Determinants of Employees’ Personal and Collective Energy Consumption and Conservation at Work

Abstract

Energy conservation in public buildings is an important means towards reducing CO₂ emissions worldwide and tackling climate change. In this context, employee behaviour has been recognised as a highly impactful factor that needs to be studied more thoroughly. In this study, we propose and investigate a behavioural model that can be utilised in energy-saving interventions in the workplace. Employing a questionnaire (N = 119 employees in three workplaces in EU countries), we identified two types of energy consumption behaviour at work: personal and collective actions. We further investigated the effect of six factors on employee willingness, as well as self-reported energy-saving habits and behaviour. We found that an employee’s profile (i.e., i. personal energy-saving norms, ii. emotional exhaustion/burnout, iii. collective energy-saving responsibility and efficacy, iv. awareness of energy wastage and knowledge of solution, v. personal comfort/comfort levels, vi. age, vii. gender, and viii. having children) determines energy-saving habits and behaviour, as well as affects willingness to alter it and to conserve energy at work. Employee willingness in turn directly affects energy-saving habits and behaviour at work. The proposed behavioural model can provide guidance towards applying energy conservation initiatives in the workplace. Behavioural interventions should accordingly primarily focus on improving personal energy-saving norms at work and be designed to be easy to follow and not overly demanding, time consuming, or pressuring. Moreover, to motivate collective energy-saving behaviours, interventions should focus on increasing employees’ collective energy-saving responsibility and efficacy, while respecting their personal comfort/comfort levels and their emotional exhaustion/burnout levels. Practical advice towards specific types of interventions is provided accordingly.

1. Introduction

The specific importance of energy conservation in organisational environments and public buildings stems from a number of facts. Climate change is an issue that has been recognised worldwide for decades, through international actions, authorities, and treaties, with the need to intensify our efforts towards reducing CO₂ emissions and protecting the environment [1,2,3]. In addition, according to the International Energy Agency, as the global economy rebounded at record speed from the COVID-19 pandemic, global energy demand in 2021 increased by 5.4%, an all-time record high, and accounted for three-quarters of today’s greenhouse gas (GHG) emissions (the remaining quarter originated from outside the energy sector; mainly from agriculture, forestry, and other land use) [4]. Furthermore, 20% of the total delivered energy worldwide is consumed in buildings [5], while the fastest-growing energy demand (avg. 1.6% per year projected until 2040) has been observed in the commercial sector [5]. Therefore, energy conservation in public buildings and workplaces is an important measure towards addressing the worldwide recognised issue of climate change. Apart from that, in the modern context of the “Anthropocene”, where human activity directly affects the climate, organisational conservation is concurrently a matter of environmental sustainability, ethics, and social justice and a matter entwined with different types of crises (environmental, health, economic, humanitarian, etc.) [6].

Organisational energy conservation has been identified as a non-primary goal that is vulnerable to multiple-goal conflicts [7]. As workplaces worldwide are a major source of carbon emissions, changing employee energy use behaviour can lead to significant carbon savings [8]. Occupant behaviour can lead to an increase or decrease of buildings’ designed energy performance by one-third [9]. However, despite the significant effect of public buildings on energy consumption, only a limited body of research focuses on employee energy consumption behaviour—one of the most important factors that could limit increased energy consumption. Moreover, research on employee energy conservation behaviour, as well as the socio-psychological influences of organisational contexts on individual energy decisions, is limited, constraining appropriate and effective policy and planning [10].

To attain the net-zero-emissions scenario (NZE), CO₂ emissions from electricity generation need to halve between 2021 and 2030, decrease by a 10% yearly rate between 2031 and 2040, and have achieved a total of 90% decrease by 2050 through the widespread deployment of efficiency and energy savings measures, including behavioural change [4]. Although identified as a major determinant of energy use in buildings with a comparable impact to technological solutions, human behaviour has not been adequately analysed in energy studies, and the question of how organisational members can be motivated to conserve energy in the workplace remains under-examined [11,12,13,14]. Therefore, there is a need for further research on the relationships between individual behavioural and social factors and energy use in the workplace [15].

Many of the behavioural changes in the NZE scenario target wasteful or excessive energy consumption, predominantly in wealthier parts of the world, in order to reduce emissions and ameliorate global inequalities in per capita energy consumption [4]. Aiming to motivate and support energy behavioural change, we aspired to fill an existing identified gap in the literature with regards to the specific personal characteristics that affect employee energy conservation behaviour at work. Thus, we reviewed existing relevant literature and existing results from semi-structured interviews in three workplaces (situated in different EU countries) to identify such personal characteristics. Accordingly, in the present research, we considered seven important factors that affect employees’ willingness to conserve energy, as well as their personal and collective energy consumption behaviour in the workplace: personal energy-saving norms, energy conservation habits, collective energy-saving responsibility and efficacy, awareness of energy wastage and knowledge of solution, personal comfort/comfort levels, emotional exhaustion/burnout, and demographic characteristics. We conducted a questionnaire survey to explore the relationships between them and assess their impact based on a behavioural model concerning personal and collective energy-saving behaviour at work. Our findings can be important to both researchers and practitioners that aspire to positively affect energy conservation behaviour at work. The insights we provide on the factors that relate to energy usage at work can support the design and execution of appropriate energy saving initiatives and interventions (such as IoT-enabled mobile applications that encourage employees to save energy and offer them the capacity to monitor their energy use in real-time) that fit the employees’ behavioural profile and needs, and, thus, increase the possibility of broad adoption and long-term success.

In the next sections, we begin by reviewing existing relative insight from the literature and analysing the previously mentioned behavioural parameters in our study. We consequently present our research methodology and analyse and discuss our results. Afterwards, we summarise our findings and discuss the theoretical and practical implications of our study, especially focusing on how such findings can contribute to the design and development of effective personalised behavioural interventions that motivate energy saving behaviour.

2. Background: Review of Literature and Formulation of Hypotheses

No matter how little agency that may in some cases be allocated by extant automated systems in modern buildings, employees are by no means passive with regards to energy consumption [16]. Hence, changing employee energy use behaviour can lead to significant savings [8], as they generally use much more energy than they need and are reluctant to change their energy use behaviour at work [17]. Energy conservation through behavioural change should be considered alongside efforts to reduce energy consumption through structural and operational changes or technological improvements [18,19], as it determines one-third of a building’s designed energy performance [9]. Space heating and cooling systems, lights, refrigerators, computers, and other equipment are identified as the largest energy consumption sources within public buildings both in the EU and the US [9]. However, energy-saving behaviours such as turning off equipment when not needed are not necessarily motivated by employees’ pro-environmental intentions [20], and motivations beyond energy reduction need to be harnessed in order to engage employees towards energy-saving at work [17], while the complexity of energy-related behaviours in the workplace is further increased due to the existence of organisational roles and work objectives [21] in addition to workplace attachment [22].

