Logistic regression

Table E1 Assessment of logistic regression model and contribution of predictors for full sample and subsamples. Full sample

Number of included cases Number of missing casesa Total number of cases Classification accuracyb Omnibus tests of model coefficientsc Cox & Snell (Pseudo R²)d Nagelkerke (Pseudo R²) Hosmer-Lemeshow teste

382 102 484 58.4% vs. 85.3% Chi² = 248.257, df = 16, p < .01*** .478 .643 Chi² = 7.211, df = 8, p = .514

Subsample Problem 109 31 140 62.4% vs. 88.1% Chi² = 88.036, df = 12, p < .01*** .554 .755 Chi² = 8.324, df = 8, p = .402

Opportunity

New Building

169 43 212 57.4% vs. 82.8% Chi² = 99.066, df = 12, p < .01***

113 19 132 74.3% vs. 87.6% Chi² = 53.640, df = 11, p < .01*** .378 .556 Chi² = 1.997 df = 8, p = .981

.444 .596 Chi² = 6.781, df = 8, p = .560

a

Logistic regression excludes all cases with missing values for any of the independent variables. Percentage of cases classified correctly in the baseline model using only the constant in the regression equation vs. the model including predictors (percentages < 80% are not acceptable). c A significant model chi² statistic indicates that the model including the predictors is significantly better than the baseline model. d Pseudo R² measures provide information on how well the dependent variable is predicted by the independent variables (Cox & Snell is acceptable from > .02 and good from > .04, Nagelkerke is acceptable from > .02, good from > .04, and very good from > .05). e Goodness of fit tests provide information on how well the predicted and observed frequencies match, a non-significant p-value indicates that the model fits the data well (Schendera, 2008; Field, 2009; Backhaus et al., 2011). b

