Ph.D. Dissertation Social Media Management for Consumer

Faculty of Computer Science and Management

Ph.D. Dissertation

Social Media Management for Consumer Awareness and Acceptance of Smart Meters

Yash Chawla

Supervisor: dr hab. inż. Grzegorz Chodak, prof. PWr

Auxillary Supervisor: dr Kamila Ludwikowska

Wrocław, 2020

2 You see things; and you say “Why?” But I dream things that never were; and I say “Why not?”

GEORGE BERNARD SHAW Nobel Laureate in Literature 4 Abstract

Innovations are entering the market so rapidly that managing these innovations and ensuring that consumers are aware of its full potential, is a huge challenge. Energy markets, around the world, have been experiencing signiﬁcant changes and an inﬂux of innovative technologies, such as Electricity Smart Meters (SM), which are an integral element of Smart Grids (SG). This study explores the consumer willingness and acceptance of SM, their preferred communication channels and recommends a social media management plan that would be effective for enhancing diffusion of SM. Results derived through an empirical survey among social media users, in four countries, show that there is still a lack of knowledge about SM among consumers and more marketing communications are required to facilitate the acceptance of SM. Social media can play a major role in these marketing communications and its effective strategy has also been discussed with empirical evidences and experiments in real business environment.

This work was partially supported by the following two projects funded by the National Science Center (NCN, Poland) Logistics, trade and consumer decisions in the age of the Internet • (Grant no. 2018/29/B/HS4/02857; PI: Prof. David Ramsey)

Segmentation of electrical energy consumers using the stage-change model: • Analysis of factors enhancing adoption of demand side management tools (Grant no. 2016/23/B/HS4/00650; PI: Prof. Anna Kowalska-Pyzalska) 6 Contents

1 Introduction 3 1.1 Motivation ...... 3 1.2 Aims and research objectives ...... 4 1.3 Methodology ...... 5

2 Innovation and Marketing 7

3 Summary of results and core articles 9 3.1 Consumers and Smart Meters ...... 9 3.1.1 RO1. Consumers’ awareness and willingness regarding SM ...... 10 3.1.2 RO2. Sources of Information and Communication Channels ...... 13 3.1.3 Publication Details (Papers 1-4) ...... 14 3.2 Social Media Effectiveness and Management ...... 15 3.2.1 RO3. Effectiveness of different content types on social media . . . . . 16 3.2.2 RO4. Social Media Management for SM ...... 17 3.2.3 Publication Details (Papers 5-7) ...... 19 3.3 Auxiliary results ...... 20

4 Conclusions 23

Acknowledgements 25

Bibliography 27

Appendix A: Papers 1-7 33

Appendix B: Co-authorship declarations 193

1 2 CONTENTS Chapter 1

Introduction

1.1 Motivation

The 21st century has been deemed as the new age of innovation, where sustainability is one of the key issues that are to be addressed. Innovations are entering into the market so rapidly that managing these innovations and ensuring that consumers are aware of its full potential is a huge challenge. One of the key issues, for managing innovation, is the innovation itself and the concurrence between technical and social elements of innovation (Xu et al., 2007). The social elements or social acceptance is a very important determinant for the smooth diffusion of any innovation in the market (Gouws and Van Rheede van Oudtshoorn, 2011). There are several models which study the diffusion of innovation, from which Roger’s model for innovation diffusion is one of the most widely used (Rogers, 2003). This model puts particular emphasis on communication channels, which are used to market the innovations to the consumers. Different users have different preferences of communication channels, even more so in this age of advanced information technology and the internet (Danaher and Rossiter, 2011). The sector of information technology, highly driven by innovation, has directly or indirectly influenced all walks of life (Ma et al., 2014). The beginning of the 21st century has seen an exponential growth in the number of internet users, with the current global user base of over 4.5 billion. Businesses have adopted new business models that allow them to utilize the opportunities that the internet has to offer (Wielki, 2010). Evolution of social media, with this huge penetration of internet services, has given the general population and businesses, an effective and cheap tool to communicate worldwide. As of 2020, the global number of social media users stand at over 3.8 billion (Kemp, 2020). Social media has been widely used for businesses to market their products and services to the consumers globally and is highly important from the aspect of direct marketing in today’s digital economy (Unold, 2003). Social media is very effective in generating personalized social influence, which enhances the effect of information diffusion and raising public awareness (Booth and Matic, 2011). In some sectors, such as e-commerce, the use of social media is quite high, whereas, in sectors, such as ’energy markets’, there is still scope for large improvement (Accenture, 2015). Energy markets, around the world, have been experiencing significant changes and an influx of innovative technologies, such as Electricity Smart Meters (SM), which are an integral element of Smart Grids (SG) (Verbong et al., 2013; Faruqui and Sergici, 2010). Soon, electricity will become a technology that is tangible and would require

3 4 CHAPTER 1. INTRODUCTION the attention and decision-making from consumers (Kowalski and Matusiak, 2019). In 2009, the Electricity Directive of the European Commission 2009/72/EC, stated that all EU Member States should roll out SM to at least 80% of consumers by 2020. One of major motives behind this was to draw consumer engagement in the energy markets and was subject to the cost benefit analysis. Consumers active engagement in the energy markets, through SM, would provide a demand side response to balance the grids and can help save energy (Soroczynski´ and Szkutnik, 2015). Usage of SM would also provide additional facilities to consumers, such as switching between suppliers, resulting in a more competitive market and lower tariffs (British Infrastruc- ture Group of Parliamentarians, 2018), and can also be looked at as a step closer to consumers gaining more control of their energy consumption (Kowalska-Pyzalska and Byrka, 2019; Weron et al., 2018). However, a number of researchers have expressed concerns regarding the low level of knowledge and engagement towards SM, such as new solutions in the energy markets (Claudy et al., 2010; Verbong et al., 2013; van der Werff and Steg, 2016; Ellabban and Abu-Rub, 2016). Raising the knowledge or awareness regarding SM would lead to higher acceptance and engagement among consumers (Kowalska-Pyzalska and Byrka, 2019). There have been several studies in the literature that have analyzed consumer awareness and acceptance of SM (as discussed in detail in Chawla and Kowalska-Pyzalska (2019); Chawla et al. (2019b,c, 2020)), but there is a gap in the literature when it comes to studies being conducted among social media users. Social media has been proven to be effective for raising awareness of consumers and also for generating social influence that increases the acceptance of innovation (Booth and Matic, 2011). Moreover, social media users have been found to be early adopters of technology and possess the power to influence the early majority (Droge et al., 2010; Lipschultz, 2017; Khamis et al., 2017). Hence, studying the factors affecting the awareness, acceptance and preferences of social media users, regarding SM, would provide insights to energy companies on how to manage their social media accounts and enhance the diffusion of SM. A combination of all these factors motivated us to study social media management for consumer awareness and acceptance of smart meters. The remainder of the thesis is structured as follows. In Sections 1.2 & 1.3, of Chap- ter 1, the aims, objectives and methodology are discussed briefly. Thereafter, in Chap- ter 2, the objectives of this thesis are discussed with regard to the two sub-disciplines of management sciences, innovation management and marketing. In Chapter 3, the detailed results, with regard to each objective and the corresponding publications, are described. In Chapter 4, the summary of main results and the drawn conclusions are presented. Finally, the Appendices that follow the Bibliography include the 7 papers constituting the thesis and scanned co-authorship declarations for these papers.

1.2 Aims and research objectives

The main aim of this thesis is to highlight the factors for social media management, which can enhance consumer awareness and acceptance of smart meters. To cater to this aim, the following objectives have been set. These objects are interesting from the point of view of basic research, as well as for their managerial or practical implications.

RO1: To investigate the attitudes, preferences and fears, regarding aware- • ness, willingness and acceptance of SM, among social media users. 1.3. METHODOLOGY 5

RO2: To explore the various sources of information regarding electricity in • general and SM in particular. RO3: To test the effectiveness of different types of content on social media • and device metrics, through which managers can interpret the results of their campaigns. RO4: To create a social media management plan that would be useful for • energy companies to enhance the diffusion of SM. The first objective particularly concentrates on class of audiences that have previously not been studied in the literature, being the social media users. There are several studies in the literature regarding consumer willingness and acceptance of SM (see e.g. Krishnamurti et al., 2012; Ellabban and Abu-Rub, 2016; Chou and Yutami, 2014; Rocha, 2016), but none of which have been conducted among social media users. There were similar studies conducted among social media users in other sectors, such as acceptance of renewable energy (see e.g. Boutakidis et al., 2014), which encouraged further study for SM. The second objective compliments the first one, being the sources of information looked at, through which the consumers (who are social media users), receive their information for electricity, in general, and SM, in particular. Although these are social media users, the proposed hypothesis was that they would also prefer to receive SM related information from multiple sources. It was also interesting to see how preferences of different communication channels would affect the preference, fears and acceptance of SM among consumers. In the third objective, through independent experiments, recommendations for the type of content and web-link placements for the social media content were made. In this part, metrics for measuring the effectiveness of social media promotions were defined, which can be used for researchers or managers. Finally, the fourth objective would be directly associated with the results of the first three objectives. The social media management plan was outlined for positively affecting the consumer willingness and acceptance of SM.

1.3 Methodology

The whole research was divided into two parts as shown in Figure 1.1. First part, which concentrated on SM (the innovation) and, the second, on social media (the marketing medium). In the first part, an empirical survey was carried out among social media users in four countries, Poland, Portugal, Indonesia and Turkey, for obtaining the results pertaining to objectives, 1, 2 & 4 (partly). In total, 2047 responses were collected from these four countries (more details are discussed in Section 3.1). The data obtained was analyzed using Tobit, Probit and Logit regression models (for detailed descriptions see Papers 1-4 in Appendix A), to understand the significant factors effecting the preferences, concerns, attitudes, awareness and willingness about SM. Thereafter, using Student’s T-test, the correlations between the variables and communication platforms (both conventional and social media) were found. This was done to recognize the marketing content that would be useful for the specific platform, which was being considered (see Paper 2). In the second part, two empirical experiments were carried out in a real business environment for fulfilling the objectives 3 & 4 (partly). In the first experiment, organic and paid promotions of different types of content was carried out on the 6 CHAPTER 1. INTRODUCTION

Facebook fan page of a Polish business and observations were recorded. Using various metrics for social media marketing effectiveness, the observations were analyzed (for a detailed description see Paper 5). The second experiment was an extension of the ﬁrst, where only organic promotions of posts, with different link placements, was carried out on the Facebook fan page of the same Polish business. Observations recorded were analysed using simple linear regression and social media metrics (for a detailed description see Paper 6).

Figure 1.1: Research Framework Chapter 2

Innovation and Marketing

There has been a wide debate in the literature about the position of ’Management Sciences’ as a scientific discipline and it’s sub-disciplines (Cyfert et al., 2014; Sudoł, 2016). Various organizations, based on different factors, have given wide range of sub-disciplines for management sciences. The European Academy of Management (EURAM) created 13 strategic interest groups covering a wide and transversal management discipline (EURAM, 2009). Academy of Management (AOM), USA, has divided management into 25 disciplines, represented by divisions and interest groups (AOM, 2010). These groups were primarily created on the basis of member interests and to group professional interaction and involvement. Cyfert et al. (2014), defined 21 sub-disciplines within the principal discipline of Management Sciences and elaborated on the scope of each division, based on scientific literature. This thesis was concerned with two out of these 21 sub-disciplines, Innovation Management and Marketing Man- agement, as shown in Figure 2.1, hence, it can have its place in the Management Sci- ences discipline. There are several aspects of innovation management. From its essence, the ty- pology and source, to the management models and commercialization of innovation (Cyfert et al., 2014), researchers have found definitive proof of an innovation’s positive performance and success with the application of innovation management processes (Tidd and Thuriaux-Alemán, 2016). There has been extensive research in innovation management, which covered the synergistic patterns between product and processes (Xu et al., 2007). More focus was given to the technical aspect of it, whereas, the social elements were not highlighted equally. Among the various concepts of innovation management, consumer adoption is one of the key social elements for an innovation’s success in society (Rogers, 1962). All consumers are not the same and have different tendencies towards accepting innovation. The literature regarding consumer adoption of innovation has largely relied upon the four elements given by Rogers (2003), the innovation itself, communication channels, time, and a social network (system). The innovative product, in this thesis, was Smart Electricity Meters (SM), which has been deemed as one of the biggest innovative developments in the energy markets which is indispensable (Van Gerwen et al., 2006). The consumer was concentrated on in this case, being the social media users, and were known in the literature as early adopters of innovative communication technology, possessing the potential to influence the early majority (Droge et al., 2010; Lipschultz, 2017). This constitutes two of the five groups of consumers segmented by Rogers (2003), innovators, early adopters, early majority, late majority and laggards. Communication channels are the link to