In office environments, energy consumption behaviour is affected by a combination of the physical context (the presence of controls over building systems or equipment), the social context (the peers’ needs, expectations, and norms), and the organisational context (the policies and expectations of the employing organisation) [20]. Moreover, employees generally lack direct financial incentives to conserve energy at work, and different motivations, as well as incentive structures, can be considered. Saving energy at work is therefore usually considered altruistic, with no personal benefits expected in return, and enacted by environmental concern along with the desire to help one’s peers and their organisation [23,24]. Hence, in order for employees to reduce their energy-consumption, we need to develop tailored interventions that create opportunities to leverage their inner commitment and autonomous motivation to conserve energy, while taking into account the demands posed by the organisational setting [25], and to create social (due to the communal nature of most workplaces) and physical opportunities for employees to conserve energy [8]. Moreover, we need to also bear in mind that organisational interventions to increase pro-environmental behaviours (PEBs) are more effective when they concurrently focus on a variety of behaviours, while targeting their antecedents may affect different behaviours at once [26]. At the same time, interventions that provide support and greater control over energy saving or utilise social influence (due to the communal nature of most workplaces that feature both personal and collective energy-consuming devices [27]) are the most promising towards motivating employees to conserve energy [8]. Participatory interventions are accordingly increasingly recognised as an effective means of enhancing organisational energy-saving behaviour [25].

According to existing insight from the literature, in organisational environments, personal and collective energy-saving at work is related to an employee’s profile in terms of their:

• Personal Energy-Saving Norms: Psychological factors found to be important for energy saving at work include an employee’s personal norms and attitudes, as well as sense of personal responsibility [8,17]. Values-Beliefs-Norms (VBN) theory suggests that pro-environmental behaviour is directly affected by personal norms for pro-environmental action [28]. Accordingly, it has been found that energy-saving personal norms can be leveraged in organisational interventions to affect employees’ energy-saving intentions and actions [19,26,29].
• Willingness to Conserve Energy: Willingness is defined as “the quality or state of being prepared to do something; readiness” [30]. Personal norms are considered a significant predictor with a direct effect and strong association on willingness to sacrifice in order to protect the environment [31]. Moreover, as concern for the environment does not automatically translate to pro-environmental actions at work, in the “value/action gap” phenomenon [7] it is important to also evaluate the willingness of employees to conserve energy. Employees holding a strong belief on the importance of energy saving are more willing to save energy at work, even at some cost on their personal comfort [15]. At the same time, although the use of energy-efficient technologies (EETs), such as automated lights and programmable thermostats, has contributed to long-term energy reductions worldwide, one of the barriers towards their successfulness is a lack of willingness to use them [32]. However, willingness to enact a behaviour is a more salient predictor of its enactment only when established habits are weak [33], as changing conscious intentions can prove too weak in the face of strong habits [34] and strong habits tend to override conscious intentions [35]. Therefore, investigating willingness and habits in the same context may provide for a more complete assessment of the participants’ behaviours.
• Energy Conservation Habits: Habit is defined as “a settled or regular tendency or practice, especially one that is hard to give up” [30]. The stressful modern work environment and automatic nature of energy-related behaviours promotes their habitual enactment and impedes changing them mainly due to the existence of a “status quo bias” [36]. When a person has a considerable commitment to, or psychological investment in, the status quo, then changing one’s habit is a burdensome act and they may exhibit a “status quo bias”, thus electing to do nothing to change their current situation or behaviour when called upon to do so, out of convenience, habit or inertia, policy (company or government), or custom, because of fear or innate conservatism or through simple rationalisation [37]. A plethora of reasons have been suggested in the literature for not adopting energy-saving habits, such as the habit of switching off desk equipment in workplaces, the inconvenience of turning off a PC and the time taken to reboot, or the perception that fellow employees do not turn their PCs off [17]. As habits strongly determine the enactment of behaviours, interventions aimed at changing energy conservation behaviour will have to address both habitual and intentional energy consumption.
• Emotional Exhaustion: Existing stressful conditions at work or high workload may create a context in which individual considerations for energy conservation are assigned a lower priority by employees. Burnout is a prolonged response to such chronic emotional and interpersonal stressors at work that leads to a decrease in productivity, effectiveness, job satisfaction, and organisational commitment [38], thus triggering a downward spiral in individual and organisational performance [39]. Employees who experience burnout exhibit the lowest in-role and extra-role performance within their organisation, thus becoming the organisation’s worst performers [40]. Therefore, as energy-saving at work is considered as an extra-role behaviour, burnout may also lead to a decrease in energy conservation actions. Exhaustion is considered the central, most widely reported, and thoroughly analysed dimension of burnout that reflects emotional stress and emerges as a result of work pressure and overload, as well as social conflict or lack of social support [41,42]. Furthermore, although it has been identified as a separate burnout dimension across a wide variety of occupations [43], high scores on exhaustion indicate the presence of burnout, as exhaustion represents the “energy” dimension of burnout [40,42]. Moreover, interventions to prevent or alleviate exhaustion can focus on reducing job demands mainly by job redesign and training programs [41].
• Collective Responsibility and Efficacy: Perceived collective efficacy is concerned with the performance capability of a group as a whole and affects the employee’s sense of mission and purpose, strength of common commitment, and how well they work together with their peers towards attaining a goal [44]. It is considered a stronger predictor of people’s self-reported pro-environmental behaviour in collective settings than self-efficacy [45]. Enhancing personal and collective energy-saving efficacy can in turn increase personal and collective energy-saving intentions and actions [46]. Energy saving at work can be considered a collective moral responsibility towards preserving the environment for future generations [47]. Typically, individuals do not act in isolation in the workplace but within groups of colleagues, whose collective efficacy influences their own [21]. This influence is further exerted in the presence of many shared workspaces, tasks, appliances, and the fact that energy-related behaviours in the workplace are regularly observed by peers [21]. In an organisational setting, employee energy conservation entails actions that could either be solely personally performed (e.g., operating a personal computer) or as a member of a group of employees (e.g., operating a space cooling device in a shared office space). In these latter cases, the responsibility to conserve energy lies in groups and is therefore shared, while the “free rider” issue arises as a result of this shared responsibility [31]. When we allocate responsibility for group outcomes, we can either ascribe individual responsibility to each member of that group or ascribe joint responsibility to individual members for their contribution to the collectively brought about outcome [48]. In this context, collective responsibilities, such as conserving energy in the workplace, cannot effectively be attended to by individual employees, especially in the case of shared electrical devices. Hence, this responsibility must be assumed both by individual employees and the groups they belong to within their work environment. In fact, collective and individual responsibility are not contradictory but can work in concert to achieve the commonly ascribed goal [49].
• Awareness of Energy Wastage and Knowledge of Solution: Knowledge on how to conserve energy in addition to awareness of the multi-dimensional problem of energy sustainability and the intricacies of energy conservation are considered as enablers of pro-environmental and more sustainable energy-consumption behaviour and furthermore as an important prerequisite for the development of pro-environmental and energy-saving norms and intention, as well as willingness, to act pro-environmentally [50,51,52].
• Personal Comfort/Comfort Levels: When it comes to energy consumption in office buildings, employees attempt to restore their comfort in the easiest possible way [53]. Research in organisational behaviour suggests that the physical setting within offices can influence employees’ behaviours in numerous ways [54,55]. Creating a comfortable working environment has a large effect on productivity (5–15%) [56]. Combined with air quality, thermal comfort (indoor temperature) has a considerable effect (6–9%) on employee performance [56,57,58]. Hence, in order to promote human comfort, health, and productivity, adequate ventilation IS recommended [59]. However, as ventilation consumes energy (the air within a building is heated or cooled), the ventilation rates in buildings must strike a balance between energy consumption and personal comfort [60]. Overall, occupants’ preferences for thermal comfort, and ventilation practices, have a very significant influence on the energy consumed on heating/cooling [61]. In essence, the perceived loss of comfort may reduce the likelihood of engaging in energy-conservation [62], due to loss aversion as it is difficult to accept that any personal sacrifices would be worth the social gain [63]. However, when provided with adequate autonomy and opportunities to behave pro-environmentally, employees are willing to incur some personal costs and act upon their personal norms towards enacting pro-environmental behaviours (PEBs) [26].
• Demographic Factors Affecting Energy Conservation: It is important to note that demographic factors should also be considered when designing behavioural interventions, as they have been correlated with energy behaviour. Engagement towards pro-environmental behaviour tends to increase with age, while women tend to have stronger environmental attitudes, concern, and behaviours than men across age [17,64]. Additionally, higher levels of motivation to conserve energy have been reported by residential users with children [65], suggesting that this may also be true for employees. Furthermore, across occupations, females have reported higher levels of burnout than males, particularly those who were relatively young or had limited working experience [43].

