How Social Interaction Induce Energy Saving Behaviors in Buildin 2023 Energy

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Energy Economics 118 (2023) 106515

Contents lists available at ScienceDirect

Energy Economics
journal homepage: www.elsevier.com/locate/eneeco

How social interaction induce energy-saving behaviors in buildings:


Interpersonal & passive interactions v.s. public & active interactions
Hao Li a, Zhao-Hua Wang a, b, *, Bin Zhang a, b, c, *
a
School of Management and Economics, Beijing Institute of Technology, Haidian, Beijing 100081, China
b
Center for Sustainable Development and Smart Decision, Beijing Institute of Technology, Beijing 100081, China
c
Center for Energy and Environmental Policy Research, Beijing Institute of Technology, Beijing 100081, China

A R T I C L E I N F O A B S T R A C T

Keywords: Programs and policies that reduce energy consumption take the home, workplaces, and communities as the unit
Individual energy-saving behavior of analysis, with little consideration of the social interactions among people. Based on social interaction theory
Interpersonal relationship and behavioral decision-making theory, this paper conducted a questionnaire study and collected >1000 samples
Public relationship
to analyze the effects of social relationships, active or passive interaction, and individual characteristics on
Active interaction
Passive interaction
energy-saving behaviors at different levels of difficulty, such as habit adjustments, equipment upgrades, and
housing retrofits. The results show that social interactions positively contribute to individual energy-saving
behaviors, but there are significant differences in the effects on different behaviors. Low-difficulty energy-
saving behaviors are susceptible to interpersonal and passive interactions, while high-difficulty energy-saving
behaviors are susceptible to public relations and active interactions. In addition, individual perceived value and
information processing depth play a mediating and moderating role, respectively, in the impact of social in­
teractions on energy-saving behaviors. Based on a new social interaction perspective, the findings of this paper
can provide scientific support for the formulation of differentiated, non-economically motivated energy-saving
policies.

1. Introduction those others (Burkitt, 2014). Social relationships and interaction styles
are two critical dimensions for evaluating social interactions (Har­
Current efforts to change energy demand patterns tend to target greaves and Middlemiss, 2020; Wolske et al., 2020). The two central
people as discrete and isolated individuals. In so doing, they ignore the social relations types are interpersonal and public (Hargreaves and
fact that individual energy needs are characterized by social interaction, Middlemiss, 2020). Interpersonal relationships are between individuals,
i.e., no one consumes energy in isolation, and most energy decisions are such as family, friends, and colleagues. Public relationships exist be­
influenced by social interactions (Crossley, 2010). When using air con­ tween individuals and organizations, such as government departments,
ditioners for cooling or heating, setting a reasonable temperature to community properties, housing agencies, energy supply, and service
coordinate the comfort needs of different individuals depends on companies. There is a clear difference between interpersonal relation­
communication and interaction (Sintov et al., 2019). Social interactions ships and public relationships. Interpersonal relationships are built at
also drive various energy-saving behavior, and many people will do the the individual level, they are based on common interests, common
same when they see their neighbors install solar systems (Bollinger and values, common experiences, etc. Public relationships are built at the
Gillingham, 2012). And comparing neighbors’ electricity bills has also organizational level, and they are based on the common goals, common
been shown to promote energy-saving behaviors (Allcott, 2011). Social interests, and common moral principles of individuals and organiza­
interactions shape individual experiences, choices, and behaviors in tions. In addition to that, the relationship between people is more per­
socialized groups, fundamentally influencing all aspects of daily life, sonal, while the relationship between people and organizations is more
including energy needs. public. Individuals use interpersonal relationships to learn and adjust
Social interactions characterize how people act as structured by their energy-use behavior, share energy-saving information and technology,
situations, others involved in those situations, and their relations with and use public relationships to obtain energy-use advice, energy-saving

* Corresponding authors at: School of Management and Economics, Beijing Institute of Technology, Haidian, Beijing 100081, China.
E-mail addresses: wangzhaohua@bit.edu.cn (Z.-H. Wang), zhangbin8706@163.com (B. Zhang).

https://doi.org/10.1016/j.eneco.2023.106515
Received 19 September 2022; Received in revised form 31 December 2022; Accepted 6 January 2023
Available online 10 January 2023
0140-9883/© 2023 Published by Elsevier B.V.
H. Li et al. Energy Economics 118 (2023) 106515

products, and energy-saving retrofit services. perspective of social interaction processes, incorporate various di­
The main types of interaction include active and passive interactions mensions of social interactions (e.g., interpersonal relationships, public
(Wolske et al., 2020). Active interaction is the communication between relationships, active interactions, and passive interactions) into a unified
different subjects, including impromptu face-to-face conversations, analytical model, and systematically investigate the roles played by
organized group meetings, and online community discussions. Passive different social interaction factors in motivating different types of in­
interaction is when people observe and think about the characteristics of dividual energy-saving behaviors, to fill the research gap and provide
their surroundings and the behavior of other individuals. Active inter­ new ideas for building a more flexible, low-carbon, and sustainable
action is when a person actively engages in communication and actively energy demand model.
expresses his or her thoughts or feelings. This type of communication The remainder of this paper will be presented as follows. Section 2
involves interaction between two people and can facilitate mutual un­ describes the research model and hypotheses of this paper. Section 3
derstanding and communication between people. Conversely, passive presents the research methods and data of this paper. Section 4 presents
interaction refers to a person not actively participating in the commu­ the research results. Section 5 gives the conclusions. Finally, Section 6 is
nication, but rather observing and recording what is happening around a discussion related to the study of this paper.
them. This type of observation usually does not involve interaction be­
tween two people, but simply observes and records. Active interactions 2. Theoretical foundation and hypothesis development
help individuals deepen their knowledge of new technologies or new
sources of energy or gain important procedural information (Berger, 2.1. Theoretical foundation
2014; Lane et al., 2018). In the absence of an objective basis, individuals
can use passive interactions to form subjective evaluations of certain Social interaction theory suggests that individuals are embedded in
behaviors by comparing and referring to the behaviors of others. social networks, that group behavior influences individual behavior, and
Social interactions are conducive to enhancing individual energy- that individual behavior becomes the driving force behind the emer­
saving behaviors. However, there is a lack of clear conclusions about gence of group behavior (Manski, 2000). Veblen, the founder of insti­
what role different social interaction factors play in the process of in­ tutional economics, was one of the first scholars to study social
dividual energy-saving behavior elicitation. A research study on resi­ interaction theory. His conspicuous consumption theory reflected social
dential solar systems applications in the United States shows that both factors’ significant influence on individual behavior (Veblen, 1899).
active and passive interactions can shorten the decision period for in­ Duesenberry (1949) and Leibenstein (1950) proposed the relative in­
dividuals considering the application of solar systems, with active in­ come hypothesis that underpins the theory of social interaction,
teractions being more effective. (Rai and Robinson, 2013). But another following Veblen. They argued that individuals are willing to imitate
study suggests active interactions do not necessarily trigger more sig­ others and compare themselves with each other and that the share of
nificant interest in solar systems than passive interactions. The key is one’s consumption expenditure in total income depends on the payment
whether individuals promote reduced uncertainty about how solar sys­ of those in the same class around him. Based on the above scholars’
tems and available subsidies work through social interactions (Palm, research, many sociological and psychological researchers have
2017). Further research is needed to answer the debate about how social explored the influence of social interaction in the form of the reference
interactions affect individual energy-saving behavior. effect, social learning, word-of-mouth effect, and network externalities
In addition, there is a clear typology of individual energy-saving (Inhoffen et al., 2019; Guo et al., 2021).
behaviors. Synthesizing the existing literature on the typology (Van Considering that no one consumes energy in isolation, most energy
Raaij and Verhallen, 1983; Parker et al., 2003; Barr et al., 2005; Wang decisions are influenced by social interactions (Crossley, 2010). There­
et al., 2022) and further considering the behavioral difficulty, individual fore, social interactions are also considered a critical factor in the gen­
energy-saving behaviors can be classified into three types: habit ad­ eration and development of individual energy-saving behaviors (Yin and
justments, equipment upgrades, and housing retrofits. Habit adjust­ Shi, 2021), which provides an essential theoretical model for the
ments reduce energy waste by changing one’s energy use habits while ongoing interactions between various social factors and individual
meeting primary energy demand. Equipment upgrades mean choosing energy-saving behaviors. Through the process of social interactions
products with higher energy efficiency when purchasing energy-using influencing individual energy-saving behaviors outlined in Section 1, we
equipment to reduce energy consumption. Housing retrofits refer to first identified four key explanatory variables that contribute to the
larger-scale energy-saving retrofitting actions such as installing insu­ emergence and consolidation of individual energy-saving behaviors,
lated facades and rooftop solar systems to reduce energy consumption in namely interpersonal relationships, public relationships, active in­
building operations. The above three individual energy-saving behav­ teractions, and passive interactions, and developed our conceptual
iors are progressively more challenging to implement. Among them, model, as shown in Fig. 1.
habit adjustments do not require the additional monetary investment At the same time, considering that individual energy-saving behavior
and have the lowest energy-saving threshold. Equipment upgrades and is essentially a behavioral decision, the individual’s decision charac­
housing retrofits are investment behaviors. The highest threshold for teristics should also be critical factors influencing their energy-saving
energy saving is for a house renovation, which requires more time and behavior (Allen et al., 2015). Based on decision theory (Simon, 1977)
money and objective conditions but also has the best energy-saving and perceived value theory (Zeithaml, 1988; Woodruff, 1997), how an
effect. individual chooses between multiple behaviors depends on the overall
Based on different social relationships and interaction styles, various perception and evaluation of the utility of the behavior, i.e., perceived
social interaction paths can be combined to influence individual energy- value. Perceived value is a crucial reference factor for individual
saving behaviors. And different paths may apply to behaviors of behavioral decisions (Neumann and Morgenstern, 1944). Studies on
different difficulties. Clarifying the above matching relationships is of behaviors such as online shopping (Zhang et al., 2015), knowledge
great value in developing differentiated, precise, and sustainable payment (Sánchez et al., 2021), electric car purchases (He and Hu,
energy-saving programs. However, most studies on individual energy- 2021), and online healthcare (Sun et al., 2022) have demonstrated that
saving behaviors have focused on attributes such as attitudes, subjec­ perceived value can positively influence individuals’ behavioral ten­
tive norms, and behavioral intentions. The role of social interaction in dencies. External stimuli are critical in forming perceived value (Keller
shaping individual energy-saving behaviors has been neglected. For et al., 1982). Social interactions are full of various types of information
these reasons, there is a lack of systematic research that integrates and environmental stimuli. Individuals involved in social interactions
multiple social interaction factors. This study aims to explore the non- receive many stimuli that create new perceptions and feelings about
economic incentives of individual energy-saving behaviors from the something, changing its perceived value. In energy decisions, rational