7 8 CHAPTER 2. MANAGING INNOVATION AND MARKETING

Figure 2.1: Management Science and its sub-discipline, addressed in this thesis. Adapted from: Cyfert et al. (2014) marketing management, where social media platforms and consumer preferences, in terms of other communication platforms, were focused on and checked. Marketing management has also seen a huge shift with the evolution of the internet and social media. There has been a major focus shift onto the customers, the ben- eﬁts of a product or a service and knowledge resources (Webster Jr, 2005). For the energy sector, Nakarado (1996), had highlighted the importance of marketing management and understanding prospective consumers of new technologies. In recent years, catching up with the global trends, energy companies have also started using social media and other digital media for reaching out to consumers. However, their efforts have not yet reached critical masses. Management of social media and other digital media marketing for the energy sector is an important area of investigation, as the consumers become more habituated to modern information and communication technologies (Barrios-O’Neill and Schuitema, 2016). Researchers have studied marketing management for the energy sector, in terms of communication management for the public sector and the need to update strategies for the diffusion of information among consumers (Bogdal, 2013), effects of the community based marketing and lack of proper information (Streimikiene and Vveinhardt, 2015), requirements for marketing strategies to be tailored for speciﬁc groups of consumers (Hille et al., 2019) and the need for energy companies to put in more effort for marketing and communicating with the consumers (Gong et al., 2019). This thesis caters to these challenges emphasized in the literature, and enrich the literature in Management Sciences domain by studying various consumer traits and attributes, as well as their role in content creation for the diffusion of innovation. Chapter 3

Summary of results and core articles

The thesis constitutes 7 core articles, divided into two parts. The first part, discussed in Section 3.1, consists of 4 articles (Papers 1-4, see Appendix A) and concerns of consumers, communication channels and smart meters. These papers were based on the analysis of survey data, collected from social media users in Turkey, Portugal, Poland and Indonesia. In each paper, the levels of awareness and willingness regarding SM were checked, as well as the various social media channels being used, sources of information regarding electricity, in general, and SM, in particular, consumer preferences and concerns regarding SM. Further in Paper 1 and Paper 2, significant factors are discussed, that need to be addressed for raising awareness of consumers regarding SM. In Paper 3 and Paper 4, models were created for the willingness to accept SM under various conditions and to understand the factors that could be addressed in raising acceptance of SM among consumers. Additionally in Paper 2, it was shown how marketing content for each specific communication platform can be deduced through student’s t-test for independent samples and different variances. The second part, discussed in Section 3.2, consists of 3 publications (Papers 5- 7, see Appendix A) and describes the effectiveness of different social media content types and the recommended social media plan. In paper 5, we discussed the first experiment and its results regarding the effectiveness of image, video and album type of posts on Facebook. in Paper 6, the second experiment and its results regarding the effectiveness of organic promotions of web links on Facebook, were discussed. In both these papers, respective metrics for measuring campaign performance, were also formulated. In Paper 7, based on the literature, we discuss recommendations for social media activities for individuals, businesses as well as government.

3.1 Consumers and Smart Meters

The first part of the thesis was concentrated on the current market situation, regarding SM, and various consumer factors, such as their preferences willingness, concerns, awareness, behaviours, sources of information, social media channels used and so on. This part consisted of two research objectives, RO1 and RO2, the results of which are discussed in the following sub-sections 3.1.1 and 3.1.2, respectively. Although several studies in the literature were related to the research objective RO1, there were no studies found to have been conducted specifically among social media users. Moreover, the outcome of RO1 was just the first piece of the overall research, as significant factors were further connected to communications channels, yielding recommendations

9 10 CHAPTER 3. SUMMARY OF RESULTS AND CORE ARTICLES for promotional content, specific to each communication channel, which has not yet been done in the literature. The results of both these research objectives were based on empirical data, collected through an online survey and conducted among social media users in Poland, Portugal, Turkey and Indonesia. The number of responses collected from each country are shown in Table 3.1. The survey was first conducted in Poland, then in Portugal, followed by Indonesia and, finally, Turkey. The reason for carrying out the survey repeatedly was to evolve the questionnaire each time with the feedback received from respondents, collaborators and peer-reviewers. In the first survey conducted in Poland, there were 53 variables with no control groups, which evolved to 63 variables and two control groups in the final survey carried out in Turkey. The list of variables of each country has been included in Papers 1-4. For practical implementation of the plan stated further in this thesis, the variables described in Paper 1, are recommended to be used.

Table 3.1: Number of responses collected from each country Country Poland Portugal Indonesia Turkey Total No. of Responses 505 518 519 504 2046

Poland and Portugal, being a part of the European Union, are mandated to follow the Electricity Directive of the European Commission established in 2009/72/EC. Turkey has announced that it would be following the European Commission’s guidelines regarding the roll-out of SM (Theron, 2015), whereas Indonesia has devised its own roll-out plan (Perindustrian, 2018). Thus, the four countries are a peculiar mix of EU counties, a country following EU policy and a country having its own independent policy. Additionally, each of these countries were found to be on similar level of SM diffusion among consumers, which was the criteria through which the countries were chosen. In each of the three countries, there was a collaboration with a local researcher, who was also the co-author for the publication for the respective countries.

3.1.1 RO1. Consumers’ awareness and willingness regarding SM In this part of the research, the goals were: (i) to check the level of awareness regarding SM (see Papers 1-4)) and deduce significant factors correlating with the awareness (see Papers 1-2); (ii) to check the level of willingness to accept SM under various conditions (see Papers 1-4)) and deduce significant factors correlating with the acceptance (see Papers 3-4). The levels of awareness and acceptance were used to determine the need for more information dissemination. On the other hand, the significant variables deduced through modelling for awareness and willingness indicated the topics, which could appeal to the consumers for the respective objective. In the literature, for each of the four countries, there was low level of awareness regarding SM and its benefits (Kowalska-Pyzalska and Byrka, 2019; Ghazvini et al., 2019; Theron, 2017; Kamarudin and Boothman, 2017). The findings re-affirmed that there is still a lack of awareness regarding SM in these countries. Figure 3.1, shows the portion of respondents who knew what a SM was in each of the four countries, which emphasised that more efforts are required to raise awareness among the consumers regarding SM. Through the analysis of survey results, it was found that 54.1% 3.1. CONSUMERS AND SMART METERS 11 of respondents in Poland, 62.7% in Portugal, 60.8% in Indonesia and 54.2% in Turkey, expressed the desire to know more details about SM. This also indicated the need for energy companies to reach out to the consumers and provide detailed information regarding SM.

Figure 3.1: Level of awareness regarding SM among the respondents in the study from various countries.

For the willingness measure, several conditions were given to the respondents and they were asked whether they would be willing to accept SM under those circumstances. Figure 3.2, shows the responses recorded. It can be seen that, apart from a couple of exceptions, the responses are clustered. This indicates quite similar willingness and acceptance conditions in all four countries. One possible reason for this similarity is that these were all social media users and were engaged with online activities that have similar attributes. However, it is essential to check the statistically signiﬁcant correlations, with the responses to all the other questions, to understand the factors responsible for these results. Hence we further carried out modelling for awareness and acceptance of SM.

Modeling for Awareness and Acceptance of SM Raising the awareness of consumers, regarding SM, would lead to a higher acceptance (Kowalska-Pyzalska and Byrka, 2019). To raise awareness or to try and convenience the consumers regarding their acceptance, it is very important to understand what attracts the consumer (Zhou et al., 2019). Understanding consumer expectations and preferences helps businesses to adopt the appropriate strategies to enhance the diffusion of products or services in the market (Fiore et al., 2017). Marketing content, created on basis of what a consumer can relate with, would attract consumers and, in turn, positively effect their decision to accept a certain message (Lee and Hong, 2016; Guelman et al., 2015). The purpose of identifying the signiﬁcant factors for awareness or acceptance of SM, is to understand the factors that the consumers can relate with, in terms of SM. To obtain the results for signiﬁcant factors effecting the knowledge / awareness regarding SM, Logistic regression models were used to analyzed the data collected from Turkey (shown in sub-section 4.10 of Paper 1) and from Portugal (shown in sub-section 4.4 of Paper 2). This gave an understanding of the factors that can be 12 CHAPTER 3. SUMMARY OF RESULTS AND CORE ARTICLES

Figure 3.2: Willingness to accept SM under various conditions among the respondents in the study from various countries. Note, that conditions De7 and De8 were not asked to respondents from Poland, hence, their values cannot be seen in the graph, and that De5* was asked to all the respondents in Poland, but was only asked to respondents who knew what an SM was for Portugal, Indonesia and Turkey. 3.1. CONSUMERS AND SMART METERS 13

Figure 3.3: Percentage of social media users recorded in the survey and the actual users* in the report by Kemp (2019). For Poland, respondents were not explicitly asked about the social media channels, hence, the data is inconclusive and not included in the figure. Actual users* percentage, indicates the percentage of internet users using a particular social media platform in the respective countries. addressed to increase the awareness regarding SM among consumers. To obtain the significant factors effecting the willingness to accept SM under various conditions, To- bit models were created for each condition. For data collected from Poland, six models were built (shown in sub-section 5.3 of Paper 3), whereas, for the data collected from Indonesia, eight models were built (shown in section 4 of Paper 4). It is recommended that the initial objective of the energy companies should be to raise awareness of the consumers, hence, content related to significant factors, found through modelling for awareness, should be used for content generation. Campaigns, for raising the acceptance of SM, would complement the awareness campaign and add further value to it.

3.1.2 RO2. Sources of Information and Communication Channels Consumers’ preferences of communication channels play a vital role in the diffusion of information, as well as addressing their concerns (Khan et al., 2013). Using multiple marketing communication channels or adopting an omni-channel strategy has become vital in the current competitive markets (Payne et al., 2017). Through the survey, respondents were given questions regarding the social media channels they were active on and sources through which they received information regarding electricity, in general, and SM, in particular. The responses are discussed in Papers 1-4. Additionally, respondents in Turkey and Indonesia were asked about the communication channel they would use to seek more information regarding SM (see Paper 1 and Paper 4). Even though the study was conducted among social media users, it was found that the consumers preferred to receive information through a variety of channels. Though 14 CHAPTER 3. SUMMARY OF RESULTS AND CORE ARTICLES the findings and preference of communication channels are discussed in Papers 1-4 (see Appendix A), the objective within this thesis was to specifically concentrate on social media channels. Figure 3.3, shows the distribution of the percentage of respondents who indicated using a particular social media platform. Alongside the responses of the survey respondents, the distribution of the percentage of internet users, who used a particular social media platform in the respective countries, is shown. On average, in all the countries, each respondent was active on at least two social media platforms and 10% of respondents, indicated to be using either six or more platforms out of the eight included in the study. This showed the users were active on multiple platforms and, because each platform had its own attractive features, it became important to understand which content would be attractive on particular platforms. For this purpose, significant negative co- relations, between the variables in the study and each of the social media channels, were examined. Negative co-relations between a variable and a social media platform suggested that there was scope for improving information / content addressing the particular variable on the co-related social media platform. This result, took the general content suggestions, obtained from the modeling for willingness and acceptance variables, towards more specificity. The detailed analysis was carried out for Portugal (see sub-section 4.5 of Paper 2), where the social media channels, as well as other conventional channels, were considered.

3.1.3 Publication Details (Papers 1-4) Paper 1 published as: Y. Chawla, A. Kowalska-Pyzalska, B. Oralhan (2020), • Attitudes and Opinions of Social Media Users Towards Smart Meters’ Rollout in Turkey, Energies, 13(3), 732.

– JCR classiﬁcation: Energy & Fuels, IF5Y = 2.990, MNiSW 140p. – My contribution amounted ca. 60%. I conceived and designed the survey; collaborated with the foreign author, managed the data from the respondents on social media; analyzed the data using Gretl and SPSS; and together with A.K-P: drafted, reviewed, edited and revised the manuscript.

Paper 2 accepted for publication as: Y. Chawla, A. Kowalska-Pyzalska, P.D. Sil- • veria (2020), Marketing and communications channels for diffusion of electricity smart meters in Portugal, Telematics and Informatics.

– JCR classiﬁcation: Information Science, IF5Y = 3.768, MNiSW 140p. – My contribution amounted ca. 50%. I conceived and designed the survey; collaborated with the foreign author, managed the data from the respondents on social media; reviewed the literature; discussed the analyzed data; drew out; the content recommendations and together with A.K-P: drafted, reviewed, edited and revised the manuscript.

Paper 3 published as: Y. Chawla, A. Kowalska-Pyzalska (2019), Public Aware- • ness and Consumer Acceptance of Smart Meters among Polish Social Media Users, Energies, 12(14), 2759.

– The journal is classiﬁed by the Journal Citation Reports (JCR) in category Energy & Fuels. Its 5-year impact factor IF5Y = 2.990 and it has 140 3.2. SOCIAL MEDIA EFFECTIVENESS AND MANAGEMENT 15

points in the ranking of the Polish Ministry of Science and Higher Educa- tion (MNiSW) for years 2018-2020. – My contribution amounted ca. 50%. I conceived and designed the survey; collected and managed the data from the respondents on social media; reviewed the literature; and together with the co-author: drafted, reviewed, edited and revised the manuscript.

Paper 4 published as: Y. Chawla, A. Kowalska-Pyzalska, W. Widayat (2019), • Consumer Willingness and Acceptance of Smart Meters in Indonesia, Resources, 8(4), 177.