Based on all of the above, we formulate the following hypotheses:

An employee’sprofile, in terms of (i) personal energy-saving norms, (ii) emotional exhaustion (burnout), (iii) collective energy-saving responsibility and efficacy, (iv) awareness of energy wastage and knowledge of solution, (v) personal comfort/comfort levels, and (vi) demographic factors (age, gender, children), will affect:

H1: 

their willingness to conserve energy at work.

H2: 

their energy-saving habits and behaviour at work.

H3: 

Employees’ willingness to conserve energy will affect their energy-saving habits and behaviour at work.

3. Materials and Methods

In line with the above, we followed a structured preliminary investigation to gain a basic understanding of employee energy consumption in workplaces, as well as their personal characteristics that may relate to it. We conducted on-site visits to prospective pilot sites and semi-structured interviews with employees in order to delve more deeply into their energy saving motivation and behaviour at work. Several common, as well as contradictory significant characteristics were identified, regarding the possible targeted energy-saving behaviours: i.e., PCs, lights, monitors, and printers can be switched off more often when not needed, air conditioning machines can be used more efficiently, and common area equipment such as coffee machines can be used optimally for energy saving. Furthermore, the varying roles of the participants across all sites seemed to affect their daily routines and opportunities to act upon specific energy conservation actions. Individual office occupants’ greater level of control over the use of computers and lighting than over other, more energy-consuming, building systems such as heating or cooling was evident. Working in shared spaces presented additional challenges, as the individual energy-saving actions of the participants also affect their colleagues’, and a level of cooperation towards common goals may be needed. Therefore, we found that in these situations, a need for ascription of collective responsibility to adhere to collective energy-saving actions arises, so that they can be adhered to by the employees more effectively. At the same time, we found that employees believe that both personal and collective actions have an impact on energy conservation at work, and deduced that, within an organisational context, both personal and collective behaviours are significant and need further studying. In addition, not all the employees consider saving energy at work to be their own personal responsibility, while the majority prefer cooperative over competitive energy-saving goals. More details on the process we followed, as well as our findings in the structured preliminary investigation that preceded the current research, can be found in [27].

3.1. Research Model

Based on the preliminary empirical insight outlined and the literature-driven hypotheses we have formulated above, we devised the following research model in our study, which can be reviewed in Figure 1.

To test our research model, we performed an online survey with employees from three different workplaces (two office buildings in Greece and Spain and a museum in Luxemburg).

3.2. Survey Instrument

Based on the insight we gained from the literature, we compiled a questionnaire to assess our participants’ profile and energy-saving behaviour. The questionnaire consisted of 24 questions divided into 8 sections. All items on the questionnaire, except for the demographics section and the questions on emotional exhaustion, were rated on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). The items on emotional exhaustion were scored on a 7-point scale, ranging from (0) “never” to (6) “every day”, as per the MBI rating scale. Higher scores indicate higher agreement in the statements and hence also higher levels of the investigated variables. The participants’ adherence to different energy-wasting behaviours at work was recorded with items adapted from an existing questionnaire [19]. Three different types of willingness in the context of energy saving at work were assessed with corresponding single-item scales: willingness to help the organisation conserve energy and willingness to alter personal energy consumption behaviour with items were adapted from the questionnaire in [19] and willingness to alter energy behaviour collectively was rated with one item we developed accordingly. The participants’ personal energy-saving norms at work were assessed by rephrasing items from an existing questionnaire [19], substituting the word “university” with “workplace”, to better reflect the context of our study. The participants’ awareness of energy wastage at work and knowledge of its solution was assessed with a two-item scale. Collective energy-saving responsibility and efficacy was assessed with a single-item scale. Employee self-perception of the level of personal comfort at work was assessed with a two-item construct. Emotional exhaustion was assessed utilising two items adapted from the exhaustion subscale of the Maslach Burnout Inventory (MBI-GS) [66] that cover the specific aspects of “quantitative demands” and “relations at work”, as we considered these two parameters the most potentially impactful on employee energy saving behaviour at work, that are also representative of the experience of work-related stress. Finally, the participants’ demographic profile was assessed by employing three questions regarding the participant’s age, gender, and whether they had children or not. The complete questionnaire and its descriptive statistics can be found in Section 4.2 of the paper.

3.3. Participants and Procedure

We contacted the employees of: (i) a municipal IT-support office situated in Greece, (ii) an electricity regulation authority in Spain, and (iii) an art museum in Luxembourg, by email and invited them to participate in the survey (a total of all 144 employees were invited to participate). The three sites were also equipped with the necessary IoT infrastructure to facilitate IoT-enabled feedback provision in future experiments in the context of an EU-funded project. The questionnaire was answered through an online platform by 119 participants in total (82.6% response rate), with the majority of ages between 18 and 45 years old (66.1%), while males outnumbered females (55.7% vs. 44.3%), and the majority had children (55.7%).

We performed a multiple-step analysis on the collected answers: (i) Reliability analysis: an analysis on all the sub-sections of the questionnaire, except the demographics section, to determine the reliability of all the multi-item constructs. (ii) Descriptive statistics: to gain insight into the general trend of our samples’ characteristics on all the factors surveyed through the questionnaire. (iii) Analysis of correlations: to explore the relationships between our variables. (iv) Multiple linear regression: to gain more in-depth insight into the mathematical relationship between the measured variables, as well as their directionality, and test the validity of our research model and the assumed hypotheses. All statistical analyses were performed using IBM SPSS Statistics v.23.