2
H. Li et al. Energy Economics 118 (2023) 106515

Fig. 1. Conceptual model.

individuals decide whether to implement energy-saving behaviors based individuals obtain through trust mechanisms (Luo and Xi, 2012) and
on the perceived value level to maximize utility. In other words, further influence individuals’ behavioral decisions (Yin and Shi, 2021).
perceived value is a crucial decision basis for individual energy-saving The complexity of social relations increases the difficulty of subdi­
behavior. vision. Different social relations can also be transformed into each other.
In addition, different ways of processing available information may Based on the research demand, this paper mainly studies interpersonal
result in individuals acquiring different perceived values for the same and public relationships in social relations, which are also the most
behavior (Ren and Huang, 2018). In behavioral decision-making, some important social relations. When defining individual relationships, we
individuals may process information rather superficially, reacting give priority to whether they can be divided into interpersonal re­
quickly to interactions based on intuition. In contrast, others analyze the lationships. For example, the relationship between individuals and the
acquired information in depth, weighing the pros and cons as much as enterprises they work for can be regarded as the relationship between
possible before forming a judgment to make unbiased decisions. Evi­ individuals and colleagues in energy behavior research. The workplace
dence from outside the energy domain suggests that attitudes formed is the scene of energy use, and individuals and their colleagues are the
through this deliberate information processing pathway are more du­ main body of energy use.
rable and likely to translate into behavior (Stiff and Mongeau, 2016). In Close interpersonal relationships with family and friends shape
social interactions, how individuals process the information they obtain when, where, and to what extent individuals use energy. That is, in­
in different depths also affects their behavioral choices. dividuals’ energy needs and energy-saving decisions are not only
The perceived value and depth of information processing play an dependent on their preferences but are also influenced by the needs and
essential role in the path of social interactions on individual energy- behaviors of others (Burningham and Venn, 2017). These relations with
saving behavior. Therefore, we added perceived value and informa­ family and friends are often based on strong emotional bonds around
tion processing depth to the hypothesized model (as shown in Fig. 1), care, intimacy, love, and friendship. These relationships are achieved
which is essential to refine the relationship between individual energy- through a range of interpersonal practices, including the practices of
saving behavior and the four explanatory variables mentioned above. parenting, friendship, and intimacy (Jamieson, 2016). It is through the
The conceptual model in Fig. 1 contains three separate analytical regular implementation of these practices that social relationships with
models for three different types of energy-saving behaviors. In these family and friends have a significant impact on energy demand (Har­
models, individual social relationships and interaction styles share greaves and Middlemiss, 2020). There are more restrictions on energy
explanatory variables, while perceived value and information process­ demand when living with a family than when living alone. In households
ing depth are measured separately in the three types of energy-saving with infants and young children, there is a non-negotiable need for naps,
behaviors. feedings, or bedtimes (Burningham and Venn, 2017; Nicholls and
Strengers, 2015). In households with school-age or adolescent children,
such concerns have gradually evolved to include regular schedules for
2.2. Hypothesis development
school hours, homework, and after-school activities (Powells et al.,
2014). Such relations are not exclusive to the home either. At work too,
2.2.1. Social relationships and individual energy-saving behaviors
people must negotiate with their co-workers about thermostat settings in
Individuals’ energy needs are shaped by their relationships with
public offices, as well as communicate about whether lights or other
other subjects regarding consensus, conflict, cooperation, and compe­
equipment should be turned on or off.
tition. In communications involving energy use, people also tend to talk
A community study from the Netherlands showed that good inter­
about relationships with other people or organizations to explain how
personal networks contribute to spreading energy efficiency information
and why they use energy in the way they do (Hargreaves and Mid­
and intentions and ultimately drive the popularity of housing energy
dlemiss, 2020). This information about energy use and management is
efficiency retrofits in the community (Wilde, 2019). And a growing body
exchanged through extensive social interactions, but the exchange is
of research shows that children who are well-educated about energy
based on mutual trust (Granovetter, 1973). Based on trust theory,
efficiency become effective drivers of household energy efficiency
relationship strength determines individuals’ trust level (Wang et al.,
(Boudet et al., 2016). However, when good interpersonal relationships
2007). And social interactions with high trust can give individuals more
are lacking, and consensus cannot be formed among individuals
access to high-quality resources, and social interactions with low trust
regarding the use of shared equipment (e.g., air conditioning), it may
will limit the availability of resources (Yang et al., 2018). In social in­
result in energy waste or additional energy demand (Sintov et al., 2019).
teractions, social relationships can influence the quality of resources