– Scopus classiﬁcation: Environmental Science: Management, Monitoring, Policy and Law, Cite Score = 2.60; MNiSW 100p. – My contribution amounted ca. 60%. I conceived and designed the survey; collaborated with the foreign author, managed the data from the respondents on social media; analyzed the data using Gretl and SPSS; and together with A.K-P: drafted, reviewed, edited and revised the manuscript.

3.2 Social Media Effectiveness and Management

Once the content material was derived through modeling and testing, as discussed in section 3.1, the next step was to decide the type of content to be used. Some platforms, such as Facebook, LinkedIn, Twitter and so on, allow for different content types to be displayed, including images, videos, multiple photos, text, web links or combinations of these, whereas other platforms, like YouTube, Instagram and so on, have a fixed type of content that can be disseminated on them. Facebook is currently the 3rd most visited site on the internet and has the largest user base among social media platforms, with over 2.45 billion users globally (Kemp, 2020). In terms of business promotions, 89% of advertisers prefer to use Facebook (Zote, 2020). Large numbers of active users, wide dimensions of promotion options, usability of content type and easy access to basic campaign insights, are some of the reasons for this immense popularity (Pongpaew et al., 2017). Hence, it was decided to conduct experiments on Facebook, with a vigorous experiment design and detailed recording of observations. While doing so, we kept in mind that the same could be replicated on other platforms as well. Experiments were conducted in real business environments, specifically, on the Facebook fan page of a live business in Poland. Contact with energy companies was attempted in each of the four countries, however, consent was not granted to run the experiments through their social media handles. This remains as the future scope of research, based on the results within this thesis. As establishing the method of conducting the experiment and calcu- lating the performance of various post types was concentrated on, it was justifiable to carry out the experiment on a live social media handle of a business, even though the business belonged to a different sector. In sub-section 3.2.1, the results are discussed of the experiments and effectiveness metrics of various types of posts. Thereafter, in sub-section 3.2.2, the social media plan is explained, which combines the discussions in sub-sections 3.1.1, 3.1.2 & 3.2.1. 16 CHAPTER 3. SUMMARY OF RESULTS AND CORE ARTICLES

3.2.1 RO3. Effectiveness of different content types on social media There were two experiments conducted to fulfilling this research objective. The first experiment and results are discussed in Paper 5, and the second are discussed in Paper 6. To check the effectiveness of the different types of content, the experiment needed to first have simple measures, which would provide an understanding of the campaign performance and has also been discussed in Papers 5 & 6. In the literature, there was an observed lack of agreement about which social media marketing indicators would accurately define campaign performance (Lamberton and Stephen, 2016). Content can been seen as having three distinct aspects: quality (e.g. interactivity, vividness, education, entertainment, information), valence (e.g. emotions, tonality, rating variance), and volume (counts and volumes) (Peters et al., 2013). This study only sought to measure the difference between the interaction patterns, so it concentrated on the third aspect, which is the volume. Volume related measures are easy, as observations can be recorded through the insights given by the social media platform. These measures can be treated as quality indicators, especially if the analysis is based on the observed action and mutual relations, which is what we did in the experiments. In the first experiment (see Paper 5 in Appendix A), three types of posts were considered, as shown in Figure 3.4, as well as constructed metrics related to goal attainment, content valence, content quality and content volume. Each could be calculated for individual post types, compared mutually and calculated for the campaign overall. Each of the metrics, for the first experiment, are described in Section 3 of Paper 5.

Figure 3.4: Three post types used in the ﬁrst experiment.

The results of the experiment showed that videos (post Type 2) had a greater reach, whereas, the images (post Type 1) had greater engagement. However, in terms of business goals, video was arguably more effective. The results also suggested that, even though a number of clicks were generated, they were not the desirable clicks. 3.2. SOCIAL MEDIA EFFECTIVENESS AND MANAGEMENT 17

Only a small portion of the clicks were on the link, which would lead the user to the landing page with more information or the product buying option, in case of this experiment. Hence, there was a need to ﬁnd a more effective post type, which would take the users to the company website with more information. This led to experiment two, described in Paper 6. In this experiment, four different post types were used, with a prime focus on weblinks only. Social media is considered as a way to generate trafﬁc for the websites (Treadaway and Smith, 2012). In the scope of this thesis, regarding SM, it was observed that the energy companies had good content and information regarding SM on their websites. Generating more visits by the consumers on such pages, with the use of social media, would be step in the desired direction. Organic reach through Facebook page posts, especially with a weblink, has died out in the past few years. Organic reach refers to how many people you can reach for free on Facebook by posting to your page (Boland, 2014). Facebook’s algorithm and its interpretation of posts is one of the reasons for the declined reach (Cooper, 2020). Through this second experiment, the focus was to examine where would the best location would be to place a weblink in a post to get the maximum organic reach possible. Links were placed in four different locations, as shown in Figure 3.5, the results of which are described in Section 4 of Paper 6.

Figure 3.5: Four different post types, based on positioning of the web links, used in the second experiment.

Results showed that, placing the link in the comment was the most effective, in terms of the majority of metrics. It also generated the weblink clicks, as desired. Another important ﬁnding was the graph of metrics revealed the optimum time of posting, as well as the time interval between two posts. This would ensure maximum efﬁciency of each post, without any hindrance from the researcher’s activities.

3.2.2 RO4. Social Media Management for SM This part of the thesis brings all the previous results together to form a social media plan outline, which can be used by managers in energy companies. Based on 18 CHAPTER 3. SUMMARY OF RESULTS AND CORE ARTICLES the literature findings, Paper 7 briefly outlines the recommendations for individuals, businesses and governments to effectively use social media. Results, obtained in the previous parts of this thesis, add detailed inputs to those preliminary findings. A four step social media plan is recommended, which would follow a similar methodology used in this thesis. Figure 3.6 shows the outline of the plan that is recommended.

Figure 3.6: Outline of social media management plan to enhance awareness and acceptance of SM

Step 1 : Understand the consumers and the effectiveness of content presentation This step would include conducting a survey among consumers that are targeted. Responses to variables listed in Paper 1, are recommended to be obtained. Responses collected can then be used to determine the level of acceptance and awareness. De- pending on the results, goals can be set, in terms of raising the level of acceptance and awareness among consumers. The survey data would also indicate the interest among consumers, social media channels they use, as well as where they would prefer to seek information regarding SM. The second action, within this step, would be to test the different post types and calculate their effectiveness. For this purpose, the experiment design methodology, respectively used in Papers 5 and 6, can be utilized. The same can also be used for social media platforms, other than Facebook.

Step 2 : Analysis of the data and derivation of the content In this step, the collected data should be analysed using regression models. De- pending on the goal set forth, either of the methods, used in Papers 1-4 (see Appendix A), can be utilized. The yielded signiﬁcant factors should be used to create content outlines, as described in sub-section 3.1.1. Moreover, in this step, the test of correlations, as described in Paper 2, should be carried out. This test should be targeted speciﬁcally for social media platforms, which were preferred by the consumers. How- 3.2. SOCIAL MEDIA EFFECTIVENESS AND MANAGEMENT 19 ever, the platforms in consideration can be expanded in number, as well as, carried to non-social media channels. Lastly, through the results obtained and analyzing the social media metrics, this would yield effective content types for respective platforms. Additionally, the productive publishing times and interval between the posts should also be noted from this analysis.

Step 3 : Content creation and campaign design Using the content material obtained through the regression models and t-tests, the effective content type recognized through the experiment results were promotional content. As a result, the promotional campaign should be designed, taking into account the effective time of posting, platforms preferred by the consumers and the content created in this step. At the same time, the metrics must be selected that would be used to measure the outcomes of the campaign. Once these preparations are completed, the campaign execution should be initiated.

Step 4: Monitoring and feedback Once the campaign is online on social media, careful and constant monitoring is a must (Stauss and Seidel, 2019). This is to ensure that the comments are responded to, unnecessary comments are deleted and any customer queries are sufﬁciently answered. Social media is a two-way or multi-way communication medium and, to harness its full potential, it is important to engage with the consumer (Abeza et al., 2013). Depending on the intensity of the campaign, that is, the number of posts being published with time, observations should be recorded to analyze the performance through the metrics. This can be scheduled during the campaign, if the campaign is designed for four or more days, or, it can be at the end of the campaign if it is for a shorter duration. Performance should be analyzed and recorded and, the performance, along with any feedback received during monitoring, should be taken into account while creating the next campaign.

3.2.3 Publication Details (Papers 5-7) Paper 5 published as: G. Chodak, Y. Chawla, A. Dzidowski, K. Ludwikowska • (2019), The effectiveness of marketing communication in social media., Proceed- ings of the 6th European Conference on Social Media, ECSM 2019: University of Brighton, UK, 13-14 June 2019 / Ed. by Wybe Popma and Stuart Francis. Sonning Common: Academic Conferences and Publishing International Lim- ited, pp. 73-81.

– Indexed in Web of Science. – My contribution amounted to ca. 35%. I, along with G.C. developed the concept of the paper and the experimental design; along with G.C. conducted the experiment; analyzed the data; along with the co-authors drafted, reviewed, edited and revised the manuscript.

Paper 6 submitted for review as: Y. Chawla, G. Chodak (2020), Social Me- • dia Marketing for Businesses: Organic Promotions of Web-Links on Facebook, Journal of Business Research

– JCR classiﬁcation: Business, IF5Y = 4.747, MNiSW 140p. 20 CHAPTER 3. SUMMARY OF RESULTS AND CORE ARTICLES

– My contribution amounted to ca. 60%. I, along with the co-author developed the concept of the paper, the experimental design and reviewed the literature; along with co-author conducted the experiment; collected the observations and analyzed the data; along with the co-authors drafted, reviewed, edited the manuscript.

Paper 7 published as: Y. Chawla, G. Chodak (2018), Recommendations for • social media activities to positively inﬂuence the economic factors., In: Double- blind peer-reviewed proceedings part I. of the international scientiﬁc conference Hradec Economic Days 2018,January 30-31, 2018, Hradec Králové / [ed. by Pavel Jedlicka,ˇ Petra Marešová, Ivan Soukal]. Hradec Králové : University of Hradec Králové, 2018. s. 328-338. (Hradec Economic Days, ISSN 2464-6059; vol. 8 (1)).

– Indexed in Web of Science. – My contribution amounted to ca. 80%. I developed the concept of the paper; carried out the literature review; analysed observations from social media and drew out the recommendations; drafted, reviewed, edited and revised the manuscript.

3.3 Auxiliary results

During the course of my doctoral studies, I have published three more papers, have one paper under review after revisions and currently advancing the work carried out under this thesis with more experiments in other countries. The published articles listed under this section are not an integral part of this thesis. Nevertheless, for completeness of this section, I brieﬂy summarize the main results obtained in the following papers:

Y. Chawla, G. Chodak, K. Ludwikowska (2019a), Importance and Recommen- • dations for Trainers’ Use of Online Social Media as A Soft Skill to Positively Inﬂuence Trainees and Peers. In proceedings: 34th International Business Information Management Associa- tion Conference (IBIMA), pp: 2477-2487. Conference paper: CORE B, MNiSW 70p. Y. Chawla (2019) Education for Sustainable Development and Careers with the • use of Social Media. In proceedings: 34th International Business Information Management Associa- tion Conference (IBIMA), pp: 5675-5684. Conference paper: CORE B, MNiSW 70p. Y. Chawla (2020) Social media presence of managers at private universities: a • case study from India, In Book: Capacity building in higher education institutions, pp: 28-49. Book Chapter: MNiSW 20p.