The Pearson product-moment correlation coefficient (r) is presented for all correlations reported as part of the correlation analysis, along with the level of statistical significance, indicating the confidence levels of correlations calculated. To assess the internal consistency of the multi-item scales, we calculated and reported Cronbach’s alpha coefficient, which should ideally be above 0.70. However, the reliability of a scale may vary depending on the sample, as well as the number of questions in a scale. Therefore, although the Cronbach alpha score of the personal energy-saving behaviour scale was lower than the 0.70 threshold (a = 0.660), we deemed it a reliable scale, bearing in mind that it featured acceptable inter-item correlation and that according to existing literature it is frequently difficult to achieve acceptable Cronbach’s alpha values on scales with a small number of items [67,68]. Linear regression analyses were performed using the enter method, having ascertained the adequacy of our sample size for the analyses performed, according to existing prescriptions in the literature such as [69,70,71,72]. The squared value of the multiple correlation coefficient (R square) provides an indication of the variance in the dependent variable that is explained by a model. In our analysis, we utilised primarily the adjusted R square statistic, as it tends to provide a better estimate of the true population value, especially in limited samples [67,73]. We utilised the significance level of the F-ratio to test whether the overall regression models were a good fit for the data. The b-coefficients (B) can be used together with the models’ intercept values in the regression tables, to derive regression model equations, by utilising the formula “Predicted Variable” = Intercept + Σ (B_i x “Predictor Variable”_i), where i represents the different predictors in the model [67,73]. As a general guideline, b-coefficients (B) and F-ratios are considered statistically significant if their p value is lower than 0.05 (p < 0.05) [67]. Standardised regression coefficients—beta (β) coefficients—(and their p-values) were also utilised to compare the relative strength of the predictors in each model, in predicting the value of the predicted variable [67,74].

Interpreting a regression coefficient that is statistically significant does not change based on the R² value [73,74]:

• Studies that attempt to explain human behaviour tend to usually produce R² values less than 50%. This is due to the inherent increased difficulty in predicting human behaviour compared to other physical processes, which in turn leads to a greater amount of unexplainable variation in behavioural models.
• Although low R² values can warn of imprecise predictions, in such cases the regression models can be quite useful in identifying the nature, directionality, and strength of the relationships between variables.
• Even in models with a relatively low R², statistically significant p-values continue to identify relationships, and coefficients can be interpreted in the same way. Therefore, although they are not as suitable in generating precise predictions, regression models with a low R-squared value and statistically significant independent variables can help in drawing important conclusions about the existing relationships between the variables in a model.

4. Results

4.1. Personal vs. Collective Energy-Saving at Work

The reliability reported by Scherbaum et al. for the scale on which we based our own self-reported behaviours scale was sufficiently high, with an internal consistency (a = 0.71), while one factor was revealed for all the items [19]. As per the questionnaire section on employee self-reported energy-saving behaviour, in our study, we found that—if examined as one scale—the five energy saving behaviours we surveyed featured lower internal consistency (a = 0.636), with a number of low (even below 0.150) inter-item correlations in a number of cases. We hence performed an exploratory factor analysis to explore if the items in the scale indeed reflected a unique construct. More specifically, a principal components analysis (PCA) was followed after the suitability of our data for factor analysis was assessed. Indeed, an inspection of the correlation matrix revealed the presence of many coefficients of 0.3 and above. The Kaiser-Meyer-Olkin value was 0.608, exceeding the recommended value of 0.6 and Bartlett’s test of sphericity reached statistical significance (p < 0.001), supporting the factorability of the correlation matrix. The results from the PCA revealed the presence of two components with eigenvalues exceeding one (>1), explaining 43.54% and 30.05% of the variance, respectively, and hence the two-component solution explained a total of 73.59% of the variance. An inspection of the scree plot revealed a clear break after the second component, as is evident in Figure 2.

Therefore, in adopting Catell’s (1966) scree test [75], we concluded that the results of this analysis supported the existence of two separate sub-scales, recording the different dimensions of energy saving at work (personal vs. collective). To aid in the interpretation of these two components, Oblimin rotation was performed. The rotated solution also revealed and confirmed the presence of a simple structure, with both components featuring a number of strong loadings and all variables loading substantially on only one component [67]. The pattern and structure matrix obtained in the PCA analysis can be reviewed in

Table 1. Pattern Structure Matrix for PCA with Oblimin Rotation of Self-Reported Energy-Saving Behaviour items.

Item	Pattern Coefficients *		Structure Coefficients *		Communalities
	Component 1	Component 2	Component 1	Component 2
When I am finished using my computer for the day, I turn it off.	−0.030	0.851	−0.064	0.853	0.728
When I leave a room that is unoccupied, I turn off the lights.	0.029	0.877	−0.007	0.876	0.768
When I leave my work area, I turn off the air conditioner(s).	0.857	−0.021	0.858	−0.056	0.737
When I leave my work area, I turn off the printer(s).	0.852	−0.083	0.855	−0.118	0.739
When I am the last to take coffee in the afternoon at work, I turn the coffee machine off.	0.839	0.100	0.835	0.066	0.708

* Note: Major loadings for each item are marked in bold font.

.

This resulting factor structure was also in line with our observations during the visits we conducted at the pilot sites. Indeed, deactivating PCs and switching off the lights in empty workspaces were actions that could be performed individually and were therefore personally ascribed. In contrast, switching off the A/Cs, printers, and coffee machines required coordination with co-workers, as they were collaboratively used in a seamless way, and it was impossible to know if, for example, someone else was going to use the printer, drink coffee, or return to a workspace (that would require air conditioning) later on in the day.

In exploring the descriptive statistics of the items included in the newly formed sub-scales, we noted that the behaviours surveyed through the first sub-scale {(i) When I am finished using my computer for the day, I turn it off, and (ii) When I leave a room that is unoccupied, I turn off the lights} exhibited high mean values. Therefore, they were both behaviours that are in general adhered to by the majority of our sample to a high degree. If we take into account that these behaviours (turning off the PC, as well as the lights) are also behaviours that can be performed at home, there might also be room to assume a level of spillover effect at play in this specific case. Based on all of the above, this sub-scale was regarded as representative of “personal energy-saving behaviour” at work. The reliability of the scale (a = 0.660) was deemed acceptable—although it was below the 0.700 threshold regularly reported in the literature, also taking into account that it was a two-item scale with acceptable levels of inter-item correlation (0.499) (as already delineated in the materials and methods section).

With regards to the items in the second sub-scale {(i) When I leave my work area, I turn off the air conditioner(s), (ii) When I leave my work area, I turn off the printer(s), (iii) When I am the last to take coffee in the afternoon at work, I turn the coffee machine off}, their mean values were significantly lower than the ones reported for the “personal energy-saving behaviour” scale. Thus, these behaviours seem to be performed a lot less consistently by the users, making them less popular in our sample. As this was already analysed, this may be a product of the fact that a certain level of cooperation is required for their adherence, making them more complicated to perform. According to our observations at the pilot sites, this second sub-scale at the same time records collective energy-saving behaviours at work and was hence named “collective energy-saving behaviours”. The reliability of this scale was high (a = 0.807).

