Resources, Conservation & Recycling 134 (2018) 91–99

Contents lists available at ScienceDirect

Resources, Conservation & Recycling

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

Full length article

Exploring the effects of normative factors and perceived behavioral control T

on individual’s energy-saving intention: An empirical study in eastern China
⁎
Xingjun Rua, Shanyong Wangb, , Shuai Yana
a
School of Management, Hangzhou Dianzi University, Hangzhou, Zhejiang Province, 310018, PR China
b
School of Management, University of Science and Technology of China, Hefei, Anhui Province, 230026, PR China

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

Keywords: Previous researches have explored the critical determinants that influence individual’s energy-saving intention.
Extended theory of planned behavior However, these researches rarely focus on individual’s energy-saving intention in developing countries and
Subjective injunctive norm limited research has been conducted to explore the combined effects of normative factors and perceived be-
Subjective descriptive norm havioral control. In this study, an extended theory of planned behavior (TPB) model is employed to narrow this
Personal moral norm
gap. The model is empirically tested using questionnaire survey data collected from 450 respondents in eastern
Energy-saving intention
China. The findings reveal that perceived behavioral control is the most decisive factor for individual’s energy-
saving intention. Meanwhile, attitude towards energy-saving and personal moral norm are also important fac-
tors. Two interaction terms (i.e., perceived behavioral control and subjective injunctive norm, and perceived
behavioral control and subjective descriptive norm) are negatively associated with energy-saving intention,
which implies that social norm plays an important role in motivating energy-saving behavior in the Chinese
context. Furthermore, social norm increases the energy-saving intention of individuals who exhibit low per-
ceived behavioral control. Based on the results, implications for improving individual’s energy-saving intention,
limitations of the study and suggestions for further research are discussed.

1. Introduction to predict and interpret different pro-environmental behavioral inten-

tions and behaviors, such as energy-saving behavior (Ajzen, 1991; Chen
Energy conservation is an active effort to reduce human impacts on and Knight, 2014; Gao et al., 2017; Sarkis, 2017; D’Oca et al., 2018).
the environment (Park and Kwon, 2017). In this connection, govern- Moreover, TPB is more suitable for comprehending consumer energy
ments worldwide have launched various schemes to address the issue efficiency behaviors, such as private energy conservation behavior
and promote energy conservation practices among individuals. In this (Sarkis, 2017). TPB comprises attitude, subjective norm, and perceived
regard, governments around the world have launched various programs behavioral control (De Leeuw et al., 2015; Chen et al., 2016). TPB holds
to deal with this problem and facilitate individual’s energy-saving be- that an individual’s behavioral intention is determined by positive as-
havior (Yue et al., 2013; Du et al., 2017). Intervention measures such as sessment of behavior (attitude), social pressure advocating behavior
commitment, goal setting, information, and modeling would affect (subjective norm) and perceived ease of implementing specific behavior
energy effectiveness (Abrahamse et al., 2005). A large number of these (perceived behavioral control). Several studies have adopted TPB to
measures rely on the thinking of transforming energy-saving practices empirically analyze some pro-environmental behaviors, such as electric
into a more normalized activity and encouraging people to participate vehicles adoption behavior (Shi et al., 2017b), recycling behavior at
in energy-saving practices regularly (Hori et al., 2013). It is essential for home (Greaves et al., 2013), general pro-environmental behavior (PEB)
scholars and policymakers to get a thorough understanding of in- in workplace (Blok et al., 2015), energy efficiency behavior of in-
dividual’s intention to save energy in daily life. dividuals in large organizations (Chen and Knight, 2014), and solar
Energy-saving has attracted wide attention, and several theoretical water heaters and alternative fuel vehicles adoption behavior (Chen
frameworks have been used to explore individual’s energy-saving in- et al., 2016). However, some studies showed that subjective norm does
tention, such as social cognitive theory, the drivers-needs-actions-sys- not always lead to more pro-environmental behavior when considering
tems framework, and the theory of planned behavior (TPB) (D’Oca the different degree of perceived behavioral control (Castanier et al.,
et al., 2017). Among these theory frameworks, TPB is commonly used 2013; Shi et al., 2017b). Donald et al. (2014) suggested that subjective

⁎
Corresponding author at: School of Management, University of Science and Technology of China, No. 96, Jinzhai Road, Hefei, Anhui Province, 230026, PR China.
E-mail address: [email protected] (S. Wang).

https://doi.org/10.1016/j.resconrec.2018.03.001
Received 18 September 2017; Received in revised form 28 February 2018; Accepted 1 March 2018
Available online 22 March 2018
0921-3449/ © 2018 Elsevier B.V. All rights reserved.
X. Ru et al. Resources, Conservation & Recycling 134 (2018) 91–99