In Chawla et al. (2019a), we highlighted that, in the 21st century, it is very important for the trainers and teachers to have a strong social media presence. Presence has the potential to influence students / trainees, which would create more engagement between them and also enhance the training transfer. Based on the findings from the 3.3. AUXILIARY RESULTS 21 literature regarding effective marketing habits and good practices of social media, we created a guideline, which can be used by trainers and teachers, to increase their effectiveness on social media. The guidelines provide direction but leave ample scope for the teacher or trainer to integrate their own personality traits on social media. This ensures that their virtual personality is a true reflection of themselves. In Chawla (2019), I discussed how social media can be effective in engaging students towards education for sustainable development and, at the same time, have a positive influence on their careers. The work was a part of my research on developing a change project for the “Baltic University Programme (BUP) Teachers Course on Edu- cation for Sustainable Development”, organized from September 2018 to March 2019. It addresses the increasingly complex problem the world is facing, which requires innovative and sustainable solutions. Communication and networking among stakeholders around the world would facilitate the development, implementation and scalability of such innovative and sustainable solutions. Increasing popularity and global coverage of social media makes it a vital role-player in the process of communication and networking for education in sustainable development and students’ career prospects. I also conducted a pilot study of the implementation of this change project among students of engineering and business administration at an Institution in India. Results show that there is an increased level of awareness regarding Sustainable Development Goals, stronger communication skills on virtual platforms, increased ability to use So- cial Media tools for gaining precise and required information, increased ability to prepare and outline the skills and knowledge required for employment in that career path and an increased ability to discuss sustainability issues with peers around the world. In Chawla (2020), I discussed the outcomes and importance of effective social media usage by managers at private universities, through a literature review and case studies from India. In this information age, society is being shaped based on knowledge and shadowed by a high-tech global economy. Higher educational institutions are expected to ensure the flow of skilled and knowledgeable human resources into the market. India’s higher education system is the third largest in the world, next to United States and China. The role of the manager in this system is undoubtedly crucial and, in this information age, the role of their social media presence is equally crucial. I established a connection between the social media presence of universities and that of their managers by analyzing documents, relevant scientific literature and monitoring online social media presence. I also conducted two field experiments with the managers at two private universities in India to check their willingness to participate in social media activities. Through the analysis, I found that managers are not willing participants when it comes to increasing their social media presence. In addition to these publications, there are several works in progress, which are expected to be completed in the coming months. We have collected data for the survey regarding SM from two other countries, India and Brazil, which will now be analyzed. Currently, we are in the process of collecting responses from 5 other countries including, Australia, Spain, Germany, Russia and Malaysia, which would give us further insight into this area. Three other works are also in progress. The first being: "Opinion of students and young professionals towards a model for social media management to enhance career opportunities", for which I have already collected the data and plan to analyze after the completion of this thesis. The PPS (personal, professional and social) model for social media management, was presented as a poster in the 6th European Conference on Social Media, where it was awarded the first prize for best poster. The 22 CHAPTER 3. SUMMARY OF RESULTS AND CORE ARTICLES second is a book chapter for a Springer book series, "Handboook on Climate Change" and my chapter is titled, "Use of Social Media by climate change organizations for public relations". The third is a case study about waste management and the concept of a circular economy. During the course of my PhD, I have reviewed three international conference papers and one journal article for the World Electric Vehicle Journal, a MDPI journal. Chapter 4

Conclusions

The main aim of this thesis was to highlight a social media management plan, which can enhance consumer awareness and acceptance of smart meters. This was achieved by fulﬁlling the following four objectives:

RO1: To investigate the attitudes, preferences and fears, regarding aware- • ness, willingness and acceptance of SM, among social media users. RO2: To explore the various sources of information regarding electricity, in • general, and SM, in particular. RO3: To test the effectiveness of different types of content on social media • and device metrics, through which managers can interpret the results of their campaigns. RO4: To create a social media management plan that would be useful for • energy companies to enhance the diffusion of SM.

Although, there were previous studies in the literature regarding consumer awareness and acceptance of SM, to the best of our knowledge, this was the very first conducted among social media users. Social media management strategies have been discussed by several researchers with different perspectives, however, this thesis is the very first which discussed a detailed social media plan for enhancing the diffusion of SM in countries where there is a low diffusion of SM. The interdisciplinary nature of the thesis, combining two sub-disciplines of Management Science, Innovation Management and Marketing Management, from the theoretical and managerial aspect, opens new horizons for similar research with other innovative products in energy markets, as well as other sectors. Social media management outlined in this thesis, with its various parts described and performed in Papers 1-7 (see Appendix A), provides a comprehensive framework, which can be utilized as a whole and also in parts, depending on the company’s goals. In parts, it can seen as determining consumers’ willingness to accept SM, consumers’ awareness regarding SM, communication channels & marketing content for promoting SM and the effectiveness of social media marketing for businesses. As a whole, they fuse together to fulfill the main aim of this thesis. Some of the key findings of our thesis includes: (i) There being a low awareness and acceptance of SM among consumers in Poland, Portugal, Indonesia and Turkey; (ii) Increases in knowledge or awareness positively impacts acceptance and reduces fears; (ii) Privacy concerns, regarding SM, are lower among social media users; (iii) Consumers, previously having invested in energy saving, or energy saving devices and having other

23 24 CHAPTER 4. CONCLUSIONS smart devices at home (which can connect to internet) are more likely to accept SM; (iv) Less than half the respondents, possessing knowledge of SM, were willing to pay for a SM; (v) Even though the study was conducted among social media users, they preferred using a variety of communication channels to get information regarding SM; (vi) There is scope for energy companies to use social media to enhance the diffusion of SM; (vii) The insights of post performance given by Facebook can be misleading, hence, performance should be judged through the proposed metrics; (viii) Posts with videos were found to be more effective, as compared to images or photo albums on Facebook; (ix) Posts, which have text in the caption and web link as a comment, were more effective, as compared to posts where the link was in the caption. Due to the novel results obtained in this thesis, we already see several future research horizons. First of all, it would be interesting to investigate the recommended plan by collaborating with energy companies. This could help in further optimizing the plan outlined in this thesis. Within the scope of this thesis, we concentrated only on social media channels, but it would be interesting to check its broader application with other digital and non-digital marketing channels. The intended scope of application for this thesis is for the countries at early stages of SM diffusion, which prompts the scope to check if similar methodology can be used to enhance the engagement of consumers in countries with a high diffusion of SM. To summarize, in this thesis, we devised a social media management plan, which can used by energy companies to enhance the diffusion of SM. The novel plan, derived in the thesis fill the identified gap in the management science literature as well as opens new horizons for further research. The flexible nature of the plan, and its non- geographical bounding scope, warrants that it can be implemented by energy companies in various countries. This is the case because the plan takes into account specific target consumers and also the business environment where the energy company is located. Hence, the managerial implications of the outcome in this thesis a has wide potential. Acknowledgements

I have been fortunate to receive blessings, mentoring and support from teachers, parents, friends and colleagues over the years. First and foremost, I would like to express my deepest love and gratitude towards my parents who have always encouraged me to go the extra mile and gave me the independence to make difficult decisions. Moving from India to Poland and leaving a settled life was a hard choice to make. It would not have been possible without the due support of my parents, friends, Dr. Kamila Ludwikowska - my co-supervisor for this thesis, Dr. Devanshu Patel (President - Parul University, India), Mr. Denish Patel (Executive Vice President, RK University, In- dia, Mr. Mohit Patel (Vice President, RK University India) and Mr. Prabhjeet Singh (General Manager - Glinks International, Dubai). I would like to express my sincere gratitude to them. It was only because of the efforts and guidance of Dr. Ludwikowska, that i was able to find my scientific supervisor and carry out the enrollment process at the university. Her support and encouragement throughout my doctoral studies has been imperative in achieving my goals. I would like to pay special regards and gratitude to my supervisor, Prof. Grzegorz Chodak, who always went above and beyond to support and guide me throughout the course of my doctoral studies. Be it lessons in research or life lessons of humility and leadership, his wisdom always enlightened me. The flexibility and freedom he granted me, has made a huge difference for me to find the niche, which best suited my interests and skills. He convincingly guided and encouraged me to be professional and do the right thing, even when the road got tough. Without his persistent help, the goal of this thesis would not have been realized. It was also because of his efforts, that I was able to apply and secure a teaching & research position at the Department of Operations Research and Business Intelligence, Faculty of Computer Science and Management, Wrocław University of Science and Technology, Poland. I am greatly indebted to him and feel highly fortunate to have had him as my scientific supervisor. The research in this thesis would also not have been possible without the support and guidance of Prof. Anna Kowalska-Pyzalska. It was with her expertise in innovation management and Prof. Chodak’s expertise in marketing management, that the objectives of this thesis were able to be realized. Analyzing the data and writing manuscripts for impact factor journals, are two important things that Prof. Kowalska- Pyzalska taught me. This helped me in fulfilling the objectives of my research as well as would benefit me further in my scientific career. I feel privileged to have had the opportunity to collaborate with her and learn from her. I would like to express my deepest gratitude to Prof. Rafał Weron, who has always challenged me to do more and at the same time guided as well as supported to fulfill those challenges. He also enabled me to participated in various international conferences and workshops, through which i have got ample exposure to the scientific community. I also want to Dr. Katarzyna Maciejowska, who has always guided me

25 26 CHAPTER 4. CONCLUSIONS in data analysis as well as didactics. Her constant motivation and encouragement has always been a driving force for me. I would also like to acknowledge the support of Ms. Wieslawa Darska, who always made extra efforts to ensure that my ofﬁcial documentation in Polish is taken care of. I also thank, the Dean, teachers, researchers, administrators and staff members at the Faculty of Computer Science and Management, who have been supportive throughout the course of my doctoral studies. Last but not the least, I would like to acknowledge and thank all my teachers and mentors, who have inculcated the knowledge and values that I possess. Bibliography

G. Abeza, N. O’Reilly, and I. Reid. Relationship marketing and social media in sport. International Journal of Sport Communication, 6(2):120–142, 2013. Accenture. The new energy consumer: Unleashing business value in a digital world, 2015. URL https://www.accenture.com/_acnmedia/accenture/ next-gen/insight-unlocking-value-of-digital-consumer/ pdf/accenture-new-energy-consumer-final.pdf. Last Accessed: 13 March 2020. AOM. Divisions & interest groups of the academy of management, 2010. URL http: //aom.org/DIG/. Last Accessed: 24 February 2020. D. Barrios-O’Neill and G. Schuitema. Online engagement for sustainable energy projects: A systematic review and framework for integration. Renewable and Sus- tainable Energy Reviews, 54:1611–1621, 2016. M. Bogdal. Communication management in public sectors: the case of the Polish energy sector. Comunicacao Publica, 8:7–23, 2013. B. Boland. Organic reach on facebook: Your questions answered, 2014. URL https://www.facebook.com/business/news/ Organic-Reach-on-Facebook. Last Accessed: 10 March 2020. N. Booth and J. A. Matic. Mapping and leveraging influencers in social media to shape corporate brand perceptions. Corporate Communications: An International Journal, 16(3):184–191, 2011. D. Boutakidis, S. Aggelopoulos, A. Pavloudi, K. Soutsas, et al. Attitudes and opinions of social media users on renewable energy. Journal of Environmental Protection and Ecology, 15(4):1727–1734, 2014. British Infrastructure Group of Parliamentarians. Not so smart – a comprehensive investigation into the roll-out of energy smart meters. Technical report, British In- frastructure Group of Parliamentarians, July 2018. Y. Chawla. Education for sustainable development and careers with the use of social media. In K. S. Soliman, editor, Proceedings of the 34th International Business Information Management Association Conference (IBIMA), pages 5675–5684. In- ternational Business Information Management Association, Nov. 2019. Y. Chawla. Social media presence of managers at private universities: a case study from india. In K. Ludwikowska, A. Bienkowska,´ and K. Tworek, editors, Capacity building in higher education institutions, chapter 2, pages 28–49. Difin, Warsaw, Poland, 2020. Y. Chawla and A. Kowalska-Pyzalska. Public awareness and consumer acceptance of smart meters among polish social media users. Energies, 12(14):2759, 2019. Y. Chawla, G. Chodak, and K. Ludwikowska. Importance and recommendations for trainers’ use of online social media as a soft skill to positively influence trainees and peers. In K. S. Soliman, editor, Proceedings of the 34th International Busi-

27 28 BIBLIOGRAPHY

ness Information Management Association Conference (IBIMA), pages 2477–2487. International Business Information Management Association, 2019a. Y. Chawla, A. Kowalska-Pyzalska, and P. D. Silveira. Marketing and communications channels for diffusion of smart meters in Portugal. HSC Research Re- ports HSC/19/05, Hugo Steinhaus Center, Wroclaw University of Technology, Dec. 2019b. URL https://ideas.repec.org/p/wuu/wpaper/hsc1905. html. Y. Chawla, A. Kowalska-Pyzalska, and W. Widayat. Consumer willingness and acceptance of smart meters in indonesia. Resources, 8(4):177, 2019c. Y. Chawla, A. Kowalska-Pyzalska, and B. Oralhan. Attitudes and opinions of social media users towards smart meters’ rollout in turkey. Energies, 13(3):732, 2020. J.-S. Chou and I. G. A. N. Yutami. Smart meter adoption and deployment strategy for residential buildings in indonesia. Applied Energy, 128:336–349, 2014. M. C. Claudy, C. Michelsen, A. O’Driscoll, and M. R. Mullen. Consumer awareness in the adoption of microgeneration technologies: an empirical investigation in the republic of ireland. Renewable and Sustainable Energy Reviews, 14(7):2154–2160, 2010. P. Cooper. How the facebook algorithm works in 2020 and how to make it work for you, 2020. URL https://blog.hootsuite.com/ facebook-algorithm/. Last Accessed: 3 March 2020. S. Cyfert, W. Dyduch, D. Latusek-Jurczak, J. Niemczyk, and A. Sopinska.´ Subdyscy- pliny w naukach o zarz ˛adzaniu–logikawyodr˛ebnienia,identyﬁkacja modelu kon- cepcyjnego oraz zawartos´c´ tematyczna. Organization and Management, 161, 2014. P. J. Danaher and J. R. Rossiter. Comparing perceptions of marketing communication channels. European Journal of Marketing, 2011. C. Droge, M. Stanko, and W. Pollitte. Lead users and early adopters on the web: the role of new technology product blogs. Journal of Product Innovation Management, 27:66–82, 2010. O. Ellabban and H. Abu-Rub. Smart grid customers’ acceptance and engagement: an overview. Renewable and Sustainable Energy Reviews, 65:1285–1298, 2016. EURAM. Strategic interest groups, 2009. URL http://www.euram-online. org/community/strategic-interest-groups.html. Last Accessed: 24 February 2020. A. Faruqui and S. Sergici. Household response to dynamic pricing of electricity: a survey of 15 experiments. Journal of regulatory Economics, 38(2):193–225, 2010. M. Fiore, R. Silvestri, F. Contò, and G. Pellegrini. Understanding the relationship between green approach and marketing innovations tools in the wine sector. Journal of Cleaner Production, 142:4085–4091, 2017. A. Ghazvini, S. Ramos, J. Soares, R. Castro, and Z. Vale. Liberalization and customer behavior in the Portuguese residential retail electricity market. Utilities Policy, 59: 100919, 2019. T. Gong, D. Li, Y. Liu, G. Wang, and H. Zhu. Analysis of marketing strategy of electricity selling companies in the new situation. Journal of Physics: Conference Series, 1187:022043, 2019. T. Gouws and G. P. Van Rheede van Oudtshoorn. Correlation between brand longevity and the diffusion of innovations theory. Journal of Public Affairs, 11(4):236–242, 2011. L. Guelman, M. Guillén, and A. M. Pérez-Marín. A decision support framework to BIBLIOGRAPHY 29