References

2003/30/EC. Ellsworth-Krebs, K., Reid, L., 2016. Conceptualising energy prosumption: exploring energy production, consumption and microgeneration in Scotland, UK. Environ. Plan A. 48 (10), 279–288. http://dx.doi.org/10.1016/j.rser.2016.06.013. Feola, G., Binder, C.R., 2010. Towards an improved understanding of farmers' behaviour: the integrative agent-centred (IAC) framework. Ecol. Econ. 69, 2323–2333. http:// dx.doi.org/10.1016/j.ecolecon.2010.07.023. Field, A., 2009. Discovering statistics using SPSS Third edition. SAGE Publications, London, England. Forsén M., Market status for ground source heat pumps in Europe, 2008, Report Ground-Reach Project Deliverable 2: Stockholm, Sweden. Friege, J., Chappin, E., 2014. Modelling decisions on energy-efficient renovations: a review. Renew. Sustain. Energy Rev. 39, 196–208. http://dx.doi.org/10.1016/ j.rser.2014.07.091. García-Maroto, I., García-Maraver, A., Muñoz-Leiva, F., Zamorano, M., 2015. Consumer knowledge, information sources used and predisposition towards the adoption of wood pellets in domestic heating systems. Renew. Sustain. Energy Rev. 43, 207–215. http://dx.doi.org/10.1016/j.rser.2014.11.004. GEA, 2012. Global Energy Assessment - Toward a Sustainable Future, Key Findings, Summary for Policymakers and Technical Summary. Cambridge University Press, Cambridge UK and New York NY, (USA and the International Institute for Applied Systems Analysis, Laxenburg, Austria). Grubler, A., et al., 2012. Policies for the Energy Technology Innovation System (ETIS). In: Johansson, T.B. (Ed.), Global Energy Assessment: Towards a Sustainable Future. Cambridge University Press, Cambridge, 1665–1743. Jang, Y.J., Kim, W.G., Lee, H.Y., 2015. Coffee shop consumers' emotional attachment and loyalty to green stores: the moderating role of green consciousness. Int. J. Hosp. Manag. 44, 146–156. http://dx.doi.org/10.1016/j.ijhm.2014.10.001. Karytsas, S., Theodoropoulou, H., 2014. Public awareness and willingness to adopt ground source heat pumps for domestic heating and cooling. Renew. Sustain. Energy Rev. 34, 49–57. http://dx.doi.org/10.1016/j.rser.2014.02.008. Knoeri, C., Binder, C.R., Althaus, H.-J., 2011. An Agent Operationalization Approach for Context Specific Agent-Based Modelin. J. Artif. Soc. Soc. Simul. 14. Kranzl, L., Hummel, M., Müller, A., Steinbach, J., 2013. Renewable heating: perspectives and the impact of policy instruments. Energy Policy 59, 44–58. http://dx.doi.org/ 10.1016/j.enpol.2013.03.050. Lillemo, S.C., Alfnes, F., Halvorsen, B., Wik, M., 2013. Households' heating investments: the effect of motives and attitudes on choice of equipment. Biomass-. Bioenergy 57, 4–12. http://dx.doi.org/10.1016/j.biombioe.2013.01.027. Mahapatra, K., Gustavsson, L., 2010. Adoption of innovative heating systems-needs and attitudes of Swedish homeowners. Energy Effic. 3, 1–18. http://dx.doi.org/10.1007/ s12053-009-9057-7. Mahapatra, K., Gustavsson, L., 2009. Influencing Swedish homeowners to adopt district heating system. Appl. Energy 86, 144–154. http://dx.doi.org/10.1016/ j.apenergy.2008.03.011. Mahapatra, K., Gustavsson, L., 2008. An adopter-centric approach to analyze the diffusion patterns of innovative residential heating systems in Sweden. Energy Policy 36, 577–590. http://dx.doi.org/10.1016/j.enpol.2007.10.006. Maya Sopha, B., Klöckner, C. a., Hertwich, E.G., 2011. Exploring policy options for a transition to sustainable heating system diffusion using an agent-based simulation. Energy Policy 39, 2722–2729. http://dx.doi.org/10.1016/j.enpol.2011.02.041. Mayring, P., 2010. Qualitative Inhaltsanalyse: Grundlage und Techniken (11. aktual. und überarb. Auflage). Weinheim und Basel. Beltz, Verlag. MAXQDA Software. Software for Qualitative Data Analysis1989–2016, VERBI Software – Consult – Sozialforschung GmbH, Berlin, Germany.