norm failed to influence the intention to drive. The relationship be- various contexts, such as recycling behavior at work, general PEB in
tween subjective norm and PEB may vary depending on the context and workplace, and green products purchasing behavior. Similarity, in the
the behavior studied (Donald et al., 2014; Yazdanpanah and Forouzani, context of energy-saving behavior, it can be assumed that if an in-
2015). dividual believes that energy-saving is relevant and helpful to reduce
To improve TPB's predictive power, scholars tried to incorporate carbon emissions and improve the environment, he will maintain a
additional determinants in the TPB, such as personal moral norms, and positive attitude and may form an energy-saving behavioral intention.
divide specific fundamental determinants, for example, divide sub- Thus, it is hypothesized that:
jective norms into the subjective injunctive norm and subjective de-
H1. Attitude towards energy-saving positively affects individual’s
scriptive norm (De Leeuw et al., 2015). In previous studies, personal
intention to save energy.
moral considerations (personal moral norm) and two types of social
pressure (subjective injunctive norm and subjective descriptive norm)
were confirmed to be critical factors influencing individual’s pro-en- 2.2. Subjective norm
vironmental behavioral intentions (Wang et al., 2014; De Leeuw et al.,
2015; Shi et al., 2017a). However, previous studies rarely explored Subjective norm is defined initially as individual’s perception that
individual’s energy conservation behavior in the Chinese context and most people who are essential to him consider he should or should not
considered the personal moral norm and two types of the social norm. perform a behavior (Ajzen, 1991). Fishbein and Ajzen (2011) argued
Furthermore, it is still not clear whether there is a difference in energy that subjective norm should consist of two components: subjective in-
conservation behavior when considering the varying degrees of nor- junctive norm (SIN) and subjective descriptive norm (SDN). Subjective
mative factors and perceived behavioral control. Thus, it is crucial to injunctive norm refers to behaviors commonly approved or dis-
develop a more thorough understanding of what motivates Chinese approved, which is equivalent to the subjective norm in TPB. Subjective
individual’s energy conservation behavior, which has significant im- descriptive norm refers to behaviors demonstrated by essential re-
plications to deal with the increasingly serious energy overuse problems ferents in a given social environment (Wang et al., 2016). Individual
in China. likes to comply with the expectations and behaviors of significant
The rest of this paper is organized as follows. We will first review others. Blok et al. (2015) indicated that the demonstrative behaviors of
the previous literature and outline the research hypotheses in Section 2. managers could motivate employee’s PEB in Netherland. The stronger
In Section 3, the data and research methodology are presented. Data expectations and behaviors perceived by the significant others, the
analysis and results are introduced based on structural equation ana- more likely of the individual to engage in behavior (Greaves et al.,
lysis in Section 4, followed by the discussion and implication in Section 2013; Shi et al., 2017a). In other words, with the higher SIN and SDN
5. In the final section, the findings and policy implications for moti- perceived, individuals would be more likely to conduct the behavior
vating individual’s energy-saving intention, limitations of the study and (De Leeuw et al., 2015). This situation is also suitable for energy-saving
suggestions for further research are discussed. behavior. If individuals realize that essential referents (e.g., family
members, great friends, celebrities) think he ought to save energy in
2. Literature review and research hypotheses daily life, or essential referents conduct energy-saving behavior as a
role model, he will perceive pressures and intend to save energy. Thus,
Individual’s energy-saving intention refers to the individual’s self- it is hypothesized that:
commitment to participate in energy-saving behaviors (Yang et al.,
2016). Literature has analyzed individual’s energy-saving intention by H2. Subjective injunctive norm positively affects individual’s energy-
extracting variables and conceptualizing models from social psy- saving intention.
chology. The theory of planned behavior (TPB) is preferred for under- H3. Subjective descriptive norm positively affects individual’s energy-
standing individual’s energy-saving intention and behavior (Wang saving intention.
et al., 2016). According to this theory, individual’s behavioral intention
is decided by the attitude towards this behavior, subjective norm and
perceived behavioral control. Attitude refers to individual’s negative or 2.3. Perceived behavioral control
positive evaluation to conduct the particular behavior (Ajzen, 1991;
Wang et al., 2016). Subjective norm refers to the perception of social It is widely believed that perceived behavioral control is a crucial
pressure from essential others (Ajzen, 1991; Abrahamse and Steg, determinant of behavioral intention in many studies (De Leeuw et al.,
2013). Perceived behavioral control refers to the individual’s assess- 2015; Botetzagias et al., 2015; Lizin et al., 2017). Some external con-
ment of self-efficacy and perceived control ability in accomplishing the ditions, such as facility availability, time, cost, knowledge of the be-
behavior (Ajzen, 1991; Shi et al., 2017b). havior and skills, may be free from personal control, and thus affect
Although TPB is widely used, there is a problem with its in- their intention to participate in a specific practice. If individuals have a
completeness (Wang et al., 2016). Thus, some studies supported to in- more significant control over themselves, they will have a stronger
corporate additional factors into TPB to improve its validity, such as intention to accomplish a particular behavior (Webb et al., 2013). This
personal moral norm (Wang et al., 2016; Shi et al., 2017a) and sub- phenomenon is also suitable for individual energy-saving behavior
jective descriptive norm (De Leeuw et al., 2015). Furthermore, execu- (Wang et al., 2014). If individuals feel easy and have relevant knowl-
tive functions can also improve the predictive power of TPB in the field edge and skills to save energy in daily life, they will be more likely to
of conventional implementation of behavioral intentions (Allan et al., form the intention to save energy (Donald et al., 2014). So we come to
2011, Hall et al., 2008, Corradi et al., 2013). the next hypothesis:
H4. Perceived behavioral control positively affects individual’s energy-
2.1. Attitude saving intention.

Attitude is a critical determinant to influence behavioral intention

in TPB. If an individual likes a specific behavior, then the intention to 2.4. Personal moral norm
conduct such behavior will be high (Yang et al., 2016). Several studies,
such as Greaves et al. (2013), Blok et al. (2015), Yazdanpanah and To examine the effectiveness of TPB, many studies have expanded
Forouzani (2015) and Yadav and Pathak (2017) noted the significance TPB by incorporating additional factors, such as personal moral norm
of attitude in predicting individual’s pro-environmental behavior in (Wang et al., 2016; Shi et al., 2017a). Personal moral norm refers to the

92
X. Ru et al. Resources, Conservation & Recycling 134 (2018) 91–99

Fig. 1. The conceptual model.

personal belief that conducting in a particular way is right or wrong Abrahamse et al. (2009) indicated that perceived behavioral control
(Schwartz, 1977). As a normative indicator, personal moral norm is negatively moderated the relationship between personal moral norm
considered as a social pressure as the subjective norm does in some and car use intention for commuting. Shi et al. (2017a) have drawn the
studies; however, what it stresses is the aspects of individual inter- similar conclusions. Castanier et al. (2013) indicated that the interac-
nalization and emotion (Bamberg et al., 2007; Shi et al., 2017a). It tion of subjective norm and self-efficacy negatively influences drunken
relates to a feeling of obligation that people hold, which leads to per- driving behavior, whereas the interaction of subjective norm and per-
form the specific actions (Wang et al., 2016). Personal moral norm ceived behavioral control positively influences drunken driving beha-
plays a vital role in the field of PEB (Shi et al., 2017a). Prior studies vior. Shi et al. (2017b) demonstrated that perceived behavioral control
have added personal moral norm into TPB theoretical framework. Some positively moderated the relationship between subjective norm and
empirical studies based on extended TPB found that personal moral electric vehicles purchase intention. It implies that the interaction of
norm as an additional predictor increased the variance to explain PEB perceived behavioral control and normative factors influence PEB sig-
intention (Bamberg et al., 2007; Shi et al., 2017b). Abrahamse and Steg nificantly. Differences in research conclusions may come from the
(2009) argued that high personal moral norm will improve households’ specific research questions (Castanier et al., 2013). In the current study,
energy usage. Thus, it is reasonable to suppose that individual with high we attempt to explore the combined effects of perceived behavioral
personal moral norm will have a more moral obligation and responsi- control and normative factors on individual’s energy-saving intention.
bility to save energy. So we propose the hypothesis: It will contribute to comprehend the interactive mechanisms of TPB
determinants and provide suggestions for policymakers. Thus, it is hy-
H5. Personal moral norm positively affects individual’s energy-saving
pothesized that:
intention.
H6a. The combined effect of the subjective injunctive norm and
perceived behavioral control significantly influence individual’s
2.5. Combined effects of perceived behavioral control and normative factors energy-saving intention.