implement optimal personalized marketing interventions. Decision Support Systems, 72:24–32, 2015. S. Hille, S. Weber, and T. Brosch. Consumers’ preferences for electricity-saving programs: Evidence from a choice-based conjoint study. Journal of Cleaner Produc- tion, 220:800–815., 2019. Y. Kamarudin and T. Boothman. Powering the nation: Indonesian power industry survey, 2017. URL https://www.pwc.com/id/en/ energy-utilities-mining/assets/power/power-survey-2017. pdf. Last Accessed: 11 March 2020. S. Kemp. Digital 2019: Global internet use accelerates, 2019. URL https:// wearesocial.com/global-digital-report-2019. Last Accessed: 19 February 2020. S. Kemp. Digital 2020: 3.8 billion people use social media, 2020. URL https://wearesocial.com/blog/2020/01/ digital-2020-3-8-billion-people-use-social-media. Last Accessed: 18 February 2020. S. Khamis, L. Ang, and R. Welling. Self-branding,‘micro-celebrity’and the rise of social media inﬂuencers. Celebrity studies, 8(2):191–208, 2017. M. Z. Khan, A. K. Miankhel, and A. Nawaz. Information & communication technology and ‘individual’: Prospects & concerns. Global Journal of Computer Science and Technology, 2013. A. Kowalska-Pyzalska and K. Byrka. Determinants of the willingness to energy monitoring by residential consumers: A case study in the city of wroclaw in poland. Energies, 12(5):907, 2019. J. Kowalski and B. E. Matusiak. End users’ motivations as a key for the adoption of the home energy management system. International Journal of Management and Economics, 55(1):13–24, 2019. T. Krishnamurti, D. Schwartz, A. Davis, B. Fischhoff, W. B. de Bruin, L. Lave, and J. Wang. Preparing for smart grid technologies: A behavioral decision research approach to understanding consumer expectations about smart meters. Energy Policy, 41:790–797, 2012. C. Lamberton and A. T. Stephen. A thematic exploration of digital, social media, and mobile marketing: Research evolution from 2000 to 2015 and an agenda for future inquiry. Journal of Marketing, 80(6):146–172, 2016. J. Lee and I. B. Hong. Predicting positive user responses to social media advertising: The roles of emotional appeal, informativeness, and creativity. International Journal of Information Management, 36(3):360–373, 2016. J. H. Lipschultz. Social media communication: Concepts, practices, data, law and ethics. Routledge, 2017. L. Ma, C. S. Lee, and D. H.-L. Goh. Understanding news sharing in social media. Online Information Review, 2014. G. Nakarado. A marketing orientation is the key to a sustainable energy future. Energy Policy, 24:187–193, 1996. E. M. Payne, J. W. Peltier, and V. A. Barger. Omni-channel marketing, integrated marketing communications and consumer engagement. Journal of Research in In- teractive Marketing, 2017. K. Perindustrian. Making Indonesia 4.0. Ministry of Industry, Republic of Indonesia, 2018. 30 BIBLIOGRAPHY

K. Peters, Y. Chen, A. M. Kaplan, B. Ognibeni, and K. Pauwels. Social media metrics—a framework and guidelines for managing social media. Journal of interactive marketing, 27(4):281–298, 2013. W. Pongpaew, M. Speece, and L. Tiangsoongnern. Social presence and customer brand engagement on facebook brand pages. Journal of Product & Brand Management, 26(3):262–281, 2017. M. Rocha. Data privacy and social acceptance of smart meters. Smart Grid Handbook, pages 1–9, 2016. E. M. Rogers. Diffusion of Innovations. The Free Press, New York, USA, 1962. E. M. Rogers. Diffusion of innovations. free press. New York, 551, 2003. B. Soroczynski´ and J. Szkutnik. New possibilities in electrical energy efficiency. In 2015 16th International Scientific Conference on Electric Power Engineering (EPE), pages 13–16. IEEE, 2015. B. Stauss and W. Seidel. Social media complaints. In Effective Complaint Manage- ment, pages 451–468. Springer, 2019. D. Streimikiene and J. Vveinhardt. Community based social marketing for implementation of energy saving targets at local level. Amfiteatru Economic Journal, 17: 723–734, 2015. S. Sudoł. Zarz ˛adzaniejako dyscyplina naukowa. Przegl ˛adOrganizacji, 4:4–11, 2016. A. Theron. Eastern europe to meet eu smart meter targets, says report, 2015. URL https://www.smart-energy.com/regional-news/europe-uk/ eastern-europe-to-meet-eu-smart-meter-targets-says-report/. Last Accessed: 8 March 2020. A. Theron. Turkey to install 3.6m smart meters annually until 2020, 2017. URL https://www.smart-energy.com/regional-news/ africa-middle-east/turkey-smart-meters/. Last Accessed: 19 February 2020. J. Tidd and B. Thuriaux-Alemán. Innovation management practices: cross-sectorial adoption, variation, and effectiveness. R&D Management, 46(S3):1024–1043, 2016. C. Treadaway and M. Smith. Facebook marketing: An hour a day. John Wiley & Sons, 2012. J. Unold. Basic aspects of the digital economy. Acta universitatis lodziensis folia oeconomica, 167, 2003. E. van der Werff and L. Steg. The psychology of participation and interest in smart energy systems: Comparing the value-belief-norm theory and the value-identity- personal norm model. Energy research & social science, 22:107–114, 2016. R. Van Gerwen, S. Jaarsma, and R. Wilhite. Smart metering. Leonardo-energy. org, 9, 2006. G. P. Verbong, S. Beemsterboer, and F. Sengers. Smart grids or smart users? involving users in developing a low carbon electricity economy. Energy Policy, 52:117–125, 2013. F. E. Webster Jr. A perspective on the evolution of marketing management. Journal of Public Policy & Marketing, 24(1):121–126, 2005. T. Weron, A. Kowalska-Pyzalska, and R. Weron. The role of educational trainings in the diffusion of smart metering platforms: An agent-based modeling approach. Physica A: Statistical Mechanics and its Applications, 505:591–600, 2018. J. Wielki. The impact of the internet on the development of web-based business models. Journal of Internet Banking and Commerce, 15(3), 2010. BIBLIOGRAPHY 31

Q. Xu, J. Chen, Z. Xie, J. Liu, G. Zheng, and Y. Wang. Total innovation management: a novel paradigm of innovation management in the 21st century. The Journal of Technology Transfer, 32(1-2):9–25, 2007. F. Zhou, M. K. Lim, Y. He, and S. Pratap. What attracts vehicle consumers’ buying. Industrial Management & Data Systems, 2019. J. Zote. 55 critical social media statistics to fuel your 2020 strategy, 2020. URL https://sproutsocial.com/insights/ social-media-statistics/. Last Accessed: 10 March 2020. 32 BIBLIOGRAPHY Appendix A: Papers 1-7 34 BIBLIOGRAPHY Paper 1

Y. Chawla, A. Kowalska-Pyzalska, B. Oralhan Attitudes and Opinions of Social Media Users Towards Smart Meters’ Rollout in Turkey energies

Article Attitudes and Opinions of Social Media Users Towards Smart Meters’ Rollout in Turkey

Yash Chawla 1,*,† , Anna Kowalska-Pyzalska 1,† and Burcu Oralhan 2 1 Department of Operations Research and Business Intelligence, Faculty of Computer Science and Management, Wroclaw University of Science and Technology, 50-370 Wroclaw, Poland; [email protected] 2 Faculty of Economics and Administrative Sciences, Nuh Naci Yazgan University, Erkilet Dere Mah., 38170 Kocasinan, Turkey; [email protected] * Correspondence: [email protected]; Tel.: +48-69-329-0935 † Current address: Wyb. Wyspianskiego 27, 50-370 Wroclaw, Poland.

Received: 12 January 2020; Accepted: 4 February 2020 ; Published: 7 February 2020

Abstract: Increasing the efficiency of electricity transmission is nearing the top of the agenda in many countries around the world. Turkey, the world’s most newly industrialized country, is no different. Modernizing the current transmission grids to smart grids (SG) and the national rollout of smart meters (SM), are some of the measures taken by the government to meet the growing demand for electricity. Consumer acceptance and engagement are among the most important elements for the success of SG and SM, however, there have not been much studies done among Turkish electricity consumers. This purpose of this study is to fill this void, by detailing the attitudes, awareness and expectations among Turkish citizens regarding SM and listing recommendations for energy companies based on the findings. Through an online questionnaire, responses from 504 social media users were collected and analyzed. Results show that the consumers are open towards the acceptance of SM, but there is a need to raise awareness and knowledge through proper communication channels. The study has also revealed that a range of conventional and digital channels need to be actively used in order to enhance consumer willingness to accept SM. Increasing social interactions regarding SM is one of the key recommendations detailed by the authors.

Keywords: smart meters; consumer knowledge; Turkish electricity market; social acceptance; social media

1. Introduction Nowadays, in most countries, a great effort is being made to increase the efficiency of electricity generation, transmission and consumption. Governments have proposed some national strategies, legislation and described various ways to achieve the ambitious goals of CO2 reduction, increasing green energy and energy efficiency. These actions are motivated, not only by political (fulfillment of obligations e.g., Paris Climate Agreement) or economic incentives (i.e., lower costs), but also by environmental and social ones. Because of the negative impact of energy intensive industries and energy consumption on climate change, new technologies and a higher level of awareness among people is needed. The achievement of such goals starts with the national exchange programs of traditional electricity meters into smart ones, allowing for better communication between producers, distributors, sellers and consumers [1–4]. Thanks to smart meters (SM), access to real-time data regarding the electricity consumption is possible [5,6]. From the consumers point of view, access to such information, if combined with varied electricity prices (i.e., different consumption dependent on the time of the

Energies 2020, 13, 732; doi:10.3390/en13030732 www.mdpi.com/journal/energies Energies 2020, 13, 732 2 of 27 day, or even real-time prices), may be very valuable and may lead to energy reduction and money savings [7–11]. SM deployment may also bring significant advantages to energy suppliers, by eliminating manual monthly meter readings and enabling the monitoring of the power system in real time. It is also a step towards demand-side management, by introducing dynamic pricing, encouraging more efficient use of the electricity and providing responsive data for balancing electric loads in order to reduce blackouts. Exchange of current meters to SM, in the power system, may also allow the ability to avoid the capital expense of building new power plants by optimizing the usage of the existing resources [12,13]. Turkey observes a growing demand for electricity over the last decade. This is because of increased consumption in both residential as well as industrial sectors. Turkey, which is experiencing rapid industrial growth, is among the countries in the world which are leading producers of agriculture-related products, textiles, automotives, transportation machinery, building materials, home appliances and electronics. Hence, the demand for electricity is bound to increase further. To satisfy the rising demand, Turkish energy companies must ensure the appropriate level of supply and take care of the energy efficiency. Recently, Turkey’s largest electricity distribution and retail group—CLK Enerji—announced plans to replace most electricity meters in Turkey with SM. The SM rollout will start with covering four distribution networks in 11 provinces. During this rollout, different SM and communication infrastructure will be tested (https://www.dnvgl.com/cases/exploring-smart- metering-in-turkey-85585 (accessed 6 November 2019)). According to the transition plans, prepared by Energy Market Regulatory Authority (EMRA) and Association of Distribution System Operators (ELDER), Turkey would be replacing 80% of the current electricity meters with SM by 2035. The primary aim of this transition is to improve the efficient usage of electricity along with reduction in losses and power cuts (for more details see: https://balkangreenenergynews.com/turkey-sets-roadmap-smart- grids-plan-worth-over-4-billion/ (accessed 6 November 2019)). As per the roadmap presented in the Turkey Smart Grid 2023 strategy, individual and region specific requirements for the SM upgrade have to be determined by each of the distribution system operators by 2020. Based on individual requirements, the appropriate technology will be applied in each case. The roadmap also takes into account a broader integration of small-scale renewable energy generators within the power system. Because of the Turkish rollout of SM, there is a great need to investigate the consumers’ attitudes and opinions towards this transition. As the literature has proven many times, consumers’ acceptance is needed to enhance smooth diffusion of any technology innovation [14–16]. Smart meters themselves are not user-friendly, but if combined with smart metering information systems (platforms, SMP), such as internet widgets or mobile apps, may share the information about electricity consumption (and prices) automatically in real-time [17,18]. The access to this information may lead further to some behavioral changes connected with energy conservation. Within our paper, we plan to explore the current attitudes and expectations towards smart meters of the residential consumers in Turkey. In particular, we focus on social media users, who, based on the up-to-date findings from the literature [19] are a social group, which is perceived to be more open-minded towards innovations and IT-based solutions than the rest of society. We focus on two following research questions: (1) how does the knowledge about SM relate to the consumers’ attributes (such as age, education or income), their preferences and fears and willingness to accept/install SM; (2) what are the communication channels used by consumers in order to learn about the energy market and smart meters. As the literature proves, consumer acceptance and willingness to engage with the technology has the same level of importance as technological advances leading to increase in energy efficiency, if not more[ 14,20]. As far as we are aware from studying the up-to-date literature, the attitudes of Turkish citizens towards SM rollouts have not been extensively investigated. Within this study, we aim to fill in this void. Our paper contributes to the existing literature, not only by adding some valuable information about Turkish SM rollouts, but also by shedding some light on the general determinants of consumers’ knowledge about SM and the usage of communication channels between energy companies Energies 2020, 13, 732 3 of 27 and residential consumers. The article is structured in the following sections: Section2 details the crucial findings regarding consumers attitudes and opinions towards SM in the light of the up-to-date studies and research. Section3 details the Turkish transition of the traditional power grid into a smart one, with an emphasis on the role of SM and advanced metering infrastructure (AMI). Section4 shows the research framework and the methodological background of the study and describes the results. In Section5, the conclusions and recommendations are presented. Thereafter in Section6 we describe the limitations of the study and the new horizons it opens for further research.