We proceeded to analyse the relationships between the variables we measured, based on the premise that the two categories of behaviour (personal and collective energy-saving) are distinct between them. The results in connection to our behavioural model are further explained in the following sub-sections.

4.2. Descriptive Statistics

The complete questionnaire we employed, along with descriptive statistics and scale reliabilities observed in the context of this study, can be found in

Table 2. Questionnaire items, descriptive statistics, means, standard deviations, scale reliabilities, and sample characteristics (N = 119). Items in sections A, Bi, Bii, and C were adapted from [19]. Items in section G were adapted from [66].

Questionnaire Section		Mean	S.D.	Reliability
A. Self-Reported Behaviours *
i. Personal Energy-Saving Behaviour				a = 0.660
When I am finished using my computer for the day, I turn it off.		6.56	1.02
When I leave a room that is unoccupied, I turn off the lights.		6.56	0.86
ii. Collective Energy-Saving Behaviour				a = 0.807
When I leave my work area, I turn off the air conditioner(s).		4.65	2.31
When I leave my work area, I turn off the printer(s).		3.86	2.22
When I am the last to take coffee in the afternoon at work, I turn the coffee machine off.		5.06	2.01
B. Willingness *
i. Willingness to help the organisation conserve energy
I would help the organisation I work for to conserve energy.		6.40	0.80
ii. Willingness to alter personal energy consumption behaviour
I would change my daily routine to conserve energy.		5.91	1.39
iii. Willingness to alter energy behaviour collectively
I would change my energy-consumption behaviour at work, if others do so.		4.09	1.97
C. Energy-Saving Personal Norms *				a = 0.839
Energy conservation is something to be concerned about.		6.36	1.09
Conserving energy and natural resources is important to me.		6.27	1.08
Conserving energy is not my problem. (Reversed)		6.10	1.50
I have a responsibility to conserve energy and resources.		5.99	1.47
The organisation I work for should conserve energy.		6.25	1.11
I should help the organisation I work for conserve energy.		6.12	1.14
D. Awareness of Energy Wastage and Knowledge of Solution *				a = 0.720
I am aware of energy costs and possible sources of energy wastage.		5.88	1.20
I am aware of ways to save energy at work.		5.82	1.22
E. Collective Energy-Saving Responsibility and Efficacy *
Saving energy is a collective effort. Doing it individually has no impact at all.		4.10	2.20
F. Personal Comfort/Comfort Levels *				a = 0.763
The quality of air at work is satisfactory.		3.53	1.81
Climate conditions at work are comfortable.		3.92	1.73
G. Emotional Exhaustion/Burnout **				a = 0.726
Working with people all day long requires a great deal of effort. I feel I work too hard at my job.		3.08	2.15
		2.49	2.08
H. Demographics	Group	Count		%
Age	18–35	23		20.0%
	35–45	53		46.1%
	45–65	39		33.9%
Gender	Male	63		54.8%
	Female	52		45.2%
Do you have children?	Yes	64		55.7%
	No	51		44.3%

* Items were rated on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Higher scores indicate higher agreement in the statements and hence also higher levels of the investigated variables. ** Items were rated on a 7-point Likert scale ranging from (0) “never” to (6) “every day”. Higher scores indicate higher frequency of adherence to the investigated variables.

.

Notably, the levels of adherence to personal energy-saving behaviours at work were on average higher than for collective energy-saving behaviours. Furthermore, the mean levels of our samples’ answers on the personal norms questions were very high (>6.0/7.0), indicating that the participants were on average positively positioned towards energy saving at work. At the same time, although their willingness to help the organisation they work for to conserve energy was on average very highly rated (6.4/7.0), their willingness to alter their personal energy consumption behaviour at work was weaker (5.9/7.0) and their willingness to alter their energy behaviour as a collective even more so (4.1/7.0). Furthermore, the participants on average described themselves as aware of energy wastage at work, as well as the means to solve it (>5.8/7.0), although the levels of collective energy-saving responsibility and efficacy recorded were significantly lower (4.1/7.0). Moderate scores on personal comfort at work were also recorded regarding both the quality of air and climate conditions (3.5 and 3.9/7.0). Finally, the participants on average seemed to experience moderate mean levels of emotional exhaustion at the surveyed workplaces (<3.0/6.0).

4.3. Multiple Linear Regression Results

A series of five separate multiple regression analyses were performed to validate our research hypotheses that an employee’s profile (as defined by the parameters we identified in our preliminary research and through existing insight from the literature) will affect their (H1) willingness to conserve energy at work and (H2) energy-saving habits and behaviour at work. The obtained results can be reviewed in

Table 3. Summary of Regression Analyses—Employees’ Profiles vs. Willingness to Save Energy at Work.

	Willingness to Help theOrganisationConserve Energy			Willingness to Alter Personal Energy Consumption Behaviour			Willingness to Alter Energy Consumption Behaviour Collectively
	B	SE B	β	B	SE B	β	B	SE B	β
Personal Energy-Saving Norms	0.416	0.089	0.483 **	0.632	0.147	0.459 **	0.089	0.244	0.043
Emotional Exhaustion (Burnout)	−0.068	0.040	−0.155	0.051	0.065	0.072	−0.014	0.108	−0.013
Collective Energy-Saving ResponsibilityandEfficacy	0.012	0.034	0.031	0.048	0.056	0.080	0.203	0.093	0.225 *
Awareness of Energy Wastage and Knowledge of Solution	−0.158	0.075	−0.210 *	0.101	0.123	0.085	−0.360	0.204	−0.200
Personal Comfort/Comfort Levels	0.054	0.044	0.106	−0.001	0.073	−0.001	0.061	0.121	0.050
Age	−0.006	0.009	−0.069	−0.017	0.014	−0.114	0.011	0.023	0.049
Gender	0.434	0.147	0.266 **	0.064	0.242	0.024	−0.282	0.402	−0.072
Children	0.126	0.153	0.077	−0.034	0.253	−0.013	−0.316	0.420	−0.080
R2		0.316			0.273			0.107
Adj. R2		0.261			0.214			0.035
Intercept		4.726			1.868			4.364
F		5.78 **			4.69 **			1.50

N = 119, * p < 0.05, ** p < 0.01.

and

Table 4. Summary of Regression Analyses Employees’ Profiles vs. Energy Habits and Behaviour at Work.

	Personal Energy-Saving Behaviours			Collective Energy-Saving Behaviours
	B	SE B	β	B	SE B	β
Personal Energy-Saving Norms	0.250	0.102	0.283 **	−0.296	0.222	−0.151
Emotional Exhaustion (Burnout)	−0.118	0.045	−0.259 **	0.137	0.099	0.136
Collective Energy-Saving ResponsibilityandEfficacy	−0.019	0.039	−0.049	0.178	0.085	0.208 *
Awareness of Energy Wastage and Knowledge of Solution	−0.020	0.085	−0.025	0.003	0.186	0.002
Personal Comfort/Comfort Levels	−0.043	0.050	−0.083	0.290	0.110	0.250 *
Age	0.002	0.010	0.019	0.016	0.021	0.076
Gender	−0.109	0.167	−0.065	0.165	0.366	0.045
Children	0.112	0.175	0.066	0.515	0.382	0.138
R2		0.160			0.174
Adj. R2		0.093			0.108
Intercept		5.620			1.868
F		2.38 *			2.63 *

N = 119, * p < 0.05, ** p < 0.01.