Normative factors in recent studies contain two categories: extrinsic H6b. The combined effect of the subjective descriptive norm and
normative factor and intrinsic normative factor. In TPB, social influence perceived behavioral control significantly influence individual’s
is expressed as the notion of subjective norm (Wang et al., 2016). It energy-saving intention.
emphasizes the perceived pressure from essential referents who influ- H6c. The combined effect of the personal moral norm and perceived
ence individual’s behavior (Ajzen, 1991). The two components of sub- behavioral control significantly influence individual’s energy-saving
jective norm have different focus. Subjective injunctive norm focuses on intention.
individual’s perception of what significant others (e.g., family members,
friends, and celebrities) think they should do, whereas subjective de- Fig. 1 displays the conceptual model to be examined in the current
scriptive norm focuses on the actions that these significant others’ have study.
done (De Leeuw et al., 2015). Both notions of the subjective injunctive
norm and subjective descriptive norm are derived from extrinsic social 3. Research methodology
pressure (Wang et al., 2016), so they are considered to be extrinsic
normative factors. Personal moral norm emphasizes on internalization, 3.1. Questionnaire
emotion and individual’s sense of responsibility, therefore, it is an in-
trinsic normative factor (Bamberg et al., 2007; Shi et al., 2017a). The measurement items employed in the current research, namely,
The direct relationships between energy-saving behavioral intention attitude, subjective injunctive norm, subjective descriptive norm, per-
and its determinants are demonstrated in most empirical studies (Xu ceived behavioral control, personal moral norm, and individual’s en-
et al., 2015; Yang et al., 2016; Sarkis, 2017). Nevertheless, the inter- ergy-saving intention were adopted from Webb et al. (2013), Wang
action of different determinants may be more complicated than the et al. (2016), Yang et al. (2016) (See Appendix A). All constructs were
studies have mentioned (Wan et al., 2017; Shi et al., 2017a). A few measured by using multi-item scales. For all measurement items, a
studies have indicated that normative factors do not necessarily lead to seven-point scale was employed to indicate the extent to which re-
more energy-saving intention or behavior when considering perceived spondents approve of these items, where 7 represents the most agree
obstacles and cost (Castanier et al., 2013; Shi et al., 2017b). In other and 1 represents the most disagree. Also, respondents were required to
words, individual’s perceived behavioral control may negatively impact provide demographic information in the questionnaire. At the begin-
the relationships between normative factors and behavioral intention. ning of the questionnaire, we introduced the research background and

93
X. Ru et al. Resources, Conservation & Recycling 134 (2018) 91–99

told respondents that this study is interested in their daily energy Table 1
conservation behaviors, and then gave an example of what specific Respondent profile (N = 450).
energy conservation activities involved. According to Greaves et al.
Characteristics Frequency Percent (%)
(2013) and Yang et al. (2016), four daily energy-saving activities are
selected, which are “switch off lights whenever leaving the room”, Gender Male 215 47.8
“switch off PCs whenever leaving the desk”, “switch off power supply of Female 235 52.2
Age Under 20 103 22.9
appliances whenever it’s not working”, “Set the appliances’ control to
21–30 127 28.2
save energy (for example, setting up proposed temperature for air- 31–40 90 20.0
conditioner)”. Finally, we declared that the questionnaire is only for 41–50 77 17.1
scientific research and ensured that the respondents’ answers are con- 51 and above 53 11.8
fidential. Education Junior middle school or 68 15.1
below
Three procedures were implemented to refine the questionnaire
Senior high school 93 20.7
items for improving measurement accuracy. First, items selected from Junior college or bachelor 202 44.9
prior studies were translated into Chinese sentences. Second, a focus degree
group including five professors and eight graduate students were in- master’s degree or PhD 87 19.3
Location Hangzhou 145 32.2
vited to check the Chinese sentences to ensure that they were prone to
Ningbo 106 23.6
read and comprehend. Hereafter, some wordings in the questionnaire Wenzhou 114 25.3
were adjusted based on respondents’ feedback. Third, the measurement Jinhua 85 18.9
items were assessed and improved in a pilot study to ensure the content Monthly Income Less than 3000 77 17.2
validity. (RMB) 3001–5000 130 28.9
5001–7000 131 29.1
More than 7001 112 24.8
3.2. Sample and procedure