3
H. Li et al. Energy Economics 118 (2023) 106515

The role played by interpersonal relationships varies in the energy- 2.2.2. Interaction styles and individual energy-saving behaviors
saving behavior of individuals with different difficulty levels. An Social interaction is a meaningful way for individuals to become
essential use of relational networks is to link different individuals into a embedded in social groups and an essential medium for macro-groups to
group and pursue common goals through collective action (Savelli and influence micro-individuals (Granovetter, 1985). Based on norm acti­
Morstyn, 2021). In this process, individuals can receive appropriate vation theory, individuals are motivated to engage in pro-social
support from interpersonal networks (Turner and Turner, 2013; Carlsen behavior not only by their interests and intentions but also by beliefs
et al., 2020). In individual energy-saving scenarios, individuals can learn about morality, i.e., personal norms (Schwartz, 1977). Further, ac­
and adjust their energy use behavior, obtain energy-saving information cording to the stimulus-organism-response theory (Mehrabian and
and technical advice, etc., through interpersonal relationships. This Russell, 1974), all behavior of an individual is the formation of an as­
provides favorable conditions for the smooth implementation of energy- sociation between stimulus and response. Social interactions are
saving behaviors. However, as the difficulty of implementing energy- necessary scenarios for activating personal norms, and activated per­
saving behaviors increases, individuals’ needs for resources such as in­ sonal norms can influence individuals’ pro-social behaviors (Donald
formation, technology, and capital also increase. The support that in­ et al., 2014), such as energy-saving behaviors.
dividuals receive from interpersonal relationships may not be able to Active interactions mean that people actively engage in communi­
meet the corresponding needs. For example, professional advice from cation and actively express their thoughts and feelings. This type of
energy service companies and community property companies may have interaction usually involves interaction between multiple people and
a greater impact on individuals’ behavioral decisions for housing can facilitate mutual understanding and communication between peo­
retrofits. ple. Active interaction is important for building and maintaining social
As a result, the following hypothesis is proposed in this paper: relationships and is an important way for people to gain information and
learn new knowledge. And active interactions can facilitate the rapid
H1. Interpersonal relationships have a significant positive effect on
dissemination of energy information in social networks and give in­
individual energy-saving behaviors, but the effect of interpersonal re­
dividuals rich information stimulation. For example, it may help deepen
lationships decreased sequentially as the difficulty of energy-saving
an individual’s knowledge of new technologies or new sources of en­
behaviors increased.
ergy, reduce uncertainty when adopting new products and services, or
Extensive public relationships with institutions and communities convey crucial procedural information (Berger, 2014; Lane et al., 2018).
shape how much energy individuals use and what can be done to Word-of-mouth effects created by active interactions have been shown
manage or reduce energy demand. to play a positive role in energy-saving acts such as new energy vehicle
These relations may be formal (legal, contractual) or informal and purchases (Osei-Frimpong and McLean, 2018) and rooftop solar appli­
are based on the provision, management, and receipt of energy and cations (Bollinger et al., 2019). Adopters of energy-efficient technolo­
energy services as a customer or user, or as a member of a community of gies often perceive that effective proactive interaction improves the
place or practice. For example, good public relationships structured by quality of available information (Rai and Robinson, 2013). Numerous
housing leases can encourage landlords to improve the energy efficiency studies have also shown that if a new energy technology is discussed
of appliances to reduce tenants’ energy bills (Hoppe, 2012). If tenants more frequently in social circles, it is more likely to be accepted and
are reluctant to raise energy costs with landlords for fear of eviction, adopted (Wolske et al., 2020).
then landlords may also have no incentive to make energy efficiency The effect of active interactions on energy-saving behaviors may
investments. Similarly, implementing smart meter penetration or other vary across difficulties. Simple habit-adjusting behaviors require mini­
energy efficiency intervention programs in numerous countries faces mal cost and cognitive analysis and are more likely to be directly
significant barriers due to the lack of trust between customers and en­ motivated by the external environment (Allen et al., 2015; Truelove and
ergy suppliers (Darby, 2010; Hess, 2014; Rosenow and Eyre, 2016; Gillis, 2018). In contrast, highly challenging energy-saving behaviors
Sovacool et al., 2017). Under current energy efficiency policies, home­ require significant time and money and involve multiple decision stages.
owners are not subject to strict requirements from rental agents, energy Individuals need to acquire sufficient information through active social
providers, or utilities. The lack of trust and incentives prevents most interactions to support their energy-saving decisions (Klöckner, 2014;
homeowners from accepting smart meters or outlets, resulting in low Klöckner and Nayum, 2016).
participation rates in energy efficiency interventions. As a result, the following hypothesis is proposed:
Public relationships with institutions and communities play a critical
H3. Active interactions have a significant positive effect on individual
role in whether individuals adopt energy recommendations and how
energy-saving behaviors, and the effect of active interactions increased
they use them.
sequentially as the difficulty of energy-saving behaviors increased.
Public relationships may play a different role in individual energy-
saving behaviors of varying difficulty. Habit adjustments are simple to Passive interactions based on observational thinking about the
implement, do not involve additional purchases, and are rarely con­ behavior of others and the surrounding environment also have signifi­
strained by organizations such as government agencies, community cant behavioral influence effects. For example, observers are likelier to
properties, and energy service companies, resulting in a limited role for act the same if they see others comply with message signs encouraging
public relations in such behaviors. But equipment upgrades and housing water conservation (Aronson and O’Leary, 1983). This effect has been
retrofits are investment behaviors in which public relations between demonstrated in other scenarios (Sussman et al., 2013). According to the
individuals and energy-efficiency product manufacturers can play an social comparison theory (Festinger, 1954), individuals form evalua­
important role (Wade et al., 2016). Especially in the case of housing tions or decisions by comparing and referring to the behavior of others in
renovation, which involves multiple parties such as homeowners, the absence of an objective basis. Social learning theory (Bandura and
property owners, and service providers, complex public relationships Walters, 1977) also states that individuals learn new behaviors and
are essential to successful implementation (Aiken et al., 2017). form, reinforce, or change their attitudes toward these behaviors
As a result, the following hypothesis is proposed in this paper: through observation and imitation of others’ behaviors, propaganda
models, and social norms (Wang et al., 2022). Social identity theory also
H2. Public relationships have a significant positive effect on individual
reveals similar insights (Hogg, 2016). These sociological theories help
energy-saving behaviors, and the effect of public relationships increased
explain why adopting rooftop solar systems and new energy vehicles
sequentially as the difficulty of energy-saving behaviors increased.
spreads gradually from sporadic households to surrounding
communities.
Like active interactions, passive interactions may also exert different

4
H. Li et al. Energy Economics 118 (2023) 106515

influential effects on energy-saving behaviors of different difficulties. lasting impact depends on individuals’ motivation to engage in social
Passive interactions are the acquisition of information by observing the interactions and the depth to which they process the interaction infor­
external environment or the behavior of others and do not affect or mation (Wolske et al., 2020). Based on heuristic and analytical dual-
interfere with the external environment or others. In social interactions, system information processing theory (Kahneman, 2011), some in­
passive interactions can help people better understand the emotions, dividuals may process information rather superficially, reacting quickly
attitudes, and intentions of others. And passive interactions help people to interactions based on intuition, while others analyze the acquired
improve their communication skills and be better able to cope with information in depth, weighing the pros and cons as much as possible
different social situations. In addition, passive interactions can enhance before forming a judgment, to make unbiased decisions. In short, in­
people’s observation skills and help them better analyze and understand dividuals obtain much information from effective social interactions and
their surroundings. Observations and reflections on the behavioral and process it at different depths based on their interaction motivations and
environmental characteristics of individuals around them can influence decision-making styles, influencing their behavioral choices. Based on
individual energy-saving behaviors through social norms. Social norms the above analysis and considering the critical influence of information
are more conducive to motivating low-efficacy energy-saving behaviors in decision theory for value calculation, this paper focuses on the
(Schultz et al., 2007; Ayres et al., 2013), such as habit adjustments. In moderating role of the depth of information processing in the pathway
contrast, the influence of social norms is often very limited in high- of active and passive interactions acting on perceived value and pro­
energy efficiency behaviors (Wolske et al., 2020). This is not to say poses the following hypothesis.:
that passive interactions cannot play a role in large-scale energy in­
H6. The information processing depth moderates the paths through
vestments. The standard view is that passive interactions stimulate an
which active and passive interactions act on perceived value.
individual’s initial awareness of and interest in investment-based energy
conservation. It is important to note that this paper does not consider the possible
As a result, the following hypothesis is proposed: moderating role of information processing depth in the paths through
which interpersonal and public relations act on perceived value. This is
H4. Passive interactions have a significant positive effect on individual
because information processing depth can be significantly influenced by
energy-saving behaviors, but the effect of passive interactions decreased
social relationships. People usually judge the credibility and importance
sequentially as the difficulty of energy-saving behaviors increased.
of information based on their relationships with others and will tend to
pay more attention to the information provided by people with whom
2.2.3. The mediating role of perceived value
they have close relationships. In addition, social relationships may affect
In social interactions, information dissemination and word-of-mouth
the way people understand and interpret information, leading to
effects are influenced by the strength of relationships between subjects
different reactions to the same information. When processing informa­
in social networks (Eric and Kevin, 2014). Individuals may react and
tion, people are influenced by their relationships with other people or
evaluate other individuals or organizations with different relationship
organizations, which in turn may have an impact on the processing of
strengths differently. For example, when the relationship is close,
information. In conclusion, the depth of information processing does not
greater trust reduces risk and uncertainty in the decision-making process
fit to be considered as a moderating factor in the path through which
(Dodds et al., 1991), and the perceived value of the behavior is higher
interpersonal and public relations act on perceived value.
(Zeithaml, 1988), which is more conducive to the individual’s energy-
saving behavior. Conversely, when the relationship is weak, in­
3. Data and methods
dividuals may question the reliability of the information they receive,
which increases the uncertainty they perceive in the decision-making
3.1. Data collection
process (Osgood and Tannenbaum, 1955) and does not contribute to
energy-efficient decisions. Therefore, different relationship strengths in
This paper is based on a quantitative study to analyze the influence of
social interactions can affect individuals’ perceptions of energy-saving
individuals’ social interaction activities on their energy-saving behav­
behaviors and impact perceived value, influencing energy-saving
iors utilizing a questionnaire. The scales used in this paper refer to
decisions.
existing mature scales. Some scales are modified with expert opinions
Further, according to stimulus-organism-response theory (Mehra­
for oral and localization to help respondents better understand the
bian and Russell, 1974), external stimuli affect an individual’s final
contents of the questionnaire. Before the official questionnaire was
behavioral decision by triggering an internal organism response. In so­
distributed, a small-scale pre-survey was conducted in February–March
cial interactions, individuals receive information or environmental
2022. Specifically, university students were recruited as volunteers and
stimuli from external sources, either actively or passively, and expand
used the opportunity to return home during the holidays to distribute
their perceptions from multiple perspectives, which provides more basis
electronic questionnaires in large communities and supermarkets. An
for assessing perceived value (Cui et al., 2022). A large amount of pos­
online system automatically tallied the answers. Based on the pre-survey
itive energy-saving evaluation information will increase individuals’
results, the reliability and validity of the questionnaire were initially
perceived value of energy-saving from several levels, leading them to
tested, and further adjustments were made to the questions to form the
form a more vital willingness to save energy. When a large amount of
official questionnaire.
information stimulates an individual, it will first trigger changes in the
Considering the questionnaire collection period and sample
individual’s perception of the feasibility, economy, and social evalua­
coverage, we chose to cooperate with the professional data collection
tion of energy-saving behavior, i.e., affect the individual’s perceived
platform “Wenjuanxing” in the formal survey stage. This platform has a
value and eventually stimulate different behavioral responses.
sample base of millions of members, covering people of different ages,
As a result, the following hypothesis is proposed in this paper:
professions, and communities, and has collected >10 billion question­
H5. Perceived value plays a mediating role in the pathway of social naires. In this study, 1300 questionnaires were distributed through the
interactions on individual energy-saving behaviors. platform in May 2022. Invalid questionnaires that were not logical or
had the same evaluation value for six consecutive options were deleted,
2.2.4. The moderating effect of information processing depth and finally, 1019 valid questionnaires were obtained, with a valid re­
Information is an essential resource that individuals obtain from covery rate of 78.4%. The ratio of male to female respondents is close to
social interactions and is a crucial basis for individuals’ value analysis. 1:1. The distribution of demographic variables such as age, education,
Information stimuli can influence individuals’ beliefs or attitudes in income, and residential attribute variables is more reasonable, as shown
social interactions, and the ability of information stimuli to have a in Table 1. The sample characteristics are consistent with reality. The