2. Consumers’ Attitudes towards Smart Grids and Smart Metering: Literature Review

2.1. Technology Acceptance Among models investigating and explaining the circumstances and the process of technology acceptance, the most popular ones are: unified theory of acceptance and use of technology (UTAUT); technology acceptance model (TAM); value-based adoption model (VAM); and theory of planned behavior (TPB) [21,22]. The comparative analysis of all four models, presented in the work of [21] has shown that for products that are innovative but have minimal practical value (e.g., smart products based on artificial intelligence solutions, to which SM also belongs), the technology acceptance highly depends on consumers’ level of interest in technology than in its functional aspects. Dependent on the model, various issues of the decision making process are emphasized. For example, in TAM, the perceived usefulness of the innovation, simplicity in usage and the behavioral intention to use are the major components of the model. The TPB model includes additional attitudes towards the innovation and the subjective norms. The UTAUT and VAM models compare the cost and benefits connected with the innovation adoption (i.e., effort and performance expectancy and enjoyment versus perceived fee) [21]. In exploring consumer acceptance and engagement towards innovative technologies, products and services, Roger’s model of innovation diffusion (DoI) is often taken into account [23–28]. In this model, four main elements are integrated: the innovation itself, the social network, in which innovation spreads, as well as the time and communication channels. According to the DoI model, the rate of consumers adoption, that is the number of individuals who have started using a particular innovation in a certain time period, depends on: the innovation’s complexity or simplicity, how much can an advancement be experienced with and seen by the others (e.g., neighbors or friends) and compatibility with one’s values, needs and past experiences [27,29]. Studies in the literature, on acceptance of smart grid technology, including smart metering and SMP, can be classified into two groups. The first group investigates the main factors and attributes responsible for the smart grids (SG) and SM technology acceptance (see e.g., [1,30]) and the second explores the relations between those factors (see e.g., [15,31]). Among the factors explaining the SG and SM technology acceptance, the literature mentions, apart from financial benefit, eco-friendliness and cyber and privacy security, as well as an understanding of the technology. The researchers emphasize that raising the awareness of consumers and improving the level of information they have on the economic and financial effectiveness of SG would play a major role in the technology acceptance process[22,32].

2.2. Smart Metering Acceptance The transition of the traditional power systems into smart ones will not be possible and effective without the development of smart markets and smart customers [33]. Smart markets include, first of all, advanced metering infrastructure (AMI), metering information management systems, demand management tools and energy trading. Because of those new components of the power system, consumers may now actively engage in the energy market by providing vital inputs from the demand side to balance the grid[ 34]. In particular, smart meters, combined with enabling technologies, enables the monitoring of energy consumption on a real time basis. In that way, consumers may have a Energies 2020, 13, 732 4 of 27 positive impact on increasing energy efficiency (i.e., by lowering energy consumption or shifting it from on-peak into off-peak hours) [17]. The role of the electricity customer in the transition of the existing grids into the smart grid is vital. Smart meters installed at each household enable access to the real-time information on consumption for the energy companies as well as for the consumers. The latter may get access to such data mainly through smart metering platforms (SMP), such as web-based interfaces and smartphone applications often combined with in-home displays or intra-networks [17,18,34]. Recently, a lot of studies have been carried out to investigate the social acceptance and awareness of smart grids and smart metering. For consumers, the awareness about the opportunities they get, and the knowledge on how to use SMP, is fundamental. Without any awareness, knowledge and, finally, acceptance of smart metering together with SMP, this novelty will never spread successfully in the market [11,14,17,35], leading to an insufficient increase of energy efficiency. While examining the diffusion of SM, various aspects of its acceptance have been explored, as presented in Table1. Generally, studies have revealed that the awareness, interest and knowledge about SM, among residential consumers, is limited all over the world [11,16,34,36,37]. Energy is an abstract commodity, so people do not commonly talk about it [38]. Consumers also have concerns and fears regarding the privacy and security of data provided by SM, installation visits, adverse health consequences, portability in terms of changing suppliers and disconnection of meters on a prepayment basis [1,8,18,34,39]. As the literature indicates, most of the consumers’ fears rise from the myths and misunderstanding of how SM works. Again, the lack of awareness and knowledge continues to be the main obstacle to the smooth diffusion of SM among residential consumers. Another aspect of the studies have also shown that rising knowledge about the benefits and options given by SM and SMP may lead to positive behavioral changes (i.e., lower energy consumption or lowering consumption during on-peak hours by shifting it to off-peak hours). The potential of SM is especially emphasized if energy companies provide additional tools for demand-side management and demand response (DSM/DR), such as dynamic electricity tariffs. In that case, feedback received via SM and SMP, regarding electricity consumption as well as prices, may have a great impact on the increase of energy efficiency [10,40]. On the other hand, there is uncertainty regarding persistence of the consumers’ engagement, where it would sustain over a longer period of time or will disappear due to the discouragement and lack of motivation [17,34,41]. Finally, although there is a large portion of studies investigating the effectiveness of feedback provided by SM and SMP (see for example [9,10,38,40,41]), there are very few papers exploring the role of communication channels (traditional versus modern ones) in terms of SM diffusion (e.g., [42,43]. Within our paper we want to fill this gap. Energies 2020, 13, 732 5 of 27

Table 1. Literature review: research aspects of the consumers’ acceptance and engagement towards smart meters (SM) and smart metering platforms (SMP).

Issue Investigated Citation Summary of Findings Consumers’ acceptance [1,15,30,31, The acceptance starts with some level of awareness and knowledge. and engagement 34,44–46] So far, many of the consumers are unfamiliar with the terms SG and SM. Consumers express interest in conserving energy, but usually do not know how to achieve this goal. Consumers are willing to accept SM if it enables them to save energy (and money), has no negative effect on one’s health and does not increase privacy concerns and fears. The privacy concern due to access to the private information about one’s usage of electricity and hence, presence at home, and the health concern, as a result of increased electromagnetic exposure or wireless smart meters, belong to the most common disadvantages of SM. Willingness to pay for [11,34,36,47] Consumers are interested in SM especially if they do not need to pay for SM/willingness to install implementation. They also expect receiving an access to some supporting SM technologies (such as in-home displays or SMP) in order to optimize energy usage. They appreciate high levels of automatic adjustment of energy consumption of their home appliances, as in most cases they do not want to pay much attention to current electricity prices and hence adjust their consumption behaviors and habits. Incentives and barriers to [1,16,31,32, To the most common barriers belong: consumers decision making based adoption 34,39,48] on limited information, uncertainty of choice, lack of knowledge and understanding leading to negative perceptions and beliefs, negative word of mouth, discomfort of usage (e.g., change of habits) and privacy and security concerns. Among most popular incentives there are the willingness to protect the environment, to save energy and money and to adjust to social norms. Effectiveness of feedback [7,9,10,17,18, In most cases only consumers who are predominately aware of what SM provided by SM 40,41,49] is, are interested in receiving feedback about their energy consumption. It is still not clear how to attract consumers’ attention and engagement in a longer time period, especially if the achieved savings are not very impressive. The perceived possibility to monitor energy via SM and SMP has a great impact on adoption. The role of knowledge, [50,51] Knowledge and awareness of SM and SMP are mainly influenced by: effectiveness of education advertising, social impact and education and training. In addition, and training a certain level of skills related to the usage of an internet platform or a mobile application is required to use SMP. To attract consumers to SMP, providers of those services should rather apply training and educational programs tailored to individual groups of consumers, rather than general education and marketing campaigns. The manipulation of the time that the consumer needs to make a decision (e.g., by means of promotion in a given time interval) affects the rate of diffusion. It is also crucial to maintain the appropriate level of knowledge and skills among consumers, acquired during training, for example by reminding them about SMP by means of text messages, e-mails or information brochures. Role of communication [18,32,42,43, Outreach and communication should try to increase familiarity and channels 52,53] demonstrate the financial and environmental benefits of SM and SG. At present both, the outreach and communication between energy providers and the consumers are insufficient to improve the understanding of SM and its effective usage. Social media nowadays seems to be a convenient medium for using social impact and spreading information about the innovative solutions, applications and services.

3. Smart Grids in Turkey

3.1. Turkish Electricity Market The Turkish electricity market, as described by Colak et al. [54], has followed typical mile steps in its development: starting with the unbundling of the generation, transmission and distribution in 1994, through the establishment of EMRA (Energy Market Regulatory Authority) and a few levels of market opening between 2004 and 2011. Energies 2020, 13, 732 6 of 27

Nowadays, in Turkey, the Ministry of Energy and Natural Resources and Republic of Turkey Energy Market Regulatory Authority (EPDK) are responsible for the energy policies. These government departments have completed the necessary studies and have been conducting various projects to adjust the existing power system and the current energy market infrastructures to smart grid concept [55].

3.2. Establishment of Smart Grids The beginning of the smart grid approach towards the Turkish power system began around ten years ago by means of a policy proposed by the Turkish Ministry of Energy in the so-called Natural Resources Strategic Plan (2010–2014). This strategy aimed to increase the share of renewable energy within the energy supply without stating in detail the role of smart grid in reaching this aim [54,56]. Shortly after 2014, Turkish distribution system operators (DSO) started some pilot projects aimed at advanced metering infrastructure (AMI), including smart meters (SM), but also DSM/DR tools. Currently, the government proposed a project called Turkey Smart Grid 2023 Vision and Strategy Roadmap (TSG2023), which is going to be implemented in the years 2016–2020. This roadmap is aimed at providing directions on the 2035 smart grid vision to the distribution companies, in short and medium terms, by pointing out the necessary priorities [33]. According to the TSG2023 roadmap, at least 80% of customers should be equipped with SM in the coming years (Read more in: Turkey Smart Grid 2023 Vision and Strategy Roadmap Summary Report, (2018). Republic Of Turkey Energy Market Regulatory Authority, Strategy Development Department Ankara, Turkey).

3.3. Smart Grid Pilot Projects Turkey has embarked on the development of smart system technology and some initiatives have already taken solid steps in this direction [57]. First of all, as already mentioned, there is a vision and implementation process starting with the rollout of smart meters. This rollout has been started by the Turkish Electricity Transmission Corporation (TEIA¸S)and˙ is then being continued within the distribution companies in Turkey (Akcanca, M. A., and Ta¸skın,S. (2011). Akıllı ¸sebeke uygulanabilirli˘giaçısından türkiye elektrik enerji sisteminin incelenmesi. Akıllı ¸Sebekelerve Türkiye Elektrik ¸SebekesininGelece˘giSempozyumu, 26–27. (in Turkish)). In TEIA¸S,the˙ Automatic Meter Reading System (OSOS) project has also been carried out for the remote automatic reading of the smart meters belonging to the users of the electricity transmission system. With the provisional acceptance of the project, at of the end of 2012, automatic data collection via OSOS was performed from approximately 2761 digital electricity meters installed in 948 different locations (TEIA¸S,Annual˙ Activity Reports, http: //www.teias.gov.tr/FaaliyetRaporlari.aspx, 13 October 2013 (accessed 5 November 2019)). Within this project, the real-time monitoring system was developed throughout the country. Real-time monitoring of all electrical magnitudes and power quality parameters of the electricity transmission system has been realized [58].