The questionnaire survey was conducted online. Questionnaires validity and need to be checked. Thus, Harman’s one-factor test is im-
were sent to individuals who reside in 4 cities in Zhejiang province, plemented to examine the common method bias. The results show that
namely, Hangzhou, Ningbo, Jinhua, and Wenzhou. Zhejiang province is all the measurement items are divided into five factors with eigenvalues
one of the wealthiest areas in eastern China. Of course, along with the higher than 1.0, which explained 71.05% of the total variance. The first
high economic prosperity in Zhejiang province is the excessive energy factor explains only 26.92% of the variance, which is less than the
use, both in the household sector and industry domain, which has benchmark value of 50.0% (Harman, 1976). The results suggest that
brought severe environmental pollution problems (MEPC, 2016). In common method bias is unlikely to be a severe problem in this research.
fact, most of the respondents were urban middle-class citizens, because
the survey was conducted online. The questionnaires were collected
3.3. Statistical analysis
from January to March of 2017 by the company SoJump, which has a
vast database containing seventeen million authentic specimens from
The data analysis of this study was conducted using Structural
the internet. Additionally, the temperature is relatively lower in Zhe-
Equation Modeling (SEM) technique and followed the two-step ap-
jiang province from January to March. Thus people use air-conditioners
proach for assessing the measurement and structural models respec-
more frequently during this period. The respondents were recruited by
tively (Anderson and Gerbing, 1988). SEM technique is a commonly
SoJump from an online database of volunteers reside in Zhejiang pro-
used method to test relationships between constructs based on their
vince. SoJump requests volunteers to take part in a survey, and anyone
assigned indicators (Chin, 1998a). SEM is a powerful statistical research
on the roster allowed attending. Respondents were provided con-
technique, and is good at analyzing the relationships between multiple-
vertible credit which can be changed for money or used to buy goods
item constructs (Iacobucci, 2009). There are two principal statistical
and services. 1000 invitation letters with the URL of an electronic
methods for SEM test, namely, covariance-based SEM and variance-
version of the questionnaire were sent to volunteers in the target areas
based partial least squares (PLS) methods. Covariance-based SEM is
through e-mail, and 532 online responses were received. 82 invalid
suitable for theories’ verification or comparison, while PLS can be ap-
questionnaires with the same answer for most items, answer time is less
plied to complex structural equation models with a large number of
than 2 min or more than 5 min, and with logical errors were removed.
constructs. Furthermore, PLS is suitable for developing a theory with
450 valid responses were obtained finally. The respondent profile is
many relationships to be examined and does not require normal-dis-
shown in Table 1.
tributed input data (Reinartz et al., 2009; Hair et al., 2011). Con-
ANOVA test was adopted to explore whether there were significant
sidering PLS advantages mentioned above, this paper chooses to use
differences in the questionnaires among four regions. It is shown that
PLS to do an empirical study. With the help of statistical software
respondents’ energy-saving intention and all of its determinants are not
SmartPLS 3.2, we analyzed the collected survey data and tested the
significantly different among four regions (at 5% significance level).
hypotheses.
Thus, it is reasonable to incorporate the samples collected from the four
areas as a single one in the following data analysis. We also assessed the
potential non-response bias and Common Method Bias (CMB) of this 4. Data analysis and results
study. T-tests were conducted to compare the early respondents (those
who returned the completed questionnaires within ten days) and the There are two steps to accomplish PLS analysis, namely, measure-
late respondents (those who returned the completed questionnaires ment model and structural model (Chin, 1998b). The measurement
during the last ten days) (Armstrong and Overton, 1977). The results model is employed to examine the relationships between constructs and
suggest that there are no significant differences between these samples the corresponding items, and the structural model is adopted to test the
regarding gender, age, education, location, and income. Thus, it can be relationships between these constructs. In the following data analysis,
concluded that non-response bias is not a major concern in this re- we first tested the measurement model to evaluate the reliability and
search. Considering the mono-methodological nature of the study and validity of the constructs, and then we tested the structural model to
the item characteristic effects, common method bias may threaten the examine the relationships among the hypothesized constructs.

94
X. Ru et al. Resources, Conservation & Recycling 134 (2018) 91–99

Table 2 2015). The square root values of AVE for each latent variable are all
Confirmatory factor analysis. larger than the correlation between a pair of constructs, thus supporting
discriminant validity (Wang et al., 2016). Based on the above test results,
Construct Indicator Standardized Cronbach’s Composite AVE
loading alpha Reliability it is indicated that measurement model has a good performance of con-
struct reliability, convergent validity, and discriminant validity (Table 3).
Attitude ATT1 0.835*** 0.835 0.890 0.669
towards ATT2 0.811***
energy- ATT3 0.817*** 4.2. Structural equation modeling
saving ATT4 0.808***
(ATT)
Intention to ESI1 0.896*** 0.848 0.908 0.767 Structural equation model was adopted to explore the relationships
save ESI2 0.848*** between the constructs. The model is run by adopting a bootstrap re-
energy ESI3 0.882*** sampling routine with 450 collected data and 1000 subsamples. This
(ESI) bootstrap resampling routine produces random subsamples, which are
Subjective SDN1 0.878*** 0.812 0.889 0.727
descriptive SDN2 0.852***
conducted to evaluate the significance of hypotheses (Chin, 1998b).
norm SDN3 0.826*** Moreover, R2 values of 0.25, 0.50, and 0.75 are regarded as weak,
(SDN) moderate, and substantial, respectively in a structural model of PLS
Perceived PBC1 0.849*** 0.823 0.895 0.739 (Hair et al., 2011). Wetzels et al. (2009) suggested a single criterion of
behavioral PBC2 0.900***
goodness of fit (GoF) for PLS based on the average AVE and the average
control PBC3 0.829***
(PBC) R2. GoF score of 0.35, 0.50, and 0.61 is regarded as small, medium, and
Personal moral PMN1 0.911*** 0.902 0.938 0.836 large respectively (Latan and Ghozali, 2012). All four models are con-
norm PMN2 0.918*** sidered fit. The results are shown in Table 4.
(PMN) PMN3 0.913*** Model 1 tested the original TPB, included the original three vari-
Subjective SIN1 0.890*** 0.881 0.926 0.807
injunctive SIN2 0.889***
ables in TPB, and it is accounted for 43.7% of the variance in the en-
norm (SIN) SIN3 0.916*** ergy-saving intention. Model 1 showed that the three TPB variables
(i.e., attitude, subjective norm, and PBC) positively influenced energy-
***
Note: Significant at p < 0.001. saving intention.
Model 2 and Model 3 included two additional variables respec-
4.1. Measurement model analysis tively, namely, subjective descriptive norm and personal moral norm.
All paths in Model 2 and Model 3 were significant. By incorporating
Confirmatory factor analysis (CFA) was adopted to estimate the additional variables, two models accounted for 45.2% and 48% of the
model’s reliability and validity. Convergent validity and composite re- variance respectively in the energy-saving intention. Cohen (1988)
liability evaluate the correlation between the items within the latent suggested that the effect of a latent predictor could be evaluated
variables (Fornell and Larcker, 1981). As shown in Table 2, the adopting the following formula that is based on the value of R2 with the
minimum value of Cronbach’s alpha was 0.812, and the minimum value proposed predictor included and excluded from the model: f2 = (R2incl
of composite reliability was 0.889, which were all greater than the −R2excl)/(1- R2incl). The values 0.02, 0.15, and 0.35 could be regarded as
recommended minimum value of 0.700 (Hair et al., 2009), thus con- small, medium, or large effect, respectively. The value of effect (f2) for
structs’ reliability was confirmed. Furthermore, we examined the con- the two additional paths in Model 2 and Model 3 was 0.028 and 0.053,
vergent validity and discriminant validity of the constructs. Convergent which both indicated small-to-medium effect.
validity means the level to which two or more items of constructs Model 4 further included the combined effect of perceived beha-
theoretically relevant to each other. It was estimated by adopting the vioral control and normative factors based on Model 3. Model 4 ac-
factor loading on the corresponding construct and construct’s average counted for 51% of the variance in energy-saving intention after the
variance extracted (AVE) (Fornell and Larcker, 1981). The factor three interaction terms were included. The value of the effect (f2) for
loadings of constructs ranged from 0.808 to 0.918, which were all the interaction terms was 0.061, which represented a small-to-medium
larger than the suggested threshold of 0.700. The AVE scores varied effect. AIC and BIC value of Model 4 are both smaller than that of Model
from 0.669 to 0.836 and were all greater than the recommended 3. Furthermore, we run the Model 3 and Model 4 for additional 50
benchmark value of 0.500 (Chin, 1998b). These results suggested that times through bootstrap resembling routine with 1000 subsamples re-
all measurement items have a robust convergent validity. spectively. The results show that R2 of Model 3 and Model 4 range from
Discriminant validity means the extent to which two or more con- 0.471 to 0.492 and 0.504–0.518 respectively, which imply that Model 4
structs theoretically should not relevant to each other (De Leeuw et al., explains the data better than Model 3.