5
H. Li et al. Energy Economics 118 (2023) 106515

Table 1
Control variables of the sample (N = 1019).
Variable Category Frequency Proportion Variable Category Frequency Proportion
(%) (%)

18–25 167 16.4 ≤70 150 14.7


26–30 363 35.6 71–90 237 23.3
Age 31–35 372 36.5 Housing area (m2) 91–110 315 30.9
36–40 91 8.9 111–130 220 21.6
>40 26 2.6 >130 97 9.5
Junior high school and
5 0.5 Before 1995 41 4.0
below
High school/technical
41 4.0 Years of residential 1995–2005 191 18.7
Educational background school
construction
Junior college 273 26.8 2006–2010 317 31.1
Undergraduate 647 63.5 2011–2015 271 26.6
Graduate and above 53 5.2 After 2015 199 19.5
≤2000 93 9.1 1 60 5.9
2001–5000 108 10.6 2 148 14.5
Monthly household income
5001–8000 284 27.9 Family resident population 3 384 37.7
(yuan)
8001–15,000 342 33.6 4 295 28.9
>15,000 191 18.7 ≥5 132 13.0
Female 519 50.9
Gender
Male 500 49.1

minimum age of the research sample in this paper was set at 18 years This indicates that it is easier to establish high trust social relationships
old, which is the mark of adulthood. Adults are considered to be capable between individuals than asymmetric relationships between individuals
of engaging in different types of energy-saving behaviors. and organizations. Regarding social interactions, the mean active
interaction validity for the respondent group is lower than the mean
3.2. Measurement method passive interaction validity. For different energy-saving behaviors, the
frequency and perceived value of the behaviors continue to decrease
In this paper, we use the main variables in a multi-item scale mea­ with increasing difficulty while the depth of information processing
surement model, including the Individual Energy-saving Behavior Scale continues to grow. It is assumed that for simple energy-saving behaviors
(referenced from Allen et al., 2015), Interpersonal Relationships Scale such as habit adjustment, individuals can make quick decisions based on
(referenced from Lechner et al., 2010), Public Relationships Scale their intuition or a small amount of rational deduction to perceive the
(referenced from Bohner et al., 2003), Active Interactions Scale (refer­ behavior’s benefits. Individuals need to think carefully about
enced from Sussman et al., 2013), Passive Interactions Scale (referenced investment-oriented energy-saving behaviors, such as equipment up­
from Ou et al., 2014), Perceived Value Scale (referenced from Park and grade and housing renovation. In particular, housing renovation is likely
Kwon, 2017), and Information Processing Depth Scale (referenced from to be abandoned because of the high cost of money and time.
Truelove and Gillis, 2018), all in the form of a 5-point Likert scale. Next, this paper conducts a heterogeneity analysis using the three
The habit-adjusted energy-saving subscales ranged from 1 (never) to types of energy-saving behaviors as the dependent variables and the
5 (always), indicating low to high frequency of the behavior, and the control variables in Table 1 as the independent variables to investigate
remaining scales ranged from 1 (not at all compliant) to 5 (very the effect of each type of resident attributes on energy-saving behaviors.
compliant), representing low to high levels of compliance. The scale and The results of the analysis are shown in Table 3. The grouping of each
question settings are shown in Supplementary Table 1. independent variable was kept in the same way as in Table 1.
As shown in Table 3, with the increase of residents’ age, the fre­
quency of ESA and ESB showed an M-shaped trend, and the highest
3.3. Descriptive analysis and heterogeneity analysis
frequency was above 40 years old. With the increase of residents’
educational background, the frequencies of ESA and ESC show an
The means, standard deviations, minimum and maximum values of
increasing trend, and the frequency of ESB shows an inverted U-shaped
all critical variables involved in this paper are presented in Table 2.
trend, with the highest educational background being Junior college.
As shown in Table 2, the respondent group has good interpersonal
With the increase of monthly household income, the frequencies of ESA
and public relationships. However, the mean value of public relation­
and ESB show an N-shaped trend, and the frequency of ESC shows an
ship strength is lower than that of interpersonal relationship strength.

Table 2
Descriptive statistics of variables.
Variables Abbreviations Means SD Min Max

Interpersonal Relationships IRS 4.13 0.484 1.00 5.00


Public Relationships PRS 3.86 0.742 1.00 5.00
Active Interactions ACI 3.92 0.783 1.00 5.00
Passive Interactions PAI 4.34 0.522 1.00 5.00
Habit Adjustments ESA 4.34 0.596 1.00 5.00
Energy-saving Behaviors Equipment Upgrades ESB 4.08 0.764 1.00 5.00
Housing Retrofits ESC 3.44 0.969 1.00 5.00
For ESA PVA 4.22 0.527 1.00 5.00
Perceived Value For ESB PVB 4.00 0.442 2.67 5.00
For ESC PVC 3.59 0.543 2.00 5.00
For ESA IDA 3.08 0.992 1.00 5.00
Information Processing Depth For ESB IDB 3.45 1.002 1.33 5.00
For ESC IDC 4.16 0.566 2.33 5.00

6
H. Li et al. Energy Economics 118 (2023) 106515

Table 3
Heterogeneity analysis of control variables.
Dependent variable Group Age Educational background Monthly household income

Mean F Mean F Mean F

1 4.269 3.733 4.229


2 4.347 4.244 4.361
ESA 3 4.377 3.312** 4.286 2.874** 4.265 3.683***
4 4.249 4.375 4.386
5 4.654 4.409 4.434
1 3.888 2.800 3.778
2 4.075 4.016 4.065
ESB 3 4.140 6.227*** 4.199 6.694*** 4.052 4.725***
4 4.018 4.056 4.139
5 4.577 3.855 4.149
1 3.337 2.200 3.183
2 3.455 3.057 3.222
ESC 3 3.444 1.408 3.454 4.077*** 3.394 7.523***
4 3.484 3.461 3.466
5 3.795 3.585 3.724

Housing area Years of residential construction Family resident population


Dependent variable group
Mean F Mean F Mean F
1 4.240 4.065 4.178
2 4.338 4.330 4.331
ESA 3 4.367 1.907 4.341 2.947** 4.336 1.611
4 4.348 4.402 4.384
5 4.440 4.343 4.371
1 3.918 3.984 3.822
2 4.031 4.044 4.016
ESB 3 4.119 2.678** 4.104 0.360 4.063 2.986**
4 4.147 4.070 4.121
5 4.131 4.089 4.194
1 3.162 3.528 3.006
2 3.367 3.286 3.223
ESC 3 3.438 6.861*** 3.496 1.671 3.502 6.581***
4 3.629 3.480 3.563
5 3.656 3.441 3.449

Notes: significant at * p < 0.10, **p < 0.05, ***p < 0.01.

increasing trend. With the increase of housing area, the frequency of ESB quantity, and the test results are shown in Table 4. The Cronbach’s alpha
shows an inverted U-shaped trend, and the frequency of ESC showed an coefficient for each subscale was above 0.6, indicating the scale has good
increasing trend. With the year of housing construction advances, the reliability (Wu et al., 2021).
frequency of ESA shows an inverted U-shaped trend. With the increase of In this paper, we used the mature scale as a template when designing
family resident population, the frequency of ESB generation shows an the scale and combined the suggestions of several experts. After the
increasing trend, and the frequency of ESC generation shows an inverted adjustment and modification in the pretesting stage, it can be considered
U-shaped trend. that the scales in this paper have good content validity. Regarding
construct validity, all three energy-saving behavior scales pass Bartlett’s
4. Results test (p < 0.001) with a high KMO value of 0.800, 0.821, and 0.842,
respectively. It indicates a correlation between the different topics and is
4.1. Reliability and validity of the measurement model suitable for factor analysis. The factor loading matrix is obtained by
orthogonal rotation using principal component analysis and the
In the pre-study stage, the reliability and validity of the question­ maximum variance method, as shown in Supplementary Table 2.
naire, scale composition, and dimensional division were initially All three types of energy-saving behavior scales have seven compo­
analyzed. The modified formal scale passed the reliability test in the nents drawn based on eigenvalues >1. The loading values of each
pretest stage. To ensure the validity of the formal questionnaire data, we component on the corresponding factor are >0.5, and the factor distri­
further tested the reliability and validity of the formal questionnaire bution structure is consistent with the expected scale design dimensions.