Achtnicht, M., Madlener, R., 2014. Factors influencing German house owners' preferences on energy retrofits. Energy Policy 68, 254–263. http://dx.doi.org/ 10.1016/j.enpol.2014.01.006. Ajzen, I., 1991. The theory of planned behavior. Orgnizational Behav. Hum. Decis. Process 50, 179–211. http://dx.doi.org/10.1016/0749-5978(91)90020-T. Backhaus, K., Erichson, B., Plinke, W., Weiber, R., 2011. Multivariate Analysemethoden: Eine anwendungsorientierte Einführung. Springer Verlag, Berlin, Heidelberg. Balcombe, P., Rigby, D., Azapagic, A., 2013. Motivations and barriers associated with adopting microgeneration energy technologies in the UK. Renew. Sustain. Energy Rev. 22, 655–666. http://dx.doi.org/10.1016/j.rser.2013.02.012. Biermayr, P., Eberl, M., Enigl, M., Fechner, H., Kristöfel, C., Leonhartsberger, K., Maringer, F., Moidl, S., Schmidl, C., Strasser, C., Weiss, W., Wopienka, E., 2016. Innovative Energietechnologien in Österreich - Marktentwicklung 2015. Reports on energy and environmental research 6/2016: Vienna, Austria. Bjørnstad, E., 2012. Diffusion of renewable heating technologies in households. Experiences from the Norwegian Household Subsidy Programme. Energy Policy 48, 148–158. http://dx.doi.org/10.1016/j.enpol.2012.04.078. Bleicher, A., Gross, M., 2015. User motivation, energy prosumers, and regional diversity: sociological notes on using shallow geothermal energy. Geotherm. Energy 3 (12), 1–12. http://dx.doi.org/10.1186/s40517-015-0032-6. Bleicher, A., Gross, M., 2016. Geothermal heat pumps and the vagaries of subterranean geology: energy independence at a household level as a real world experiment. Renew. Sustain Energy Rev. 64, 279–288. http://dx.doi.org/10.1016/ j.rser.2016.06.013. Brauch, H.G., 2013. Seven Dimensions of ‘Sustainable Transition’: Temporal, Spatial, Scientific, Societal, Economic, Political and Cultural, International Studies Association’s 54th Annual Convention, San Francisco, 2-6 April 2013. Braun, F.G., 2010. Determinants of households' space heating type: a discrete choice analysis for German households. Energy Policy 38, 5493–5503. http://dx.doi.org/ 10.1016/j.enpol.2010.04.002. Büchner, D., Schraube, C., Carlon, E., von Sonntag, J., Schwarz, M., Verma, V.K., Ortwein, A., 2015. Survey of modern pellet boilers in Austria and Germany: System design and customer satisfaction of residential installations. Applied Energy 160, 390–403. http://dx.doi.org/10.1016/j.apenergy.2015.09.055. Claudy, M.C., Garcia, R., O’Driscoll, A., 2015. Consumer resistance to innovation—a behavioral reasoning perspective. J. Acad. Mark. Sci. 43, 528–544. http:// dx.doi.org/10.1007/s11747-014-0399-0. Claudy, M.C., Michelsen, C., O’Driscoll, A., 2011. The diffusion of microgeneration technologies - assessing the influence of perceived product characteristics on home owners' willingness to pay. Energy Policy 39, 1459–1469. http://dx.doi.org/ 10.1016/j.enpol.2010.12.018. Claudy, M.C., Peterson, M., O’Driscoll, a., 2013. Understanding the Attitude-Behavior Gap for Renewable Energy Systems Using Behavioral Reasoning Theory. J. Macromarketing. http://dx.doi.org/10.1177/0276146713481605. Decker, T., Menrad, K., 2015. House owners' perceptions and factors influencing their choice of specific heating systems in Germany. Energy Policy 85, 150–161. http:// dx.doi.org/10.1016/j.enpol.2015.06.004. Decker, T., Zapilko, M., Menrad, K., 2009. Purchase behaviour related to heating systems in Germany with special consideration of consumers' ecological attitudes, Energy Engineering, Economics and Policy Conference. Orlando, USA. Directive 2009/28/EC, 2009. On the Promotion of the Use of Energy from Renewable Sources and Amending and Subsequently Repealing Directives 2001/77/ EC and