Table 3
Descriptive statistics and correlations.

Construct Convergence validity

Means SD ATT ESI SDN PBC PMN SIN

Attitude 5.387 0.514 0.818

Intention to save energy 5.527 0.561 0.544 0.876
Subjective descriptive norm 5.122 0.374 0.271 0.249 0.853
Perceived behavioral control 5.619 0.634 0.265 0.488 0.019 0.860
Personal moral norm 5.211 0.623 0.574 0.527 0.190 0.301 0.914
Subjective injunctive norm 5.578 0.493 0.569 0.409 0.176 0.152 0.504 0.898

Note: Means are measured based on average factor scores; SD means standard deviation; Values in the diagonal row (bold) are the square roots of AVEs and the others are the correlations
between constructs.

95
X. Ru et al. Resources, Conservation & Recycling 134 (2018) 91–99

Table 4
Hypotheses testing results.

Construct Model1 Model2 Model3 Model4

β t-value β t-value β t-value β t-value

H1:ATT 0.363*** 7.901 0.328*** 7.120 0.248*** 4.810 0.270*** 5.392

H2:SIN 0.146*** 3.587 0.143*** 3.568 0.086* 2.090 0.070 1.774
H5:PBC 0.370*** 9.949 0.377*** 10.162 0.342*** 8.851 0.321*** 8.715
H3: SDN 0.128*** 3.885 0.119*** 3.399 0.118*** 3.689
H4: PMN 0.215*** 4.973 0.213*** 5.598
H6a: PBC*SIN −0.147*** 3.648
H6b: PBC*SDN −0.079* 2.379
H6c: PBC*PMN 0.015 0.359
R2 0.437 0.452 0.480 0.510
GoF 0.571 0.579 0.603 0.622
f2 0.028 0.053 0.061

Note: *p < 0.05, **p < 0.01, ***p < 0.001.

the subjective injunctive norm is vital to individuals who are relatively

Low SDN
not influenced by PBC. Fig. 3 illustrates that if an individual has a low
6 High SDN
level of PBC, then a strong perception of the subjective descriptive norm
5 would cause a high level of energy-saving intention. Nevertheless,
subjective injunctive norm and subjective descriptive norm have lim-
4 ited correlation to individuals with a high level of PBC.

3
5. Discussion and implication
2
TPB establishes a good foundation for explaining the energy con-
1
servation intention. Perceived behavioral control is the most decisive
0 factor for energy conservation intention, while attitude towards energy-
saving is the second significant determinant. Two additional constructs
(i.e., subjective descriptive norm and personal moral norm) adopted in
Fig. 2. Interaction of the SDN and PBC on energy-saving intention.
the previous research raised the predictive power of TPB. The results
show that the path coefficient of subjective injunctive norm decreases
Low SIN after including the two additional factors. While taking three combined
6 High SIN effects into account, subjective injunctive norm becomes non-sig-
nificant. Therefore, Chinese individuals are more easily affected by the
5
subjective descriptive norm and personal moral norm rather than the
4 subjective injunctive norm. The interaction of PBC and PMN has no
significant effect on individual’s energy-saving intention. This result is
3 in line with the study of Shi et al. (2017b), who focused on Chinese
2 public transportation adoption intention.
The findings indicated that PBC interacts with extrinsic norms. In
1 relation to the subjective descriptive norm and subjective injunctive
0 norm, the interaction terms were negatively related with individual’s
energy-saving intention. Hence, intense social pressure decreases the
effect of PBC on energy-saving intention. Furthermore, in relation to
Fig. 3. Interaction of the SIN and PBC on energy-saving intention.
PBC, the negative interaction term could be explained that PBC plays a
more critical role for individual perceiving less extrinsic norms and
Attitude (β = 0.270, p < 0.001), perceived behavioral control insensitive to social pressure. Therefore, intervention strategies for in-
(β = 0.321, p < 0.001), subjective descriptive norm (β = 0.118, dividuals with high and low PBC should be differentiated. For people
p < 0.001), and personal moral norm (β = 0.213, p < 0.001) were with high PBC, highly propaganda strategies on the subjective de-
significantly and positively related to individual’s intention to save scriptive norm and subjective injunctive norm would be inefficiency.
energy, which supported H1, H3, H4, and H5. However, the subjective Nevertheless, for people with low PBC, intensive propaganda strategies
injunctive norm had no significant effect on energy-saving intention, on the extrinsic norms would be more efficient. This finding is opposite
which failed to provide the support for H2. The combined effects of the to the study of Shi et al. (2017b). The reason for this discrepancy may
extrinsic norm (subjective injunctive norm and subjective descriptive be due to the research context difference. Udalov et al. (2017) argued
norm) and PBC (βSIN = 0.147, p < 0.001; βSDN = −0.079, p < 0.05) that determinants may have different effects on the different types of
were significantly related to individual’s intention to save energy, energy use. Shi et al. (2017b) focused on public transportation adoption
which supported H6a and H6b. However, the combined effect of the intention and electric vehicles adoption intention. Public transportation
intrinsic norm and PBC had no significant effect on intention to save adoption means more PBC related problems concerning the con-
energy, which failed to provide the support for H6c. Figs. 2 and 3 show venience and comfort, while electric vehicles adoption means more PBC
the slope analyses for the two significant interaction terms and the error related problems concerning potential technological risk and high fi-
bars in the figures represent 95% confidence intervals. Fig. 2 shows that nancial cost. Whereas this paper studies energy-saving behaviors in