Table 4
Reliability test of the measurement model.
Variables α coefficient Number of items Variables α coefficient Number of items

IRS 0.611 3 PRS 0.799 4


ACI 0.744 3 PAI 0.820 3
ESA 0.638 3 ESB 0.782 3
IDA 0.838 3 IDB 0.820 3
PVA 0.702 3 PVB 0.706 3
ESC 0.796 3 ES 0.728 9
IDC 0.710 3 PV 0.788 9
PVC 0.837 3 ID 0.695 9
All Variables 0.871 40

Note: ES denotes the set of question items containing ESA, ESB, and ESC. PV and ID are defined in the same way.

7
H. Li et al. Energy Economics 118 (2023) 106515

This indicates that the scales designed in this paper have good construct Table 5
validity (Henseler et al., 2015). Results of the mediation effect test.
Total Direct Indirect Proportion of Type of
4.2. Direct and mediating effect tests of the measurement model effect effect effect mediating effect mediating
in total effect effect

This paper hypothesizes that the four types of social interaction IRS →
factors (IRS, PRS, ACI, and PAI) directly affect the promotion of indi­ PVA
19.80% 18.10% 1.70% 8.60%
Partial
mediation
vidual energy-saving behaviors and also hypothesizes that perceived →
ESA
value plays a mediating role in this. This paper uses stepwise regression PRS →
to test whether the above hypotheses hold for each of the three types of PVA Partial
13.40% 12.20% 1.20% 8.96%
energy-saving behaviors. → mediation
Stepwise regression is one of the currently accepted methods used to ESA
ACI →
test the mediating effect. This study constructs five regression models to
PVA Partial
test the mediating effect. Taking habit-adjusted energy efficiency as an 9.50% 8.00% 1.50% 15.79%
→ mediation
example, the specific test procedure is as follows. ESA
Model 1 is first constructed with ESA as the explanatory variable, and PAI →
PVA Partial
only the control variables (demographic and residential characteristics 22.20% 19.60% 2.60% 11.71%
→ mediation
variables) are added as references. Model 2 adds the explanatory vari­ ESA
ables (IRS, PRS, ACI, PAI) to be validated based on Model 1 to determine IRS →
the main effects. Model 3 adds the mediating variable (PVA) based on PVB
16.10% 14.10% 2.00% 12.42%
Partial
Model 1 to determine the relationship between the mediating variable → mediation
ESB
and the explanatory variable. Model 4 adds the mediating variable
PRS →
(PVA) based on Model 2 to determine the relationship between the PVB Partial
20.50% 18.00% 2.50% 12.20%
explanatory variable and the explanatory variable in the context of → mediation
controlling for the mediating variable. Finally, Model 5 is constructed to ESB
ACI →
determine the relationship between the explanatory variable and the
PVB Partial
explanatory variable. The detailed analysis results are shown in Sup­ →
23.40% 20.60% 2.80% 11.97%
mediation
plementary Table 3. ESB
The regression results for the three types of energy-saving behaviors PAI →
exhibit many common features. Taking habit-adjusted energy savings as PVB Partial
14.90% 12.40% 2.50% 16.78%
mediation
an example, the adjusted R2 is only 0.014 considering only the effect of

ESB
control variables on energy-saving behavior. After adding four explan­ IRS →
atory variables, including IRS (b = 0.198, p < 0.001), PRS (b = 0.134, p PVC Partial
13.40% 11.70% 1.70% 12.69%
< 0.001), ACI (b = 0.095, p < 0.001), and PAI (b = 0.222, p < 0.001), → mediation
the adjusted R2 improved to 0.189 and the coefficient of each explana­ ESC
PRS →
tory variable passed the significance test. PVC Partial
Further, comparing models 2 and 4, the inclusion of the mediating 23.60% 19.40% 4.20% 17.80%
→ mediation
variable PVA (b = 0.150, p < 0.001) further increases the adjusted R2 to ESC
0.204 and the coefficients of the explanatory variables IRS (b = 0.191, p ACI →
PVC Partial
< 0.001), PRS (b = 0.122, p < 0.001), ACI (b = 0.080, p < 0.01), and PAI 40.60% 32.50% 8.10% 19.95%
→ mediation
(b = 0.196, p < 0.001) all decrease and remain significant. Combined ESC
with other model results, it is clear that perceived value partially me­ PAI →
diates social interactions and individual energy-saving behavior. The PVC
12.60% 10.00% 2.60% 20.63%
Partial
complete results of the mediating effect test are shown in Table 5. → mediation
ESC
As shown from Model 4 in Supplementary Table 3, the effects of
social relationships, interaction styles, and perceived value on different
types of individual energy-saving behaviors show some variability and
regularity. 0.279, respectively). This may be because as difficulty increases, in­
Regarding social relationships, habit adjustments are more likely to dividuals become more sensitive to the benefits of energy savings (Chen
be influenced by interpersonal relationships, while equipment upgrades et al., 2020). This reinforces the effect of perceived value on energy-
and housing retrofits are more likely to be influenced by public re­ saving behavior. Thus, hypothesis H5 is supported.
lationships. Moreover, the influence coefficient of interpersonal re­
lationships on the three types of energy-saving behaviors decreases in 4.3. Moderating effect test of the measurement model
order (0.181, 0.141, and 0.117, respectively), and the influence coeffi­
cient of public relationships on the three types of energy-saving be­ Based on James and Brett (1984) and Baron and Kenny (1986) tests
haviors increases in order (0.122, 0.180, and 0.194, respectively). of moderating effects, this paper uses multiple regression analysis to test
Regarding interaction styles, habit adjustments are more likely to be the moderating effect of information processing depth in the path of
influenced by passive interaction, while equipment upgrades and social interaction validity on perceived value. The multiple regression
housing retrofits are more likely to be influenced by active interaction. analysis contains a total of three levels: the first level is the explanatory
Moreover, the influence coefficient of passive interaction on the three variable level, which is the social interaction validity; the second level is
types of energy-saving behaviors decreases in order (0.196, 0.124, and the moderating variable level, which is the information processing
0.100, respectively), and the influence coefficient of active interaction depth; and the third level is the interaction term between the explana­
on the three types of energy-saving behaviors increases in order (0.080, tory and moderating variables. From this, three regression models can
0.206, and 0.325, respectively). Thus, hypotheses H1-H4 are supported. be constructed. Specifically, Model 1 is a regression model of social in­
In terms of perceived value, the coefficients of the three energy- teractions on perceived value. Model 2 adds the information processing
saving behaviors are affected in increasing order (0.150, 0.219, and depth to Model 1, and Model 3 adds the interaction term of social

8
H. Li et al. Energy Economics 118 (2023) 106515

interactions and information processing depth to Model 2. Based on the depth information processing.
results of the multiple regressions, if the interaction coefficient is sig­ On the other hand, the moderating effect of information processing
nificant, the moderating effect is substantial. The results of the moder­ depth in the influence path of passive interaction showed the opposite
ation effect test for the three types of energy-saving behavior models are result. That is, information processing depth does not play a moderating
shown in Table 6. role in the “PAI to PVB” path but plays a moderating role in the “PAI to
As shown in Table 6, the effect of information processing depth on PVC” path. In passive interactions, individuals are more likely to observe
PVA is not significant (b = 0.005, p = 0.744) and its interaction term people around them buying or using energy-saving devices or making
coefficient with ACI (b = 0.021, p = 0.289) and PAI (b = 0.046, p = energy-saving improvements. Based on this information, individuals can
0.131) are also not significant. It can be seen that there is no significant only make a vague judgment about the perceived value of energy-saving
moderating effect of information processing depth on the “ACI to PVA” behaviors. Compared to retrofitting, purchasing is less complex and less
and “PAI to PVA” paths. As for investment-based energy savings, in­ involved, and deep processing of the limited information may not
formation processing depth has a significant positive effect on PVB (b = significantly change the individual’s judgment of its value. In contrast,
0.031, p < 0.05) and PVC (b = 0.065, p < 0.05). This indicates that in-depth information processing can reduce the uncertainty of a prob­
individuals will enhance their feelings and evaluate investment-based lematic, high-input behavior such as housing renovation (Palm, 2017),
energy-saving behavior after deep consideration. Further, the interac­ reinforcing individuals’ perceived value.
tion term coefficient between IDB and ACI is significantly positive (b = In summary, based on the results of this paper, Hypothesis 1- Hy­
0.052, p < 0.01) for the equipment upgrade. The interaction term co­ pothesis 5 are wholly supported, and Hypothesis 6 is only partially
efficient between IDC and PAI for housing renovation is significantly supported.
positive (b = 0.115, p < 0.05). This indicates a significant positive
moderating effect of information processing depth on the “ACI to PVB” 5. Conclusions
and “PAI to PVC” paths. Thus, hypothesis H6 is partially supported.
The results show that information processing depth plays different Social interaction is a critical factor in influencing and shaping in­
roles in different pathways. First, information processing depth does not dividual energy-saving behavior. No one consumes energy in isolation,
exert a direct or moderating effect on the perceived value of habit- and most energy decisions occur in social interaction. However, existing
adjusted energy savings. Possible reasons for this are that habit- studies have mainly focused on the psychological and situational factors
adjusted energy conservation requires no economic investment, the influencing individual energy-saving behavior. Little has been done to
threshold for entry into the behavior is shallow, and the perceived value explore the social interaction factors and their influence mechanisms. In
of such behavior is formed by individuals based on intuition, without the this study, we systematically investigated the role played by four cate­
need for extensive information to support decision-making and without gories of social interaction factors - interpersonal relationships, public
the need for in-depth information processing (Wolske et al., 2020). On relationships, active interactions, and passive interactions - and two
the other hand, investment-based energy efficiency requires a certain categories of individual decision characteristics - perceived value and
level of investment cost, especially in the case of housing renovation, depth of information processing - in motivating different energy-saving
which can impose a significant financial burden on individuals. As a behaviors. This study provides a new theoretical perspective to explain
result, individuals need to gather a wealth of information through social how social interactions shape individual energy-saving behaviors and
interactions or other means before deciding whether or not to imple­ explore the non-economic incentives for energy conservation and
ment such energy-saving behaviors and make a well-thought-out deci­ emission reduction. The following main findings are obtained from this
sion (Klöckner and Nayum, 2016). study.
Further, information processing depth moderates in the “ACI to PVB”
path but not in the “ACI to PVC” path. As shown in Model 1 in Table 6, 1) Interpersonal relationships and public relationships have a signifi­
the influence coefficient of active interaction on PVB is 0.178 (p < cant positive effect on individual energy-saving behaviors. Specif­
0.001), while this coefficient is as high as 0.351 (p < 0.001) in PVC. This ically, interpersonal relationships are more likely to influence habit
suggests that individuals may have formed strong beliefs and perceived adjustments, while equipment upgrades and housing retrofits are
values about housing renovation during active interactions and that the more likely to affect public relationships. And as the difficulty of
influence of information processing is greatly diminished (Kumkale implementing energy-saving behaviors increases, interpersonal re­
et al., 2010; Stiff and Mongeau, 2016). In contrast, for equipment up­ lationships’ influence decreases, and public relationships’ influence
grades, the initially formed perceived value can be reinforced by in- increases in that order.