.

A series of two additional multiple regression analyses were performed to validate our research hypothesis (H3) that an employee’s willingness to conserve energy will affect their personal and collective energy-saving habits and behaviour at work. The obtained results can be reviewed in

Table 5. Summary of Regression Analyses—Willingness vs. Energy-Saving Habits and Behaviour at Work.

	Personal Energy-Saving Behaviours			Collective Energy-Saving Behaviours
	B	SE B	β	B	SE B	β
Willingness to Help theOrganisationConserve Energy	0.347	0.100	0.338 **	−0.169	0.241	−0.072
Willingnessto Alter Personal Energy Consumption Behaviour	−0.032	0.057	−0.054	0.150	0.138	0.112
Willingnessto Alter Energy Consumption Behaviour Collectively	0.054	0.036	0.130	−0.001	0.088	−0.001
R2		0.115			0.011
Adj. R2		0.092			−0.015
Intercept		4.303			4.724
F		4.98 **			0.412

N = 119, ** p < 0.01.

.

We found that an employee’s profile explained (i) 26.1% of the variance in their willingness to help the organisation conserve energy (personal energy-saving norms, awareness of energy wastage and knowledge of solution, and gender statistically significantly), (ii) 21.4% of the variance in their willingness to alter personal energy consumption behaviour (personal energy-saving norms statistically significantly), and (iii) did not significantly explain the variance in their willingness to alter energy-saving behaviour collectively at work. Moreover, it also explained (iv) 9.3% of the variance in their self-reported personal energy-saving behaviours (personal energy-saving norms and emotional exhaustion/burnout statistically significantly), and (v) 10.8% of the variance in their self-reported collective energy-saving behaviours (collective energy-saving responsibility and efficacy and personal comfort/comfort levels statistically significantly). Finally, an employee’s willingness explained (vi) 9.2% of the variance in their personal energy-saving behaviours (willingness to help the organisation conserve energy statistically significantly), while (vii) an employee’s profile did not significantly explain the variance in their collective energy-saving behaviours. The findings from the multiple linear regression analyses in the context of our research model can be reviewed in Figure 3.

The results we have presented in the regression tables can also be utilised to construct corresponding modelling equations. More specifically, the b-coefficients (B) can be used together with the models’ intercept values in the regression tables to derive regression model equations, by utilising the formula “Predicted Variable” = Intercept + Σ (B_i × “Predictor Variable”_i), where i represents the different predictors in the model [67,73]. Therefore, to provide an example utilising the data recorded in the first regression analysis presented in Table 5, we can deduce that, according to our findings, an employee’s personal energy-saving behaviours can be modelled by the following equation: “Personal Energy-Saving Behaviours” = 4.303 + (0.347 × “Willingness to Help the Organisation Conserve Energy”)−(0.032 × “Willingness to Alter Personal Energy Consumption Behaviour”) + (0.054 × “Willingness to Alter Energy Consumption Behaviour Collectively”). Similar modelling equations can be constructed for all the remaining relationships presented in our model accordingly by utilising the figures presented in Table 3 and Table 4.

4.4. Observed Correlations

The inter-correlations between the constructs we employed in our study are presented in

Table 6. Intercorrelations between constructs.

	1. Personal Energy-Saving Norms	2. Emotional Exhaustion (Burnout)	3. Collective Energy-Saving Responsibility and Efficacy	4. Awareness of Energy Wastage and Knowledge of Solution	5. Personal Comfort/ Comfort Levels	6. Age	7. Gender	8. Children	9. Willingness to Help the Organisation Conserve Energy	10. Willingness to alter personal energy consumption behaviour	11. Willingness to alter energy behaviour collectively	12. Personal Energy-Saving Behaviours	13. Collective Energy-Saving Behaviours
2.	−0.081	1
3.	0.011	0.313 **	1
4.	0.536 **	−0.102	−0.037	1
5.	−0.057	−0.180	0.072	−0.041	1
6.	0.138	−0.137	−0.215 *	0.045	−0.001	1
7.	0.206 *	−0.034	−0.190 *	0.088	−0.215 *	0.004	1
8.	0.081	0.171	0.141	0.061	0.086	0.401 **	−0.108	1
9.	0.428 **	−0.163	−0.023	0.092	0.044	−0.042	0.332 **	0.079	1
10.	0.419 **	0.057	0.169	0.261 **	−0.006	−0.124	0.135	0.108	0.440 **	1
11.	−0.052	0.088	0.247 **	−0.191 *	0.050	0.022	−0.129	−0.039	−0.040	−0.031	1
12.	0.301 **	−0.246 **	−0.086	0.141	−0.052	0.044	0.043	0.044	0.309 **	0.091	0.118	1
13.	−0.147	0.208 *	0.304 **	−0.102	0.255 **	−0.082	−0.069	0.155	−0.023	0.080	−0.002	−0.047	1

* Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed).

.

Reviewing the most significant correlations between the constructs employed, we found that in the context of our study:

• Employee personal energy-saving behaviours are most significantly correlated with their willingness to help their organisation conserve energy (r = 0.309 **), personal energy-saving norms (r = 0.301 **), and negatively with emotional exhaustion/burnout (r = −0.246 **).
• Employee collective energy-saving behaviours are most significantly correlated with their collective energy-saving responsibility and efficacy (r = 0.304 **), personal comfort/comfort levels (r = 0.255 **), and with emotional exhaustion/burnout (r = 0.208 *).
• Employee willingness to help the organisation conserve energy is most significantly correlated with their willingness to alter their personal energy consumption behaviour (r = 0.440 **), personal energy-saving norms (r = 0.428 **), willingness to alter energy consumption behaviour collectively (r = 0.309 **), and female gender (r = 0.332 **).
• Employee willingness to alter personal energy consumption behaviour is most significantly correlated with their willingness to help the organisation conserve energy (r = 0.440 **), personal energy-saving norms (r = 0.419 **), and awareness of energy wastage and knowledge of its solution (r = 0.261 **).
• Employee willingness to alter energy consumption behaviour collectively is most significantly correlated with their willingness to help the organisation conserve energy (r = 0.309 **), personal energy-saving norms (r = 0.301 **), and negatively with emotional exhaustion/burnout (r = −0.246 **).
• Employee personal energy-saving norms are most significantly correlated with awareness of energy wastage and knowledge of solution (r = 0.536 **), willingness to help the organisation conserve energy (r = 0.428 **), willingness to alter personal energy consumption behaviour (r = 0.419 **), personal energy-saving behaviours (r = 0.301 **), and female gender (r = 0.206 **).
• Employee emotional exhaustion/burnout is most significantly correlated with their collective energy-saving responsibility and efficacy (r = 0.313 **), collective energy-saving behaviours (r = 0.208 *), and negatively with personal energy-saving behaviours (r = −0.246 **).
• Employee sense of collective energy-saving responsibility and efficacy is most significantly correlated with collective energy-saving behaviours (r = 0.304 **), willingness to alter energy consumption behaviour collectively (r = 0.247 **), male gender (r = 0.190 *), and negatively with age (−0.215 **).
• Employee awareness of energy wastage and knowledge of its solution is most significantly correlated with personal energy-saving norms (r = 0.536 **), willingness to alter personal energy consumption behaviour (r = 0.261 **), and negatively with willingness to alter energy consumption behaviour collectively (r = −0.191 *).
• Employee personal comfort/comfort level is most significantly correlated with collective energy-saving behaviours (r = 0.255 **) and male gender (r = 0.215 *).
• Employee age is most significantly correlated with having children (r = 0.401 **) and negatively with collective energy-saving responsibility and efficacy (r = −215 *).
• Female gender is most significantly correlated with employee willingness to help the organisation conserve energy (r = 0.332 **) and personal energy-saving norms (r = 0.206 *).
• Male gender is most significantly correlated with an employee’s personal comfort/comfort levels (r = 0.215 *) and collective energy-saving responsibility and efficacy (r = 0.190 *).
• Having children is most significantly correlated with age (r = 0.401 **), indicating that the older the participants, the more probable it is to have children (an expected finding).