96
X. Ru et al. Resources, Conservation & Recycling 134 (2018) 91–99

daily life, which are more routine and more accessible, therefore face norm, PBC and subjective descriptive norm are negatively associated
with lower bother about convenience, technological risk, and high fi- with energy-saving intention. Furthermore, subjective descriptive norm
nancial cost. plays a pivotal role in prompting individuals with limited energy-saving
The practical implications for governments and environmental knowledge and skills to conduct energy-saving behavior. Notably, the
protection organizations to promote energy conservation can be de- combined effect of PBC and SDN is higher than that of PBC and SIN.
duced from the gained findings. First, extrinsic norms are confirmed Implications are gained based on the findings that governments and
crucial to motivating individuals’ intention to save energy. It enhances environmental protection organizations ought to enhance energy-
the energy-saving intention of people without considerable PBC. saving behavior as a popular social trend. The promotional activities
Therefore, governments and environmental protection organizations are proposed to offer information concerning injunctive norm (i.e.,
may position energy-saving behavior as a social trend, such as conduct acceptable behaviors by social members) and descriptive norm (i.e.,
propaganda to show the popularity of energy-saving behavior. behaviors conducted by others).
Moreover, Ohanian (1990) indicated that celebrities are significant This paper contributes to the energy-saving behavior literatures by
sources of social influence. Behaviors delivered by prestigious celeb- expounding the nature and the role of extrinsic normative factors (SIN
rities and experts could be convincing to the public. Social norms are and SDN) and intrinsic normative factor (PMN) in predicting energy-
accepted by society members and dominated individuals’ behaviors saving intention. Moreover, the combined effects of normative factors
(Cialdini and Goldstein, 2004). These principles imply that people are and PBC are explored. Another explanation of energy-saving intention
tending to act desirably. Desirable behaviors can be realized efficiently is offered that focus on the direct effect of subjective norm on energy-
through social influence, study, and comparison (Abrahamse and Steg, saving intention. The current study provides promotional measures to
2013). Also, take actions to enhance PBC for individuals who are in- motivate energy conservation behavior. These measures will be espe-
sensitive to social pressure will be competent to motivate energy-saving cially conducive to governments and environmental protection orga-
intention. nizations in the areas facing with severe energy overuse problems.
It is worth noting that the combined effect of PBC and SDN is more Although we conducted a comprehensive study, the limitations of
significant than that of PBC and SIN, which means a stronger perception this paper do exist. First, this paper only relies on self-reports. Thus, it is
of the subjective descriptive norm would lead to a higher level of en- unavoidable that participants may overvalue their energy-saving in-
ergy-saving intention than the subjective injunctive norm. According to tention to satisfy social desirability. Different behaviors are subject to
self-categorization theory, individuals are more likely to categorize social desirability bias to a different extent (Gamberini et al., 2014).
themselves in a particular group and follow the behavior of most group Data collected through a questionnaire combining with an implicit
members to get group identification, or else they will lose the legality measure can help check whether desirability is at stake for a particular
and be isolated (Hogg and Terry, 2000). Thus, if an individual has a low behavior or not (Gamberini et al., 2014). Therefore, it is of great sig-
level of PBC, then tell them what essential others’ demonstrative be- nificance to focus the study on specific energy-saving behavior and
havior is better than tell them important others’ expectations. For ex- adopt implicit association test to check desirability degree. Second, this
ample, a message about the frequency or percentage of the residents’ study was conducted through an online survey in eastern China, and the
performance regarding energy conservation behavior may be more ef- sample’s selectivity of Internet users may over-represent the attitude
fective than widely disseminated public service advertising. Generally and energy-saving behavior of young people and middle-high education
speaking, encouraging energy-saving behavior as a social behavioral groups. Further research would verify the present findings by re-
trend rather than merely emphasis on social expectation may be a plicating the analysis on a large and more representative sample. Third,
reasonably effective method. this paper only studies the general energy conservation intentions. In
Promoting individuals’ energy conservation behavior needs joint fact, individual’s energy-saving behavioral mechanisms may be varied
efforts from all stakeholders (D’Oca et al., 2017). The results obtained in the workplace and household. For instance, cognitive dissonance is
in this research provide some inspiration to the stakeholders to conduct more likely to occur in the workplace than that in the household; people
energy conservation behavior. For the residential community, it is an may follow what colleagues or leaders do even if they don’t approve of
effective manner to motivate households to conduct energy-saving be- it. Thus, comparing energy-saving behavioral mechanisms in different
havior by transmitting demonstrative energy conservation information places may obtain more interesting findings. Finally, four types of en-
among the neighborhoods. In addition, energy conservation demon- ergy-saving behavior studied in this paper have a certain degree of si-
stration from managers or business executives is also influential in milarity, which may cause spillover effect. Therefore, it is essential to
leading energy-saving behavior among employees in the workplace. It examine research framework proposed in this paper in the specific
is important for policymakers to note that the formulation of energy- energy conservation behavior. Cognitive dissonance and behavioral
saving policies or measures should not be one-policy-fits-all. The gen- spillover will complicate the interpretation of the relationship between
eral policy of energy-saving may not be able to achieve the desired cognitive factors and behaviors, for example, obstacles to pro-en-
results. Researchers may get more meaningful conclusions by sub- vironmental routines can affect pro-environmental attitudes. There are
dividing and studying the different energy consumption groups. Per- still many works remaining to be studied in the future. Among them, it
ceived behavioral control is a crucial determinant of individuals’ energy could be interesting to analyze the energy conservation behavior af-
conservation behavior. Thus, building designers, operators, and man- fected by the cultural specificity.
agers should think about how to make it easier for people to reduce
energy waste.
Acknowledgements
6. Conclusion and limitation
This research was financially supported by Zhejiang Provincial
This study expanded TPB with two additional normative variables Social Science Foundation of China (Grant No. 16NDJC002Z), National
(personal moral norm and subjective descriptive norm) and analyzed Social Science Foundation of China (Grant No. 16CGL041) and Zhejiang
the combined effects of perceived behavioral control and normative Provincial Natural Science Foundation (Grant No. LQ17G030010). The
variables on individual’s energy-saving behavioral intention. The in- authors would like to express their gratitude to the usable answers of
teraction term of PBC and PMN has no significant influence on energy- survey respondents and valuable comments of the anonymous re-
saving intention. The interaction terms of PBC and subjective injunctive viewers.