3.4. Attitudes of Turkish Citizens to SG The smart grid concept, together with the introduction of SM and SMP is still a great novelty in the Turkish energy market. Most consumers are not familiar with the Turkey Smart Grid and Vision roadmap for 2030. In the SWOT analysis of smart grid infrastructure in Turkey, Colak et al. [54] have emphasized that the lack of consumer awareness about the smart grid contributes to one of its most significant weaknesses for implementation. Although the opinions of Turkish residential consumers have not been investigated extensively, in the work of Tumbaz et al. [59], the attitudes and behaviors in the smart grid context have been explored. In particular, the authors have found that the energy consumption patterns of households may lead to the identification of potential electricity savings in the residential sector, such as standby consumption and potential electricity use, which can be shifted to off-peak hours. Moreover, the authors have driven some specific policy recommendations, which can promote behavioral change by measuring the responsiveness of people to different measures and the Energies 2020, 13, 732 7 of 27 combination of measures, such as information, feedback, rewards and social influences. The results obtained from this survey were used to depict a general view of Turkish households towards electricity consumption behaviors and their energy efficiency attitudes. They clearly indicate that there should be more regulations and improvements in energy policy [59] in order to increase the responsiveness level of Turkish electricity consumers. Taking all of that into consideration, by the means of our study, we seek to explore the current attitudes and expectations of Turkish citizens, regarding smart meters, even further. There are two research questions we seek to answer: (1) how does knowledge about SM relate to the consumers’ attributes (such as age, education or income), their preferences and fears and willingness to accept/install SM; (2) what are the communication channels used by the consumers in order to learn about the energy market, in general, and smart meters, in particular. As mentioned earlier, the examination of consumers’ knowledge and its determinants on SM acceptance is important. Knowing what is important for the consumers may then lead to a better presentation of the advantages and opportunities of SM by energy companies. Moreover, the investigation of communication channels, preferred by the consumers in the energy context, may enable better communication between energy companies and their residential customers.

4. Method, Results and Discussion

4.1. Methods The research and survey framework adapted the methods followed by previous research [23–26,34]. As mentioned in Section 2.1, there are a few models explaining the technology acceptance. In our survey, we focus mostly on the diffusion of innovation (DoI) model given by Roger in 2003 [27]. In this model, four main elements are integrated: the innovation itself, the social network (environment), in which innovation spreads, as well as the time and communication channels. In our approach, we focus on two elements of the model: the social network and the communication channels. According to Roger’s model, market participants can be divided into certain groups, dependent on their propensity to adopt innovative initiatives [27,28]. The first group that plays a vital role in diffusing the innovation is described as early adopters. In case of smart meters and the enabling technology combined with them (e.g., in-home displays, mobile apps or smart plugs), the early adopters are usually consumers who are more experienced with mobile apps and other smart devices. We believe that social media users—people who are active in their social networks and use mobile apps on a regular basis, are expected to be open-minded and eager to use the new equipment or applications [19,60]. That is why our study is focused on and conducted among social media users. Figure1, shows the research and survey frameworks and the variables we used. Variable choice has been inspired by the literature review as well as our research questions. In particular, similar to the other studies, we include the group of socio-demographic variables (D1–D10), preferences and willingness to adopt SM under different circumstances (P1–P4, De1–De8, G1–G3, X1), attitudes towards energy saving and environment (A2, A31–A39) and awareness and knowledge about SM (K1–K4). What is more rare in the context of smart metering, but common to the studies of innovation diffusion, we have also included the set of possession of smart devices and personal assets (B1–B7, R1), as well as behaviors towards buying a new technology (A1) and many questions regarding communication channels and sources of information in terms of the energy market in general and SM in particular (S01–S08, S1–S15, I1, I2, I31–I45, Q1, Q21–Q35). All the variables are listed and described in Table A1 (AppendixA). In total, the proposed questionnaire refers much to the Roger’s DoI model, by including social influence and communication channels in the questions asked. To collect the data from the respondents, the survey was conducted in the form of a self-administered online questionnaire which the respondents had to fill in three phases, as shown in the survey framework and variable part of Figure1. The questionnaire could be accessed in both the English and Turkish languages, for the convenience of the residents of Turkey. At first, a convenience Energies 2020, 13, 732 8 of 27 sampling method was used and the questionnaire was distributed through social media platforms such as Facebook, LinkedIn, Twitter, Facebook Messenger, WhatsApp and so on. E-mails were also sent by the author(s), through their connections. Thereafter the snowball sampling method was used, wherein the respondents who were contacted at first, were requested to disseminate the questionnaire to their social and professional network. In total, N = 504 responses were collected between 10 September 2019 and 10 October 2019, 502 in Turkish and only 2 in English. A total of 1415 landings were recorded on the homepage where the respondents had to choose the language of their choice, out which 1303 proceeded to the page in Turkish, 13 navigated to the page in English, whereas the rest dropped out from the landing page itself. Once the data was collected, all of it was translated back to English language. Thereafter we analyze the demographics to describe the sample and show that it is representative (see Section 4.2).

Figure 1. Research framework.

We analyzed the attitudes and opinions of the respondents towards the environment, buying new technology innovation, knowledge about SM, preferences and conditions under which they would accept SM and communication channels they would prefer to use while seeking more information about SM, independently. After that we created a binary Logit model, with a dependent variable as the knowledge about SM (K1) and the other variables as the regressors.

4.2. Description of the Sample The collected sample of Turkish nationals and residents is represented by both male and female (4:3) who are primarily either single or married and are between the 18 to 35 years of age. Majority of them have at-least a high school diploma or have completed a bachelor degree. A large number of the respondents are unemployed, whereas the other are employed in public or private sectors. A small number of respondents also have their own business. Cross tabulation of age with the gender, relationship status, educational level and employment status, shown in Table2. Energies 2020, 13, 732 9 of 27

Table 2. Cross tabulation of Age in Year (D2) with Gender (D1), Relationship Status (D3), Educational Level (D4) and Employment Status (D5).

Age in Years (D2) Code Variable Options Total 18–25 26–35 36–45 46–55 56–65 66+ Male 146 64 32 18 25 3 288 D1 Female 96 55 34 27 4 0 216 Single 177 28 9 1 1 0 216 In a Relationship 55 12 0 1 0 0 68

D3 Married 10 77 54 38 27 2 208 Separated/Divorced 0 2 2 3 0 0 7 Widowed 0 0 1 2 1 1 5 No formal education 0 0 0 0 0 0 0 Primary School Only 1 1 1 2 1 0 6 High School Pass 106 12 12 6 6 1 143 D4 Bachelor Complete 126 75 33 31 21 2 288 Masters Complete 9 24 12 2 1 0 48 PhD complete 0 7 8 4 0 0 19 Job in Private Sector 59 57 34 11 8 0 169 Job in Public Sector 3 33 16 21 6 0 79 Business 11 7 8 0 0 0 26

D5 Student 0 0 0 0 0 0 0 Unemployed 169 22 8 7 0 0 206 Retired 0 0 0 6 15 3 24 Note: Cells in the table, highlighted in yellow, depict the majority of the sample.

The net household income of large majority of the respondents is lower than 10,000 TL and they pay between 0 and 325 TL (1 TL = 0.17 USD) for their monthly electricity consumption. They live in apartment or ﬂats in multi-storied buildings. The cross tabulation in the Table3 shows the distribution of relative respondents between the two variable pairs, D6 with D7 and D9 with D7. Further, we cross tabulated the area of residence of the respondents with the number of members in the household, number of children and the type of residence they live in, shown in Table4. It shows a that large part of the respondents are from cities with populations of more than 500,000, with 2 to 5 members living in the household and more than 60% having no children. The most common type of residence is an apartment or a ﬂat in a multi-storied building, as mentioned in the previous paragraph. As the study was carried out among Turkish social media users, we compared the demographics of the respondents in this study with the demographics of the social media users in Turkey, to show that the sample is representative. A large majority of social media users in Turkey are young, between the age of 18 to 35, have at least a high school diploma, are either unemployed or are in private sector jobs and live in cities (for more details see: https://datareportal.com/reports/digital-2019-turkey (accessed: 8 January 2020)). The highlighted sections in Table2, correspond to the majority of the respondents in this study as well as the demographics of Turkish social media users in actual, hence we can consider the data to be representative. Energies 2020, 13, 732 10 of 27

Table 3. Cross tabulation of monthly net household income in TL (D6) and type of residence (D9) with the month electricity expense in TL (D7).

Monthly Electricity Bill (D7) Code Variable Options Total 0 1 2 3 4 5 Prefer Not to Say 14 21 12 2 0 1 50 0–2500 1 78 21 0 1 0 101 2501–5000 3 112 34 2 0 1 152 5001–7500 0 75 31 0 0 0 106 7501–10,270 0 32 14 0 0 0 46 10271–14,885 0 13 16 0 1 1 31 D6 14,886–17,550 0 3 3 0 1 1 8 17,551–21,450 0 1 1 1 0 0 3 21,451–27,072 0 0 1 0 0 0 1 27,073–40,014 0 1 4 0 0 0 1 40,015–53,950 0 3 0 0 0 0 3 more than 53,950 0 2 0 0 0 0 2 Apartment/Flat * 2 48 21 0 0 1 72 Apartment/Flat ** 15 275 95 4 2 2 393

D9 House + 0 7 5 0 0 0 12 House ++ 1 11 12 1 1 1 27 * In a building of 4 floors or less. ** In a building with more than 4 floors. + Only group floor. ++ Multiple floors. Note: Cells in the table, highlighted in yellow, depict the majority of the sample.

Table 4. Cross tabulation of Age in Year (D2) with Gender (D1), Relationship Status (D3), Educational Level (D4) and Employment Status (D5).

Area of Residence (D10) with Population in Brackets (k = 1000) Code Variable Options Sum Village City (<50k) City (50k–100k) City (100k–500k) City (>500k) One 0 2 2 2 27 33 Two 1 3 6 5 43 58 Three 5 7 10 12 106 140

D8 Four 2 4 9 13 131 159 Five 1 4 4 9 64 82 Six or More 1 1 7 7 16 32 None 5 9 21 32 243 310 One 3 5 6 5 45 64 Two 1 3 6 8 69 87 D81 Three 0 3 4 3 26 36 Four or More 1 1 1 0 4 7 Apartment/Flat * 1 12 10 3 46 72 Apartment/Flat ** 0 7 24 42 320 393

D9 House + 5 0 2 1 4 12 House ++ 4 2 2 2 17 27 * In a building of 4 floors or less. ** In a building with more than 4 floors. + Only group floor. ++ Multiple floors. Note: Cells in the table, highlighted in yellow, depict the majority of the sample.

To make sure even more that the data is representative in terms of the usage of various social media platforms, we have asked the respondents about the social media platforms they were active on. Energies 2020, 13, 732 11 of 27

Figure2 shows the percentage of respondents (in this study) active on various social media platforms and the percentage of Turkish internet users active on the same social media platforms (for more details see: https://datareportal.com/reports/digital-2019-turkey (accessed: 8 January 2020)), which are quite similar. The average household income and the expenditure of electricity bills of a Turkish household, shown in Table3, also corresponds to the actual average of a Turkish household (For more details see: http://www.oecd.org/economy/surveys/Turkey-2018-OECD-economic-survey- overview.pdfandhttps://www.guidesglobal.com/utilities-in-turkey/ (accessed: 8 January 2020)). Finally, the average number of members in household, household type and average number of children, shown in Table4, also coincide with the actually numbers in Turkey (for more details see: https://ec. europa.eu/eurostat/statistics-explained/index.php/Household_composition_statistics (accessed: 8 January 2020)). Hence, the data collected in this study can be considered as representative of the social media users in Turkey.

Figure 2. Percentage of users active on various social media platforms.

4.3. Knowledge about SM A total of 250 out of the 504 respondents, just shy of 50% of the sample, indicated that they knew what an SM was. Of these 250, 67 already have an SM installed at home, whereas 24 are in the process of having an SM installed and 45 have a plan of getting an SM installed at their homes. This indicated that still, almost 50% of the respondents who know what an SM is, neither have an SM installed, nor have any plans of having SM installed at their homes. But still, over 60% of those who knew what an SM was expressed that they would like to have an SM installed at their home, a majority of which were those who were positive towards K2, K3 or K4.

4.4. Belongings and Assets Figure3 shows the mean and standard deviation of the belongings and assets (B1–B7) of the respondents. Respondents were either in possession of or were planning to buy a B2—ﬂat or apartment (Skewness = 0.825, Std. Error = 0.109), B3—a laptop (Skewness = 1.794, Std. Error = 0.109), B4—wiﬁ − − or internet connection at home (Skewness = 2.719, Std. Error = 0.109) and B5—home appliances that − can connect to the internet (Skewness = 1.936, Std. Error = 0.109). At the same time, the respondents − did not possess and were not even planning to buy B1—a house, B6—electric vehicle and B7—smart Energies 2020, 13, 732 12 of 27 technologies that enable energy consumption or monitoring (Skewness = 1.122, 2.273 and 1.341 respectively with Std. Error = 0.109).

Figure 3. Mean and standard deviation of belongings and assets of the respondents.