Table 6
Results of the multiple regression analysis.
Variables PVA PVB PVC

Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Model 1 Model 2 Model 3

0.144*** 0.143*** 0.080 0.178*** 0.176*** − 0.005 0.351*** 0.347*** 0.424***


ACI
(0.000) (0.000) (0.210) (0.000) (0.000) (0.936) (0.000) (0.000) (0.001)
0.207*** 0.207*** 0.065 0.148*** 0.147*** 0.210** 0.123*** 0.110*** − 0.357*
PAI
(0.000) (0.000) (0.508) (0.000) (0.000) (0.021) (0.001) (0.003) (0.060)
0.005 − 0282** 0.031** − 0.108 0.065** − 0.361*
IDX
(0.744) (0.042) (0.017) (0.348) (0.010) (0.078)
0.021 0.052*** − 0.018
ACI × IDX
(0.289) (0.003) (0.536)
0.046 − 0.016 0.115**
PAI × IDX
(0.131) (0.532) (0.013)
2.712*** 2.693*** 3.567*** 2.347*** 2.258*** 2.722*** 1.836*** 1.645*** 3.348***
Constant
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.002)
2
R 0.116 0.116 0.120 0.198 0.202 0.209 0.337 0.342 0.346
Adjusted R2 0.108 0.107 0.109 0.191 0.194 0.200 0.331 0.335 0.338
F 14.665*** 13.198*** 11.397*** 27.621*** 25.544*** 22.170*** 57.047*** 52.302*** 44.282***

Note: *P < 0.1, **P < 0.05, ***P < 0.01. IDX represents IDA, IDB, and IDC in the PVA, PVB, and PVC models.

9
H. Li et al. Energy Economics 118 (2023) 106515

2) Active interactions and passive interactions have a significant posi­ and promote energy conservation. Through official websites, govern­
tive effect on individual energy-saving behaviors. Specifically, habit ment departments can post information and policies about energy con­
adjustments are more likely to be influenced by passive interactions, servation and provide the public with knowledge and methods of energy
while equipment upgrades and housing retrofits are more likely to be conservation. Social media accounts can be used to interact with
affected by active interactions. As the difficulty of implementing Internet users, answer their questions about energy efficiency, and post
energy saving-behavior increases, the influence of passive interac­ the latest energy efficiency policies and information. In addition, gov­
tion decreases, and the influence of active interaction increases. ernment departments can organize online activities, such as organizing
3) Perceived value has a significant positive effect on individual energy- online lectures and online competitions, to attract more public partici­
saving behavior, and the impact of perceived value deepens as the pation in energy efficiency promotion activities.
difficulty of implementing energy-saving behavior increases. At the Managers should be aware that social interaction is not limited to
same time, perceived value plays a mediating role in the influence individuals. The connection of individuals with various organizations
path of social interactions on individual energy-saving behaviors. and institutions is also an essential element of social interaction. Public
4) There is a significant positive moderating effect of information pro­ relationships are crucial to the dissemination of energy-saving in­
cessing depth on two paths: the path of active interactions on the tentions and behaviors. At the macro level, building a good government
perceived value of equipment upgrades and the way of passive in­ image and credibility is the most fundamental guarantee of enhancing
teractions on the perceived value of housing retrofits. the willingness of the whole society to save energy and reduce emis­
sions. At the mesoscopic level, enhancing consumer trust in energy
6. Discussions supply and service companies relies on continuous optimization of the
business environment. At the micro level, the role of the community as
6.1. Practical implications an intermediary between the government and individuals can be
leveraged to implement energy interventions in a more targeted and
Social interaction has a significant role in promoting individual en­ embedded manner (Aiken et al., 2017).
ergy saving. Building positive and effective social interaction requires
individuals to expand the breadth and depth of their social interaction
and the government, communities, and enterprises to participate in it 6.2. Limitations and future research directions
actively and serve and guide the public’s social interaction. The impact
of social interactions on individual energy savings outweighs economic This study focuses on the effects of various social interaction factors
and technological factors in many scenarios (Nolan et al., 2008; Hand­ on individual energy-saving behavior. These factors become essential
graaf et al., 2013). Efforts to improve particular energy demands and for building a more flexible, low-carbon, sustainable energy demand
stimulate individual energy-saving potential rely on the role of social model. However, the effect of individual-level influencing factors at the
interactions in shaping individual energy-saving behaviors. In­ group level needs further investigation.
terventions to promote energy savings may be ineffective if the complex When rising from the individual microscopic level to the macro­
social interactions behind individual energy use behaviors are not scopic group level, the effect of social interaction is not a simple linear
considered (Outcault et al., 2018). Economic incentives such as time-of- superposition. Adding some incentives may improve individual energy-
use tariffs (Yang et al., 2022), tiered tariffs (Sun et al., 2020), demand saving levels, but whether it will also positively impact the group level
response (Stoll et al., 2014), and energy efficiency subsidies (Dong et al., needs further verification. A study based on the dynamics of opinion
2020) have been a meaningful way to promote energy efficiency at this transmission found that the impact of social interactions will fail when
stage. But these approaches have apparent shortcomings in applicability the behavior or willingness differs too much between individuals (Zheng
and sustainability (Yin and Shi, 2021). Policymakers and researchers et al., 2019). The same phenomenon may exist in the field of individual
have been working to find social interaction mechanisms influencing energy saving. That is, after a significant increase in the level of energy
individual energy savings to build more flexible, low-carbon, and sus­ savings of one individual, other individuals who were initially influ­
tainable unique energy demand patterns (Stern et al., 2016). enced by it may engage in social interactions negatively due to the in­
Effective and frequent social interactions between individuals crease in behavioral differences between them.
depend on an extensive and stable social interaction network. An So far, there is a paucity of studies focusing on the influence of in­
interpersonal network based on daily life obviously cannot achieve dividual energy-saving motivation on the emergence of group energy-
efficient diffusion of energy-saving intentions and behaviors. Strong ties saving. In future research, we will build a social interaction network
account for a large proportion of interpersonal networks based on daily model of macro groups based on micro-level individual energy-saving
life, while weak ties account for a small proportion. The heterogeneous influencing factors. In turn, we will explore how changes in local indi­
information brought by weak ties contains a lot of knowledge that can vidual behavior affect the overall network status and identify the social
help individuals practice energy-saving behaviors, and can help in­ network’s key individuals and relationship structures that influence the
dividuals cross social structure and class boundaries to obtain more propagation of energy-saving intentions. The above efforts are crucial
effective information. So that individuals can transmit important in­ for precisely implementing energy-saving programs and policies.
formation from other groups to individuals outside the group. In In addition, the specific performance of groups with illnesses and
contrast, strong ties among individuals imply a greater degree of overlap disabilities in terms of energy use is often overlooked. Limited by the
in social circles and a higher degree of similarity and overlap in the research method of the online questionnaire survey, this paper is unable
information and resources available to individuals within the group. to investigate the special groups with diseases or disabilities. It is
Therefore, the heterogeneity of information resources obtained through necessary to design surveys to understand the energy behavior of this
strong ties is low (Yin and Shi, 2021; Fronczak et al., 2022). special group with illnesses and disabilities and to help them use energy
Government departments should build various online or face-to-face more equitably.
social interaction platforms to guide the public to actively and proac­
tively discuss energy-saving policies and programs. It also gives full play Credit author statement
to the critical role of opinion leaders in information dissemination (Kraf-
Todd et al., 2018) to help the public access energy efficiency information Mr. Hao Li designed the methodology and wrote the manuscript.
in a timely and effective manner and shorten the energy efficiency Prof. Wang Zhaohua reviewed and edited the manuscript. Dr. Zhang Bin
decision-making cycle. Specifically, government departments can conceived and designed the study. All authors read and approved the
establish official websites and social media accounts to interact socially manuscript.