97
X. Ru et al. Resources, Conservation & Recycling 134 (2018) 91–99

Appendix A. : Constructs and measurement items.

Constructs Measurement items

Attitude towards energy-saving (ATT) I think save energy in my daily life is useful to protect the environment
I think save energy in my daily life is important to reduce carbon emissions.
I think save energy in my daily life is valuable to alleviate energy shortage issues.
I think save energy in my daily life is a wise action.
Subjective injunctive norm (SIN) My family members think that I should save energy in my daily life.
My close friends think that I should save energy in my daily life.
Others who are important to me think I should save energy in my daily life.
Perceived behavioral control (PBC) I think that I am capable of saving energy in my daily life.
I have the knowledge and skills to save energy in my daily life.
Whether or not saving energy is completely up to me.
Personal moral norm (PMN) I think I have a moral responsibility to save energy in my daily life.
Save energy in my daily life is depending on my own moral obligation.
I would feel unhappy if I don’t save energy in my daily life.
Subjective descriptive norm (SDN) My friends have taken actions to save energy.
A number of neighbors in my community I know have participated in energy-saving behavior.
Others who are important to me have participated in energy-saving behavior.
Energy-saving intention (ESI) I am willing to save energy in my daily life.
I intend to engage in energy-saving activities in my daily life.
I will make an effort to save energy in my daily life.

References psychology: an interdisciplinary framework for context and occupant behavior in

office buildings. Energy Res. Soc. Sci. 34, 240–251.
D’Oca, S., Hong, T., Langevin, J., 2018. The human dimensions of energy use in buildings:
Abrahamse, W., Steg, L., 2009. How do socio-demographic and psychological factors a review. Renewable Sustainable Energy Rev. 81, 731–742.
relate to households’ direct and indirect energy use and savings? J. Econ. Psychol. 30 De Leeuw, A., Valois, P., Ajzen, I., Schmidt, P., 2015. Using the theory of planned be-
(5), 711–720. havior to identify key beliefs underlying pro-environmental behavior in high-school
Abrahamse, W., Steg, L., 2013. Social influence approaches to encourage resource con- students: implications for educational interventions. J. Environ. Psychol. 42,
servation: a meta-analysis. Global Environ. Change 23 (6), 1773–1785. 128–138.
Abrahamse, W., Steg, L., Vlek, C., Rothengatter, T., 2005. A review of intervention studies Donald, I.J., Cooper, S.R., Conchie, S.M., 2014. An extended theory of planned behaviour
aimed at household energy conservation. J. Environ. Psychol. 25 (3), 273–291. model of the psychological factors affecting commuters' transport mode use. J.
Abrahamse, W., Steg, L., Gifford, R., Vlek, C., 2009. Factors influencing car use for Environ. Psychol. 40, 39–48.
commuting and the intention to reduce it: a question of self-interest or morality? Du, L., Guo, J., Wei, C., 2017. Impact of information feedback on residential electricity
Transp. Res. F: Traffic Psychol. Behav. 12 (4), 317–324. demand in China. Resour. Conserv. Recycl. 125, 324–334.
Ajzen, I., 1991. The theory of planned behavior. Organ. Behav. Hum. Decis. Process. 50 Fishbein, M., Ajzen, I., 2011. Predicting and Changing Behavior: The Reasoned Action
(2), 179–211. Approach. Taylor and Francis.
Allan, J.L., Johnston, M., Campbell, N., 2011. Missed by an inch or a mile? Predicting the Fornell, C., Larcker, D.F., 1981. Evaluating structural equation models with unobservable
size of intention–behaviour gap from measures of executive control. Psychol. Health variables and measurement error. J. Mark. Res. 39–50.
26 (6), 635–650. Gamberini, L., Spagnolli, A., Corradi, N., Sartori, G., Ghirardi, V., Jacucci, G., 2014.
Anderson, J.C., Gerbing, D.W., 1988. Structural equation modeling in practice: a review Combining implicit and explicit techniques to reveal social desirability bias in elec-
and recommended two-step approach. Psychol. Bull. 103 (3), 411. tricity conservation self-reports. Energy Effic. 7 (6), 923–935.
Armstrong, J.S., Overton, T.S., 1977. Estimating nonresponse bias in mail surveys. J. Gao, L., Wang, S., Li, J., Li, H., 2017. Application of the extended theory of planned
Mark. Res. 396–402. behavior to understand individual’s energy-saving behavior in workplaces. Resour.
Bamberg, S., Hunecke, M., Blöbaum, A., 2007. Social context, personal norms and the use Conserv. Recycl. 127, 107–113.
of public transportation: two field studies. J. Environ. Psychol. 27 (3), 190–203. Greaves, M., Zibarras, L.D., Stride, C., 2013. Using the theory of planned behavior to
Blok, V., Wesselink, R., Studynka, O., Kemp, R., 2015. Encouraging sustainability in the explore environmental behavioral intentions in the workplace. J. Environ. Psychol.
workplace: a survey on the pro-environmental behavior of university employees. J. 34 (34), 109–120.
Clean. Prod. 106, 55–67. Hair, Joseph F., et al., 2009. Multivariate Data Analysis: a Global Perspective, seventh ed.
Botetzagias, I., Dima, A.F., Malesios, C., 2015. Extending the theory of planned behavior Prentice Hall, Upper Saddle River.
in the context of recycling: the role of moral norms and of demographic predictors. Hair, J.F., Ringle, C.M., Sarstedt, M., 2011. PLS-SEM: Indeed a silver bullet. J. Mark.
Resour. Conserv. Recycl. 95, 58–67. Theory Pract. 19 (2), 139–152.
Castanier, C., Deroche, T., Woodman, T., 2013. Theory of planned behavior and road Hall, P.A., Fong, G.T., Epp, L.J., Elias, L.J., 2008. Executive function moderates the in-
violations: the moderating influence of perceived behavioral control. Transp. Res. F tention-behavior link for physical activity and dietary behavior. Psychol. Health 23
Traffic Psychol. Behav. 18, 148–158. (3), 309–326.
Chen, C.F., Knight, K., 2014. Energy at work: social psychological factors affecting energy Harman, H.H., 1976. Modern Factor Analysis. University of Chicago Press, Chicago IL.
conservation intentions within chinese electric power companies. Energy Res. Soc. Hogg, M.A., Terry, D.I., 2000. Social identity and self-categorization processes in orga-
Sci. 4, 23–31. nizational contexts. Acad. Manage. Rev. 25 (1), 121–140.
Chen, C.F., Xu, X., Frey, S., 2016. Who wants solar water heaters and alternative fuel Hori, S., Kondo, K., Nogata, D., Ben, H., 2013. The determinants of household energy-
vehicles? Assessing social–psychological predictors of adoption intention and policy saving behavior: survey and comparison in five major Asian cities. Energy Policy 52,
support in China. Energy Res. Soc. Sci. 15, 1–11. 354–362.
Chin, W.W., 1998a. Commentary: issues and opinion on structural equation modeling. Iacobucci, D., 2009. Everything you always wanted to know about SEM (structural
MIS Q. 22 (1), vii–xvi. equations modeling) but were afraid to ask. J. Consum. Psychol. 19, 673–680.
Chin, W.W., 1998b. The partial least squares approach to structural equation modeling. H. Latan, I. Ghozali, 2012. Partial least Squares: Concept and application path modeling
In: Marcoulides, G.A. (Ed.), Modern Methods for Business Research. Taylor and using program XLSTAT-PLS for empirical research. BP UNDIP.
Francis Inc, Mahwah, United States, pp. 295–336. Lizin, S., Van Dael, M., Van Passel, S., 2017. Battery pack recycling: behavior change
Cialdini, R.B., Goldstein, N.J., 2004. Social influence: compliance and conformity. Annu. interventions derived from an integrative theory of planned behavior study. Resour.
Rev. Psychol. 55, 591–621. Conserv. Recycl. 122, 66–82.
Cohen, J., 1988. Statistical Power Analysis for the Behavioral Sciences. L. Erlbaum MEPC (Ministry of Environmental Protection of the People’s Republic of China) China
Associates. State of the Environment Bulletin 2016 (in Chinese) (Accessed 6 June 2017). http://
Corradi, N., Priftis, K., Jacucci, G., Gamberini, L., 2013. Oops, I forgot the light on! The www.zhb.gov.cn/hjzl/zghjzkgb/lnzghjzkgb/201706/P020170605833655914077.
cognitive mechanisms supporting the execution of energy-saving behaviors. J. Econ. pdf.
Psychol. 34, 88–96. Ohanian, R., 1990. Construction and validation of a scale to measure celebrity endorsers’
D’Oca, S., Chen, C.F., Hong, T., Belafi, Z., 2017. Synthesizing building physics with social perceived expertise, trustworthiness, and attractiveness. J. Advert. 39–52.