5. Discussion and Conclusions

5.1. Summary of Findings

Reviewing our findings, we found that an employee’s profile affects their willingness to alter their energy-consumption behaviour and to conserve energy at work, as well as directly affects their energy-saving habits and behaviour. Additionally, employee willingness also directly affects personal energy-saving habits and behaviour at work. More specifically, we found that:

(i) Employee willingness to help the organisation conserve energy is most strongly affected by the level personal energy-saving norms, awareness of energy wastage and knowledge of solution, and gender. This extends on existing evidence from the literature that suggests that personal norms are considered a significant predictor with a direct effect on and strong association with willingness to sacrifice in order to protect the environment [31] and that employees holding a strong belief of the importance of energy saving are more willing to save energy at work [15]. It is also in line with the fact that knowledge on how to conserve energy, as well as awareness of the multi-dimensional problem of energy sustainability and the intricacies of energy conservation, are considered to be enablers of pro-environmental behaviour and more sustainable energy-consumption behaviour, and important prerequisites for the development of pro-environmental and energy-saving willingness [50,51,52].

(ii) Employee willingness to alter personal energy consumption behaviour is most strongly affected by personal energy-saving norms. This extends on the recorded evidence in other studies that personal norms are considered a significant predictor with a direct effect on and strong association with willingness to sacrifice in order to protect the environment [31]. Moreover, it is also in line with the fact that researchers have found that energy-saving personal norms can be leveraged in organisational interventions to directly affect employee personal energy-saving intentions [19,26,29].

(iii) Employee willingness to alter energy-saving behaviour collectively is most strongly affected by collective energy-saving responsibility and efficacy at work. This is in line with the fact that enhancing personal and collective energy-saving efficacy has been found to in turn increase personal and collective energy-saving intentions [46], as well as by the fact that in shared organisational settings, collective efficacy influences employee energy-related behaviours [21].

(iv) Employee self-reported personal energy-saving behaviour is most strongly affected by personal energy-saving norms, emotional exhaustion/burnout, and willingness to help the organisation conserve energy. This extends on existing evidence that, according to values-beliefs-norms (VBN) theory pro-environmental behaviour, pro-environmental behaviour is directly affected by personal norms for pro-environmental action [28], and personal norms are considered a significant predictor with a direct effect and strong association of willingness to sacrifice in order to protect the environment [31]. It is also in line with the fact that energy-saving personal norms can be leveraged in organisational interventions to affect employees’ personal energy-saving actions [19,26,29]. Our findings are also extend on existing evidence that employees who experience emotional exhaustion tend to also exhibit low extra-role performance within their organisation [40], while employees holding a strong belief on the importance of energy saving are more willing to save energy at work [15].

(v) Employee self-reported collective energy-saving behaviour is most strongly affected by collective energy-saving responsibility and efficacy and personal comfort/comfort levels. This finding extend on existing evidence that collective responsibility and efficacy is considered a strong predictor of self-reported pro-environmental behaviour in collective settings [45,46]. It is also in line with the fact that employee energy-saving behaviour has been proven to also depend on their perceived personal comfort [15].

Therefore, in sum, with regards to hypothesis H1, we found that an employee’s profile indeed significantly affects their willingness to alter their personal energy-consumption behaviour and help their organisation conserve energy at work. Moreover, we found that their willingness to conserve energy at work is personally and not collectively ascribed. As per H2, we confirmed that an employee’s profile directly affects their personal and collective energy-saving habits and behaviour at work. With regards to H3, we also found a direct effect between an employee’s willingness and their personal energy-saving behaviour at work. These findings have both theoretical, as well as practical, implications.

5.2. Theoretical Implications

We have multiple theoretical contributions. First of all, we have provided evidence that when examining organisational energy-saving behaviour, we should consider two different perspectives: personal vs. collective energy-saving actions. Traditionally, researchers focus on individual-level factors and decision making when examining energy-consumption behaviour, but this perspective is limited since people belong to groups and need to act collectively in several contexts [46]. A workplace (and office buildings in particular) is one such context and although a number of energy-consumption behaviours can be enacted individually, there are many behaviours that require cooperation between employees, as energy is consumed collectively through shared devices and switching off a shared device needs to be agreed-upon between its co-users [76]. Indeed, in an organisational setting, employee energy conservation entails actions that could either be solely personally performed (e.g., operating a personal computer) or as a member of a group of employees (e.g., operating a space cooling device in a shared office space). It is only in these latter cases that the responsibility to conserve energy lies in groups and is therefore shared [31]. Hence, we have adopted a more holistic approach and joined our voice with few other researchers who have stressed that typically individuals do not act in isolation in the workplace but rather within groups of colleagues, and, thus, the spotlight should be also be aimed at their collective behaviour and perceptions. This study aspires to complement such sparse studies that focus on organisational settings and examine how collective efficacy influences the personal efficacy of employees and confirm the fact that energy-related behaviours in the workplace are regularly observed by peers [21]. We also highlight that, in this research, we examined collective behaviour of employees by assessing their perceptions of collective efficacy in addition to recording their views of their actual collective energy saving actions when they use devices used at a group-level by their colleagues, such as their use of a printer.

Furthermore, we provided additional support to existing evidence that suggest personal norms are a direct descendant of energy-saving behaviour, specifically in the context of workplaces and office workers. Moreover, we provided evidence that suggests that additional parameters may affect energy-saving behaviour at work, such as the level of employee emotional exhaustion (burnout), collective energy-saving responsibility and efficacy, awareness of energy wastage and knowledge of solution, and perceived personal comfort/comfort levels. These parameters are often taken into account in organisational research but have not adequately been taken into account thus far when studying employees’ pro-environmental and energy-saving behaviour. In addition, we found that employees’ willingness (i.e., to help the organisation conserve energy, to alter personal energy consumption behaviour, and to alter energy consumption behaviour collectively) is directly affected by their personal profile and, at the same time, it also has a direct effect on the employees’ energy saving habits and behaviours. Moreover, as outlined already in the summary of our findings that we offer, our findings are in line with existing evidence in the literature. However, we stress that they extend them in the context of energy saving by employees in workplace environments as opposed to the majority of existing evidence in the literature that pertains to pro-environmental behaviour in general or to non-organisational contexts.