305

Energy Policy 102 (2017) 288–306

M. Hecher et al.

Energy Econ. 32, 129–136. http://dx.doi.org/10.1016/j.eneco.2009.06.004. Schendera, C.F.C., 2008. Regressionsanalyse mit SPSS. Oldenbourg Wissenschaftsverlag GmbH, München. Sopha, B.M., Klöckner, C.A., 2011. Psychological factors in the diffusion of sustainable technology: a study of Norwegian households' adoption of wood pellet heating. Renew. Sustain. Energy Rev. 15, 2756–2765. http://dx.doi.org/10.1016/ j.rser.2011.03.027. Sopha, B.M., Klöckner, C.A., Hertwich, E.G., 2011. Adopters and non-adopters of wood pellet heating in Norwegian households. Biomass-. Bioenergy 35, 652–662. http:// dx.doi.org/10.1016/j.biombioe.2010.10.019. Sopha, B.M., Klöckner, C.A., Skjevrak, G., Hertwich, E.G., 2010. Norwegian households' perception of wood pellet stove compared to air-to-air heat pump and electric heating. Energy Policy 38, 3744–3754. http://dx.doi.org/10.1016/ j.enpol.2010.02.052. SPSS Software. IBM Corp. Released, 2013. IBM SPSS Statistics for Windows, Version 22.0. IBM Corp, Armonk, NY. Statistics Austria, 2015. Energistatistik: Energiebilanzen Österreich 1970 bis 2014 (Erstellt am 16.12.2015) - Aufgliederung nach der Struktur der Nutzenergieanalyse 2010. file:///C:/Users/Maria/Downloads/eev_2005_bis_2014_nach_et_und_ nutzenergiekategorien_fuer_oesterreich%20(1).pdf Statistics Austria, 2016. Energiestatistik: MZ Energieeinsatz der Haushalte 2003/2004, 2005/2006, 2007/2008, 2009/2010, 2011/2012 und 2013/2014 (Erstellt am 18.01. 2016). file:///C:/Users/Maria/Downloads/einsatz_aller_energietraeger_nach_ verwendungszwecken_2003_bis_2014%20(1).pdf Stolarski, M.J., Krzyżaniak, M., Warmiński, K., Niksa, D., 2016. Energy consumption and costs of heating a detached house with wood briquettes in comparison to other fuels. Energy Convers. Manag. 121, 71–83. http://dx.doi.org/10.1016/ j.enconman.2016.05.031. Wilson, C., Grubler, A., Gallagher, K.S., Nemet, G.F., 2012. Marginalization of end-use technologies in energy innovation for climate protection. Nat. Clim. Change 2, 780–788. http://dx.doi.org/10.1038/NCLIMATE1576.

Michelsen, C.C., Madlener, R., 2016. Switching from fossil fuel to renewables in residential heating systems: an empirical study of homeowners' decisions in Germany. Energy Policy 89, 95–105. http://dx.doi.org/10.1016/ j.enpol.2015.11.018. Michelsen, C.C., Madlener, R., 2013. Motivational factors influencing the homeowners' decisions between residential heating systems: an empirical analysis for Germany. Energy Policy 57, 221–233. http://dx.doi.org/10.1016/j.enpol.2013.01.045. Michelsen, C.C., Madlener, R., 2012. Homeowners' preferences for adopting innovative residential heating systems: a discrete choice analysis for Germany. Energy Econ. 34, 1271–1283. http://dx.doi.org/10.1016/j.eneco.2012.06.009. Michelsen, C.C., Madlener, R., 2010. Integrated Theoretical Framework for a Homeowner ’ s Decision in Favor of an Innovative Residential Heating System Carl Christian Michelsen and Reinhard Madlener February 2010 Institute for Future Energy Consumer Needs and Behavior ( FCN ) Faculty of Busi. Energy. Mintzberg, H., Raisinghani, D., Théorêt, A., 1976. The structure of “unstructured” decision processes. Adm. Sci. Q 21, 246–275. NREAP, 2010. Nationaler Aktionsplan für Erneuerbare Energie. Österreichischer Biomasse-Verband: Vienna, Austria. Nyrud, A.Q., Roos, A., Sande, J.B., 2008. Residential bioenergy heating: a study of consumer perceptions of improved woodstoves. Energy Policy 36, 3159–3166. http://dx.doi.org/10.1016/j.enpol.2008.04.019. E.M. Rogers In: Fifth Edit.(Ed.), Diffusion of innovations 2003 The Free Press New York. Rotmann, S., 2016. How to create a ‘magic carpet’ for behaviour change, 4th European Conference on Behaviour and Energy Efficiency, Coimbra, 8-9 September 2016. Rouvinen, S., Matero, J., 2013. Stated preferences of Finnish private homeowners for residential heating systems: a discrete choice experiment. Biomass-. Bioenergy 57, 22–32. http://dx.doi.org/10.1016/j.biombioe.2012.10.010. Sardianou, E., Genoudi, P., 2013. Which factors affect the willingness of consumers to adopt renewable energies? Renew. Energy 57, 1–4. http://dx.doi.org/10.1016/ j.renene.2013.01.031. Scarpa, R., Willis, K., 2010. Willingness-to-pay for renewable energy: primary and discretionary choice of British households' for micro-generation technologies.

306