10
H. Li et al. Energy Economics 118 (2023) 106515

Acknowledgment Duesenberry, J.S., 1949. Income, Saving, and the Theory of Consumer Behavior. Harvard
University Press.
Eric, W.K. See-To, Kevin, K.W. Ho, 2014. Value co-creation and purchase intention in
This study is supported by National Natural Science Fund of China social network sites: the role of electronic word-of-mouth and trust-a theoretical
(Reference No. 72222017, 72174023, 72141302, 71774014, 72140002, analysis. Comput. Hum. Behav. 31, 182–189.
91746208), Key Research Projects of Philosophy and Social Sciences of Festinger, L.A., 1954. A theory of social comparison processes. Hum. Relat. 7 (2),
117–140.
China Ministry of Education (Reference No. 21JZD027), Beijing Social Fronczak, A., Mrowinski, M.J., Fronczak, P., 2022. Scientific success from the perspective
Science Foundation Project (Reference No. 20JCC108, 20JJC023), Na­ of the strength of weak ties. Sci. Rep. 12, 5074.
tional Social Science Fund of China (Reference No. 17ZDA065), Science Granovetter, M., 1973. The strength of weak ties. Am. J. Sociol. 78 (6), 1360–1380.
Granovetter, M., 1985. Economic action and social structure: the problem of
and Technology Innovation Program for Innovative Talents in Beijing embeddedness. Am. J. Sociol. 91 (3), 481–510.
Institute of Technology (Reference No. 2021CX0125), Joint Develop­ Guo, M., Yang, N., Wang, J., et al., 2021. How do structural holes promote network
ment Program of Beijing Municipal Commission of Education, and Sci­ expansion? Technol. Forecast. Soc. Chang. 173.
Handgraaf, M.J.J., et al., 2013. Public praise vs. private pay: effects of rewards on energy
ence and Technology Project of the Ministry of Housing and Urban- conservation in the workplace. Ecol. Econ. 86, 86–92.
Rural Development of the People’s Republic of China (Reference No. Hargreaves, T., Middlemiss, L., 2020. The importance of social relations in shaping
2021-K-106). energy demand. Nat. Energy 5 (3), 195–201.
He, X., Hu, Y., 2021. Understanding the role of emotions in consumer adoption of electric
vehicles: the mediating effect of perceived value. J. Environ. Plan. Manag. 18, 1–21.
Appendix A. Supplementary data Henseler, J., Ringle, C.M., Sarstedt, M., 2015. A new criterion for assessing discriminant
validity in variance-based structural equation modeling. J. Acad. Mark. Sci. 43 (1),
115–135.
Supplementary data to this article can be found online at https://doi.
Hess, D.J., 2014. Smart meters and public acceptance: comparative analysis and
org/10.1016/j.eneco.2023.106515. governance implications. Health Risk Soc. 16, 243–258.
Hogg, M.A., 2016. Social identity theory. In: McKeown, S., Haji, R., Ferguson, N. (Eds.),
References Understanding Peace and Conflict through Social Identity Theory. Peace Psychology
Book Series, Springer, Cham.
Hoppe, T., 2012. Adoption of innovative energy systems in social housing: lessons from
Aiken, G.T., Middlemiss, L., Sallu, S., Hauxwell-Baldwin, R., 2017. Researching climate eight large-scale renovation projects in Te Netherlands. Energy Policy 51, 791–801.
change and community in neoliberal contexts: an emerging critical approach. Wiley Inhoffen, J., Siemroth, C., Zahn, P., 2019. Minimum prices and social interactions:
Interdiscip. Rev. Clim. Chang. 8 (4), e463. evidence from the German renewable energy program. Energy Econ. 78 (FEB),
Allcott, H., 2011. Social norms and energy conservation. J. Public Econ. 95 (9–10), 350–364.
1082–1095. James, L.R., Brett, J.M., 1984. Mediators, moderators, and tests for mediation. J. Appl.
Allen, S., Dietz, T., McCright, A.M., 2015. Measuring household energy efficiency Psychol. 2, 307–321.
behaviors with attention to behavioral plasticity in the United States. Energy Res. Jamieson, L., 2016. Families, relationships and environment: (un)sustainability, climate
Soc. Sci. 10, 133–140. change and biodiversity loss. Fam. Relatsh. Soc. 5, 335–355.
Aronson, E., O’Leary, M., 1983. The relative effectiveness of models and prompts on Kahneman, D., 2011. Thinking, Fast and Slow. Farrar, Straus and Giroux, New York.
energy conservation: a field experiment in a shower room. J. Environ. Syst. 12, Keller, R.R., Mckean, J.R., Peterson, R.D., 1982. Preference and value formation: a
219–224. convergence of enlightened orthodox and institutional analysis? J. Econ. Issues 16
Ayres, I., Raseman, S., Shih, A., 2013. Evidence from two large field experiments that (4), 941–954.
peer comparison feedback can reduce residential energy usage. J. Law Econ. Org. 29, Klöckner, C.A., 2014. The dynamics of purchasing an electric vehicle–a prospective
992–1022. longitudinal study of the decision-making process. Transport. Res. F: Traffic Psychol.
Bandura, A., Walters, R.H., 1977. Social Learning Theory. Prentice Hall, Englewood Behav. 24, 103–116.
Cliffs. Klöckner, C.A., Nayum, A., 2016. Specific barriers and drivers in different stages of
Baron, R.M., Kenny, D.A., 1986. The moderator-mediator variable distinction in social decision-making about energy efficiency upgrades in private homes. Front. Psychol.
psychological research:conceptual, strategic, and statistical considerations. J. Pers. 7, 1362.
Soc. Psychol. 6, 1173–1182. Kraf-Todd, G.T., Bollinger, B., Gillingham, K., et al., 2018. Credibility-enhancing displays
Barr, S., Gilg, A.W., Ford, N., 2005. The household energy gap: examining the divide promote the provision of non-normative public goods. Nature 563, 245–248.
between habitual-and purchase-related conservation behaviours. Energy Policy 33 Kumkale, G.T., Albarracín, D., Seignourel, P.J., 2010. The effects of source credibility in
(11), 1425–1444. the presence or absence of prior attitudes: implications for the design of persuasive
Berger, J., 2014. Word of mouth and interpersonal communication: a review and communication campaigns. J. Appl. Soc. Psychol. 40 (6), 1325–1356.
directions for future research. J. Consum. Psychol. 24, 586–607. Lane, B.W., et al., 2018. All plug-in electric vehicles are not the same: predictors of
Bohner, G., et al., 2003. When small means comfortable: relations between product preference for a plug-in hybrid versus a battery-electric vehicle. Transp. Res. D.
attributes in two-sided advertising. J. Consum. Psychol. 13 (4), 454–463. Transp. Environ. 65, 1–13.
Bollinger, B., Gillingham, K., 2012. Peer effects in the diffusion of solar photovoltaic Lechner, C., Frankenberger, K., Floyd, S.W., 2010. Task contingencies in the curvilinear
panels. Mark. Sci. 31 (6), 900–912. relationships between intergroup networks and initiative performance. Acad.
Bollinger, B., Gillingham, K., Lamp, S., Tsvetanov, T., 2019. Promotional Campaign Manag. J. 53, 865–889.
Duration and Word-of-Mouth in Durable Good Adoption. NYU Stern School of Leibenstein, H., 1950. Bandwagon, snob, and Veblen effects in the theory of consumers’
Business, SSRN. demand. Q. J. Econ. 64 (2), 183–207.
Boudet, N.M., Ardoin, J., Flora, K.C., Armel, M., Desai, T.N., 2016. Robinson, effects of a Luo, X., Xi, L., 2012. On the opinion leader of customer word-of-mouth communication
behaviour change intervention for girl scouts on child and parent energy-saving network based on the social network analysis approach. Manag. Rev. 24 (1), 75–81.
behaviours. Nat. Energy 1, 16091. Manski, C.F., 2000. Economic analysis of social interactions. J. Econ. Perspect. 14 (3),
Burkitt, I., 2014. Emotions and Social Relations. SAGE. 115–136.
Burningham, K., Venn, S., 2017. Are lifecourse transitions opportunities for moving to Mehrabian, A., Russell, J.A., 1974. An Approach to Environmental Psychology. MIT
more sustainable consumption? J. Consum. Cult. 20, 102–121. Press, Cambridge.
Carlsen, H.B., Toubl, J., Brincker, B., 2020. On solidarity and volunteering during the Neumann, J.V., Morgenstern, O., 1944. Theory of Games and Economic Behavior.
COVID-19 crisis in Denmark: the impact of social networks and social media groups Princeton University Press, Princeton.
on the distribution of support. Eur. Soc. 23, 122–140. Nicholls, L., Strengers, Y., 2015. Peak demand and the ‘family peak’ period in Australia:
Chen, B., Xie, W., Huang, F., et al., 2020. Energy-saving and pricing decisions in a understanding practice (in) flexibility in households with children. Energy Res. Soc.
sustainable supply chain considering behavioral concerns. PLoS One 15 (8), Sci. 9, 116–124.
e0236354. Nolan, J.M., et al., 2008. Normative social influence is underdetected. Personal. Soc.
Crossley, N., 2010. Towards Relational Sociology. Routledge. Psychol. Bull. 34 (7), 913–923.
Cui, Y., Li, J., Zhang, Y., et al., 2022. The impacts of game experience and fanwork Osei-Frimpong, K., McLean, G., 2018. Examining online social brand engagement: a
creation on game loyalty: mediation effect of perceived value. Technol. Forecast. social presence theory perspective. Technol. Forecast. Soc. Chang. 128, 10–21.
Soc. Chang. 176. Osgood, C.E., Tannenbaum, P.H., 1955. The principle of congruity in the prediction of
Darby, S., 2010. Smart metering: what potential for householder engagement? Build. attitude change. Psychol. Rev. 62 (1), 42–55.
Res. Inf. 38, 442–457. Ou, X.C., Pavlou, P.A., Davison, R.M., 2014. Swift guanxi in online marketplaces: the role
Dodds, W.B., Monroe, K.B., Grewal, D., 1991. Effects of price, brand, and store of computer-mediated communication technologies. MIS Q. 38 (1), 209–230.
information on buyers’ product evaluations. J. Mark. Res. 28 (3), 307–319. Outcault, S., Sanguinetti, A., Pritoni, M., 2018. Using social dynamics to explain uptake
Donald, I.J., Cooper, S.R., Conchie, S.M., 2014. An extended theory of planned behavior in energy saving measures: lessons from space conditioning interventions in Japan
model of the psychological factors affecting commuters’ transport mode use. and California. Energy Res. Soc. Sci. 45, 276–286.
J. Environ. Psychol. 40, 39–48. Palm, A., 2017. Peer effects in residential solar photovoltaics adoption—a mixed
Dong, X., Zhang, B., Wang, B., et al., 2020. Urban households’ purchase intentions for methods study of Swedish users. Energy Res. Soc. Sci. 26, 1–10.
pure electric vehicles under subsidy contexts in China: do cost factors matter? Park, E., Kwon, S.J., 2017. What motivations drive sustainable energy-saving behavior?
Transp. Res. A Policy Pract. 135, 183–197. An examination in South Korea. Renew. Sust. Energ. Rev. 79 (11), 494–502.