Our study also confirmed the direct positive effect that age, female gender, and having children has on employee energy-saving norms, habits, and behaviours. We also found that male participants featured higher levels of perceived personal comfort at work. This is in line with the fact that engagement towards pro-environmental behaviour tends to increase with age, while women tend to have stronger environmental attitudes, concern, and behaviours than men across age [17,64].

5.3. Practical Implications

Based on our findings, practical implications also arise in the context of energy-conservation interventions in organisational environments. Importantly, when designing energy conservation interventions that, for example, utilise IoT devices for real-time monitoring/sensing of the energy consumption per employee and combine them with other mechanisms that cultivate behavioural change (e.g., gamification), the resulting behavioural model dictates the users’/employees’ personal factors that the interventions should consider and manipulate to increase energy conservation. First of all, we found that to motivate employees’ personal energy-saving behaviours, behavioural interventions should focus on increasing their willingness to help their organisation conserve energy, as well as amplifying their personal energy-saving norms, and not overly burden their daily routine or exacerbate their emotional exhaustion/burnout. Therefore, such behavioural interventions should be designed to be easy to follow and not overly demanding, time consuming, or pressuring. Moreover, in order to motivate employees’ collective energy-saving behaviours, such interventions should focus on increasing employees’ collective energy-saving responsibility and efficacy, while respecting their personal comfort/comfort levels (e.g., by not suggesting actions that would make the temperature in a room uncomfortable) and their emotional exhaustion/burnout (e.g., not bombarding employees with demanding requests towards energy-saving actions or repeatedly stressing their erroneous actions). Additionally, in order to increase employees’ willingness to conserve energy, interventions should primarily focus on improving their personal energy-saving norms at work, by increasing their awareness of energy wastage and knowledge of solution.

Employees have been known to imitate each other’s job crafting behaviours (changes they make on their job tasks), which can lead to both positive and negative organisational outcomes, depending on the specific circumstances [77,78]. Organisations can stimulate job crafting towards specific directions (increasing efficacy, expectations of goal accomplishment, and improving task performance) through training [77]. By providing individual feedback to employees regarding the way that they enact specific work-related activities, they can motivate employees to (re)consider and change their behavioural patterns [77,78]. Therefore, organisations can utilise performance feedback of energy-related behaviour to motivate employees towards crafting their job routines to include more energy-saving actions, as well as educational feedback to train employees on ways to save energy at work.

Energy has been described as invisible, abstract, and intangible, making it difficult for the average building user to understand how much energy is expended during their daily routine [21]. Real-time energy consumption feedback on specific behaviours is considered a particularly powerful driver for addressing this issue of “energy invisibility” (“making invisible energy visible”) by acting as a reference point from which to evaluate, and accordingly adjust, the economic or environmental impacts of energy behaviour [21,79]. However, although the workplace offers opportunities for energy savings, and the effect of energy-consumption-related feedback in offices can be large (7.4% on average in existing studies), it has not been adequately analysed thus far [18,80]. Feedback can be utilised towards training, modifying organisational procedures and norms, and increasing awareness levels regarding an employee’s own behaviour, as well as highlighting its consequences [14]. Moreover, personalised tailored information is more effective towards energy behaviour change [24].

Numerous IoT devices (equipped with sensing and short-range communication capabilities) are already embedded in the modern workplace environment, enabling interaction between employees and smart devices and providing more flexibility and customisation possibilities for end-users devices [81,82]. Consequently, the diverse features of IoT devices present unprecedented opportunities to increase user awareness (personal context and behavioural patterns), ambient awareness (environment/workplace context), and social awareness (patterns of social interaction) [81]. Improving energy-consuming appliances’ usage behaviour by utilising IoT-enabled feedback has accordingly been proven efficient in electricity conservation [83]. Based on all of the above, we have identified IoT-enabled, personalised, direct, and timely feedback as a means of promoting positive energy behavioural change to employees at work. Furthermore, as the content of this feedback should be aimed at priming their existing energy-conservation norms, educating them on ways to conserve energy, energy-behaviour models can be utilise in order to design personalised feedback mechanisms [84]. In essence, by taking into account employees’ personal profiles, as well as the surrounding environment (via IoT-enabled devices and sensors), researchers and practitioners can increase the chances of success in energy-saving behavioural interventions at work. In short, from a designer’s point of view, our findings could be taken into account as the user requirements that have been collected and identified in order to design and execute energy saving interventions that fit the workplace environments and the behaviour of employees.

5.4. Limitations and Suggestions for Future Work

As with all research, our work comes with its limitations. Firstly, our results have been based on a limited number of responses (119), while a larger sample of participants would have provided an even firmer basis for drawing safe conclusions. Moreover, we have yet to test our model in a real-life experiment that would record longitudinal data to prove its utility, as well as fortify or extend its theoretical and practical value. Although we have provided guidelines towards that direction, practical testing must be performed to verify them. Future research should therefore focus on the organisation of an energy-saving behavioural intervention that will test our hypotheses, preferably in IoT-equipped workplaces, providing personalised and context-relevant feedback to the participating employees. Moreover, future researchers may benefit by testing the effects of additional organisational parameters in connection to energy conservation behaviour, such as the effect of laziness or inattentiveness towards performing energy-saving actions.

Based on the R² results obtained through our multiple regression analyses, we acknowledge that in some cases low levels have been recorded. As already outlined, interpreting a regression coefficient that is statistically significant does not change based on the R² value [73,74]. Indeed, especially in studies that attempt to explain human behaviour, R² values tend to be less than 50% due to the inherent increased difficulty in predicting human behaviour compared to other physical processes [73,74]. However, although they are not as suitable in generating precise predictions, regression models with a low R-squared value and statistically significant independent variables can help in drawing important conclusions about the existing relationships between the variables in a model [73,74]. In order to further fortify our findings, future models can accordingly be tested in larger samples to further corroborate the statistical significance of the contribution of different factors in each model. In such endeavours, it is standard practice to use the recorded coefficient p-values to decide whether to include variables in the models [74]. Hence, future researchers may focus on examining the exclusion of factors that have exhibited lower contribution to the models that we presented and analysed in this study or they may wish to focus on the inclusion of additional factors that may contribute to explanatory strength.

Finally, we acknowledge that the differences recorded in the adherence of the participants in the various behavioural factors measured through our questionnaire may be studied in more depth in the future. Explaining the reasons why, for example, employees’ willingness to help their organisation conserve energy at work was found to be more prominent compared to willingness to conserve energy collectively, would be interesting towards clarifying the group dynamics of energy conservation at work.