98
X. Ru et al. Resources, Conservation & Recycling 134 (2018) 91–99

Park, E., Kwon, S.J., 2017. What motivations drive sustainable energy-saving behavior? Wang, S., Fan, J., Zhao, D., Yang, S., Fu, Y., 2016. Predicting consumers’ intention to
An examination in South Korea. Renewable Sustainable Energy Rev. 79, 494–502. adopt hybrid electric vehicles: using an extended version of the theory of planned
Reinartz, W., Haenlein, M., Henseler, J., 2009. An empirical comparison of the efficacy of behavior model. Transportation 43 (1), 123–143.
covariance-based and variance-based SEM. Int. J. Res. Mark. 26 (4), 332–344. Webb, D., Soutar, G.N., Mazzarol, T., Saldaris, P., 2013. Self-determination theory and
Sarkis, A.M., 2017. A comparative study of theoretical behaviour change models pre- consumer behavioral change: evidence from a household energy-saving behavior
dicting empirical evidence for residential energy conservation behaviours. J. Clean. study. J. Environ. Psychol. 35, 59–66.
Prod. 141, 526–537. Wetzels, Martin, Odekerken-Schröder, Gaby, van Oppen, Claudia, 2009. Using pls path
Schwartz, S.H., 1977. Normative influences on altruism 1. Adv. Exp. Soc. Psychol. 10, modeling for assessing hierarchical construct models: guidelines and empirical il-
221–279. lustration. MIS Q. 33 (1), 177–195.
Shi, H., Fan, J., Zhao, D., 2017a. Predicting household PM2.5-reduction behavior in Xu, X., Arpan, L.M., Chen, C.F., 2015. The moderating role of individual differences in
chinese urban areas: an integrative model of theory of planned behavior and norm responses to benefit and temporal framing of messages promoting residential energy-
activation theory. J. Clean. Prod. 145, 64–73. saving. J. Environ. Psychol. 44, 95–108.
Shi, H., Wang, S., Zhao, D., 2017b. Exploring urban resident’s vehicular PM2.5 reduction Yadav, R., Pathak, G.S., 2017. Determinants of consumers’ green purchase behavior in a
behavior intention: an application of the extended theory of planned behavior. J. developing nation: applying and extending the theory of planned behavior. Ecol.
Clean. Prod. 147, 603–613. Econ. 134, 114–122.
Udalov, V., Perret, J., Vasseur, V., 2017. Environmental motivations behind individuals’ Yang, S., Zhang, Y., Zhao, D., 2016. Who exhibits more energy-saving behavior in direct
energy efficiency investments and daily energy-saving behaviour: evidence from and indirect ways in china? The role of psychological factors and socio-demo-
Germany, the Netherlands and Belgium. Int. Econ. Econ. Policy 14 (3), 481–499. graphics. Energy Policy 93, 196–205.
Wan, C., Shen, G.Q., Choi, S., 2017. Experiential and instrumental attitudes: interaction Yazdanpanah, M., Forouzani, M., 2015. Application of the Theory of Planned Behaviour
effect of attitude and subjective norm on recycling intention. J. Environ. Psychol. 50, to predict Iranian students' intention to purchase organic food. J. Clean. Prod. 107,
69–79. 342–352.
Wang, Z., Zhang, B., Li, G., 2014. Determinants of energy-saving behavioral intention Yue, T., Long, R., Chen, H., 2013. Factors influencing energy-saving behavior of urban
among residents in Beijing: extending the theory of planned behavior. J. Renewable households in Jiangsu Province. Energy Policy 62, 665–675.
Sustainable Energy 6 (5), 711–720.

99