11
H. Li et al. Energy Economics 118 (2023) 106515

Parker, P., Rowlands, I.H., Scott, D., 2003. Innovations to reduce residential energy use Truelove, H.B., Gillis, A.G., 2018. Perception of pro-environmental behavior. Glob.
and carbon emissions: an integrated approach[J]. Canad. Geogr./Le Géographe Environ. Chang. 49, 175–185.
canadien 47 (2), 169–184. Turner, J.B., Turner, R.J., 2013. Social Relations, Social Integration, and Social Support.
Powells, G., Bulkeley, H., Bell, S., Judson, E., 2014. Peak electricity demand and the Springer, Dordrecht, pp. 341–356.
flexibility of everyday life. Geoforum 55, 43–52. Van Raaij, W.F., Verhallen, T.M.M., 1983. A behavioral model of residential energy use.
Rai, V., Robinson, S.A., 2013. Effective information channels for reducing costs of J. Econ. Psychol. 3, 39–63.
environmentally friendly technologies: evidence from residential PV markets. Veblen, T., 1899. The Theory of the Leisure Class. Macmillan, George Allen and Unwin,
Environ. Res. Lett. 8 (1), 014044. London.
Ren, H., Huang, T., 2018. Modeling customer bounded rationality in operations Wade, F., Shipworth, M., Hitchings, R., 2016. Influencing the central heating
management: a review and research opportunities. Comput. Oper. Res. 91, 48–58. technologies installed in homes: the role of social capital in supply chain networks.
Rosenow, J., Eyre, N., 2016. A postmortem of the green deal: austerity, energy efficiency, Energy Policy 95, 52–60.
and failure in British energy policy. Energy Res. Soc. Sci. 21, 141–144. Wang, J., Liu, B., Tao, H., 2007. Evolution of entrepreneur social networks in industrial
Sánchez, C.G., De-Pablos-Heredero, C., Medina-Merodio, J.A., et al., 2021. Relationships clusters. Sci. Sci. Manag. S.&.T. 28 (4), 169–174.
among relational coordination dimensions: impact on the quality of education online Wang, J.Q., Long, R.Y., Chen, H., et al., 2022. How do parents and children promote each
with a structural equations model. Technol. Forecast. Soc. Chang. 166 (1), 120608. other? The impact of intergenerational learning on willingness to save energy.
Savelli, I., Morstyn, T., 2021. Better together: harnessing social relationships in smart Energy Res. Soc. Sci. 87, 102465.
energy communities. Energy Res. Soc. Sci. 78 (3–4), 102125. Wilde, M.D., 2019. The sustainable housing question: on the role of interpersonal,
Schultz, P.W., Nolan, J.M., Cialdini, R.B., Goldstein, N.J., Griskevicius, V., 2007. The impersonal and professional trust in low-carbon retrofit decisions by homeowners.
constructive, destructive, and reconstructive power of social norms. Psychol. Sci. 18, Energy Res. Soc. Sci. 51, 138–147.
429–434. Wolske, K.S., Gillingham, K.T., Schultz, P., 2020. Peer influence on household energy
Schwartz, S.H., 1977. Normative influences on altruism. In: Berkowitz, L. (Ed.), behaviours. Nat. Energy 5 (3), 202–212.
Advances in Experimental Social Psychology. Academic Press, San Diego, Woodruff, R.B., 1997. Customer value: the next source for competitive advantage.
pp. 221–279. J. Acad. Mark. Sci. 25 (2), 139–153.
Simon, H.A., 1977. The New Science of Management Decision. Prentice Hall PTR. Wu, C., Yao, H., Ning, X., et al., 2021. Emergence of informal safety leadership: a
Sintov, N.D., White, L.V., Walpole, H., 2019. Thermostat wars? The roles of gender and social–cognitive process for accident prevention. Prod. Oper. Manag. 30 (11),
thermal comfort negotiations in household energy use behavior. PLoS One 14 (11), 4288–4305.
e0224198. Yang, T., Zhao, W., Zhou, M., 2018. The impact of network size on access to
Sovacool, B.K., Kivimaa, P., Hielscher, S., Jenkins, K.E.H., 2017. Vulnerability and entrepreneurial resources: moderating role of entrepreneurs’ previous experience.
resistance in the United Kingdom’s smart meter transition. Energy Policy 109, Sci. Technol. Progr. Pol. 35 (2), 1–9.
767–781. Yang, S.X., Nie, T., Li, C.C., et al., 2022. Research on the contribution of regional energy
Stern, P.C., et al., 2016. Opportunities and insights for reducing fossil fuel consumption internet emission reduction considering time-of-use tariff. Energy 239, 122170.
by households and organizations. Nat. Energy 1, 16043. Yin, J., Shi, S., 2021. Social interaction and the formation of residents’ low-carbon
Stiff, J.B., Mongeau, P.A., 2016. Persuasive communication. Guilford Press, New York. consumption behaviors: an embeddedness perspective. Resour. Conserv. Recycl. 164
Stoll, P., Brandt, N., Nordström, L., 2014. Including dynamic CO2 intensity with demand (1), 105116.
response. Energy Policy 65, 490–500. Zeithaml, V.A., 1988. Consumer perception of price, quality and value: a means-end
Sun, Y., Wang, Z., Zhang, B., et al., 2020. Residents’ sentiments towards electricity price model and synthesis of evidence. J. Mark. 52 (3), 2–22.
policy: evidence from text mining in social media. Resour. Conserv. Recycl. 160, Zhang, W., Yang, X., Wang, Q., et al., 2015. Investigation on the factors determining
104903. consumers’ use of online intermediated shopping (OIS): a behavioral intention
Sun, S., Zhang, J., Zhu, Y., et al., 2022. Exploring users’ willingness to disclose personal perspective. J. Organiz. End User Comp. (JOEUC) 27 (1), 77–97.
information in online healthcare communities: the role of satisfaction. Technol. Zheng, J.J., Zhang, B., Cheng, Y., et al., 2019. Group choice behavior in green travel
Forecast. Soc. Chang. 178. based on scale-free network. Chin. J. Manag. Sci. 27 (10), 198–208.
Sussman, R., et al., 2013. The effectiveness of models and prompts on waste diversion: a
field experiment on composting by cafeteria patrons. J. Appl. Soc. Psychol. 43,
24–34.

12

You might also like