4.5. Attitude and Behavior towards Pro-Environmental Activities The analysis of the variables A1, A2 and A31–A39 revealed that, in terms of upgrading their home appliances with newer versions (A1—Mean = 1.21, SD = 0.75) and buying new phones to get latest technologies (A2—Mean = 1.63, SD = 1.006), the respondents favored longer use of technology. They upgrade home appliances with new ones, once in three years or more, and new mobile phone in a bit lesser duration i.e., in two or more years. To confirm that this buying behavior was not dependent on the household income (D6), we checked the correlations between D6 and A1 and D6 and A2, which was found to be insignificant. A total of 21.5% of respondents also reported to have renewable energy sources installed at home, with no significant correlation with household income or knowledge about SM nor the monthly electricity expenses. Table5 shows the responses towards various behavioral questions and the mean of the responses (No — 0 and Yes — 1). It can be seen that majority of the respondents are involved in activities which are pro-environmental and are in favor of energy saving, as the mean values of 7 out of the 9 activities are over 0.5. Energies 2020, 13, 732 13 of 27

Table 5. Mean values of variables A31–A39 (N = 504).

Activity Mean

(A31) I follow organizations or proﬁles on social media that promotes saving of energy 0.27

(A32) I have searched on the internet about how to live in a eco-friendly living way 0.46

(A33) I reuse grocery bags 0.89

(A34) I have invested in energy saving appliances for my home 0.57

(A35) I regularly monitor energy consumption at home 0.7

(A36) I segregate garbage at home 0.69

(A37) I have returned home, sometimes, to ensure that I switched off the home appliances or the lights etc. 0.79

(A38) I have paid more for buying a more energy efﬁcient appliance 0.6

(A39) I have picked up trash left by somebody else while being outdoor 0.85

4.6. Preferences and Attitude towards Acceptance of SM SM, combined with in-home displays or mobile applications and other devices, offer benefits such as a detailed depiction of consumption of electricity, real time information on the consumption of electricity, ability to switch on/off supply and monitor tariffs in case of fluctuating unit rates [61]. Preferences of the respondents indicate that they would like to have these benefits. A total of 70.8% of the respondents desired to get more information about their use of electricity, 72.2% respondents said that having real time information of energy consumption would be useful for them, 81% respondents would like to have the ability of controlling the power supply to the appliances through mobile application and 76.4% respondents would prefer to have fluctuating unit rates for electricity during the day, so that they can consume more when electricity is cheaper. We asked the respondents that knew what an SM was, whether they would prefer the government to make it mandatory to install an SM for everyone. The response was pretty even with 49.2% opposing, whereas the rest were in favor of it. In an additional question asked to confirm this, over three quarters (79.6%) of the respondents wanted the government to offer SM as an option. A significant minority (13.6%) of the respondents also expressed the intention to protest if they did not have an option to say no to installing SM at their home. The same respondents were asked if they knew about government’s national rollout program for SM and 76% of them did not know about it. We checked the correlations between having information about the national rollout program (I2) and their preferences towards governmental policy (G1, G2 and G3), through Kendall’s Tau B test, which showed that there is a high and statistically significant relation between them. Increase in the information about the national rollout program increases the probability of consumers accepting the government making it mandatory to install SM for all residents (correlation co-efficient: 0.160 and p-value (2-tailed): 0.12).

4.7. Fears towards Acceptance of SM Previous studies have shown the consumers have fears regrading their data privacy, health effects and inaccurate billing [11,34,44,45,62,63]. In some cases the consumers have a fear of coping with the change (for instance from current meters to SM), because they feel that such changes result in increased expenses [64]. In this study we found that over half (51.59%) of the respondents had fear of privacy breach (F1) if companies had access to the data of their detailed electricity usage. Additionally over 43% of respondents had concerns about increased stress (F4) on them due to ﬂuctuations in electricity prices. Among the respondents who indicated to have knowledge about SM, over 64.4% (N1 = 250) Energies 2020, 13, 732 14 of 27 raised concern about inaccurate billing, whereas only 6% of respondents had a fear of any adverse effect on health because of SM.

4.8. Willingness to Accept SM under Different Conditions The respondents willingness to accept the installation of SM under various conditions was found to vary (see, Figure4) between a mean as low as 0.0814 to as high as 0.8631. Respondents were highly willing to accept SM (De1) if they were convinced that SM would help them save energy and money. This did not have statistically signiﬁcant correlation with the household income (D6) or the monthly electricity expenses (D7). The willingness to install SM is at its lowest when the respondents feel that there might be some adverse effect on health (De2) even if they would be able to save on electricity bills.

Figure 4. Mean values of De1–De8. * For De5 we have N1 = 250.

In terms of energy companies having access to data of energy usage (De3), over half of the users are concerned with the data privacy issues. We found a statistically significant correlation (coefficient: 0.22, p: 0.000) between the concern regarding data privacy (F1) and D 3. It was also interesting to − e observe similar numbers in comparing the cross tabulations of variables F1 and De3 for the respondents with K1 = 0 and K1 = 1, shown in Tables6 and7 respectively. This shows that near 20% of all the respondents, are having a data privacy concerns but are still willing to accept SM if it helps them save on electricity bills. Just over 10% of the respondents do not have privacy concerns, in case the energy companies had access to their consumption data but are not willing to accept installation of SM under this condition. A visit from a company representative (De4) or recommendation from a friend, family member or colleague (De7) would convince around 45% of the respondents to accept installation of SM. About 40% of respondents were willing to install SM if a friend, a family member or a colleague has installed SM at their home (De8). De7 and De8 are significantly related to each other (coefficient: 0.513, p: 0.000) which indicates that social influence for the willingness to accept SM would be about 50% higher if the person making the recommendation has an SM installed at his/her home. Willingness to accept SM is high, at nearly 65% if the installation of SM or upgrade to SM is free for the consumers. Among the respondents who have knowledge about SM, approximately 22% are willing to accept SM even if they have to pay for having SM at their home (De5).

Table 6. Cross tabulation of variables F1 and De3, for respondents with K1 = 0.

De3 0 1 Total 0 59 59 118 F1 1 87 49 136 Total 146 108 254 Energies 2020, 13, 732 15 of 27

Table 7. Cross tabulation of variables F1 and De3, for respondents with K1 = 1.

De3 0 1 Total 0 48 78 126 F1 1 84 40 124 Total 132 118 250

4.9. Communication Channels for SM Awareness Communication channels preferred by the consumers’ play a vital role in the information diffusion as well as raising consumers’ awareness and addressing their concerns [42,52]. The respondents were asked a variety of questions to understand the communication channels in terms of electricity market in general and for SM diffusion in particular. Figure5, shows a comparison of three important variable sets: (i) sources of information regarding electricity (S1–S15), • (ii) sources of information regarding SM (I31–I35), asked to only those who had knowledge about • SM (K = 1, N1 = 250), (iii) channels (Q21–Q35) they would prefer to use, to get more information about SM (Q1 = 1, • N2 = 275).

Figure 5. Sources of information regarding electricity (S), regarding SM (I3) and sources on which users would like to search for more information regarding SM (Q2). Note: S1–S15 correspond to the blue bars 1–15 respectively, I31–I45 correspond to yellow bars 1–15 respectively and Q21–Q35 correspond to red bars 1–15 respectively. (FRC: Friends, Relatives, Colleagues, FBM: Facebook Messenger, EC: Energy Companies, OGW: Ofﬁcial government websites, WEC: Workshops/Educational Campaigns, Tele: Telephone/SMS, SE: Search Engine).

From Figure5, it can be seen that the most common source of information regarding electricity market in general as well as SM in particular was TV News, followed by the social peers such as friends, Energies 2020, 13, 732 16 of 27 relatives and colleagues. Social media channels such as Facebook, Twitter, Whats App and YouTube were mentioned by 20% to 25% of respondents as sources from where they got some information regarding the electricity markets, but the same sources were less popular to get information about SM. Nearly 30% of the respondents who knew about SM mentioned energy companies as one of the source of information regarding SM, but the numbers dropped to 10% for the ofﬁcial government websites. Several respondents (76 out of 250, for K1 = 1) indicated that they searched for information about SM (I1), but surprisingly only 2 of them indicated a government website to be the source of information regarding SM. Overall the number of respondents who indicated search engines as a source SM were less than 3%. At the end of the questionnaire, we asked all the respondents if they would try to get more information regarding SM (Q1) after participating in this study. N2 = 275 respondents who said ’Yes’ to Q1, were asked about the channels they would prefer to use for searching the information regarding SM (Q21–Q35). Computing the aggregate answer of Q21–Q35 for each respondent, we found that for the whole sample, the mean equals 3.3782 (SD = 2.3527). Despite the fact that we conducted this study among social media users, we can clearly see from the mean and standard deviation of the aggregate answers, as well as from the graph, that the respondents’ choice of communication channels is wide and not just restricted to social media channels. In fact the conventional communication channels such as ofﬁcial government websites, energy companies themselves, TV news and social peers were amongst the most common choices. YouTube was the most common among social media channels with almost 40% of respondents indicating it as their choice, followed by Facebook, Twitter and WhatsApp. These social media channels are also among the top used platforms of the respondents in this study as well as for Turkey (see Figure2). Although search engines were chosen by only 12% of the respondents who were in the category of Q1 = 1, they have a major role to play, in order to make the content from the energy companies as well as government websites reachable for the consumers.

4.10. Modeling of Knowledge about SM among Consumers and Discussion In order to evaluate the impact of different variables on the knowledge regarding SM (K1), a binary logistic regression model was used. This model reveals the statistically significant variables which influence the knowledge about SM among the consumers. In the regression analysis, only the variables corresponding to questions asked to all of the respondents (N = 518) were considered. The detailed description of these variables can be seen in Table A1 (AppendixA). In the model, first a binary variable Yi is constructed. Yi takes value one when an i-th individual responds positively to K1 (K1 = 1) and zero when a respondent declares not to know what an SM is (K1 = 0). Thereafter the probability of consumers having knowledge about SM (K1 = 1) is assumed to be dependent on a set of variables Xi, which includes a constant and all the variables shown in group A of Table A1 (AppendixA). A general description of the logistic regression model is given in the Equation (1) below,

exi β Prob(Yi = 1) = (1) (1 + exi β)

where β is a vector of the model coefficients and Xi stands for a vector of the explanatory variables. The aim of the model is to describe the probability of having knowledge what SM is with a set of the explanatory variables. The results of the model are presented in Table A2 (AppendixA), which shows that are many insignificant variables. To eliminate the insignificant variables we carried out a sequential elimination of variables using two-sided alpha = 0.05 (similar to [65]) using the Gretl program. In this process, the variable having the highest p-value was omitted step by step until no remaining variable had a p-value greater than the cutoff (which was 0.05 in this case). The final model obtained after the elimination is shown in Table8. Energies 2020, 13, 732 17 of 27

Table 8. Results of the ﬁnal Logit regression model for dependent variable: K1 (standard errors based

on Hessian) for determination of Yi.

Variable Coefficient Std. Error z Marginal Effect p-Value const 3.423 0.642 5.335 0.000 − − D2 0.371 0.087 4.285 0.093 0.000 D10 0.264 0.111 2.381 0.066 0.017 S01 0.481 0.209 2.297 0.120 0.022 S03 0.613 0.256 2.393 0.151 0.017 S1 0.517 0.242 2.132 0.128 0.033 − − − S12 0.586 0.258 2.269 0.144 0.023 A34 0.580 0.209 2.775 0.144 0.006 A35 0.892 0.233 3.834 0.217 0.000 De6 0.550 0.238 2.314 0.136 0.021 De7 0.835 0.262 3.186 0.206 0.001 − − − De8 1.112 0.250 4.441 0.271 0.000 Mean dependent var 0.496032; S.D. dependent var 0.500481; McFadden R2 0.161733; Adjusted R2 0.127382; Log-likelihood 292.8319; Akaike criterion 609.6639; Schwarz criterion 660.3348; Hannan–Quinn 629.5403. − Results of the Logit model, detailed in Table8, show that the knowledge regarding SM is positively related with the following variables: age (D2), area of residence (D10), usage of Facebook (S01), usage of LinkedIn (S03), government website as source for information regarding electricity (S12), attitude towards investing in energy saving appliances (A34), attitude towards monitoring energy consumption in household (A35), acceptance of SM if it was a free upgrade (De6) and acceptance of SM if friends/relatives/neighbor installs it at their house (De8). This indicates a stepping up of the age group by 1 (for example from 25–35 to 36–45), while leaving all the other explanatory variables unchanged, there is a 9.3% higher probability of the consumers having knowledge about SM. Similarly there is a 6.6% more probability of consumers living in cities with higher population. In case of variables S01, S03, S12, A34, A35, De6 and De7, changing the individual negative responses to positive (while keeping all other explanatory variables unchanged) increases the probability of having knowledge about SM by 12%, 15.1%, 14.4%, 14.4%, 21.7%, 13.6% and 27.1%, respectively. Two variables, TV channels as a source of information regarding electricity (S1) and acceptance of SM if one of the friends/relatives/neighbors recommends it (De7) influences the knowledge in the opposite way. Changing the responses of these variables individually to positive, while keeping all the other explanatory variables unaltered, decreases the likelihood of having knowledge regarding SM by 12.8% and 20.6%, respectively. The prediction capabilities of the final Logit model is shown in Table9, which shows that 69.25% of the responses were correctly predicted by the model.

Table 9. Prediction capabilities of the ﬁnal Logit model.

Predicted % Correct Yi = 0 Yi = 1 Y = 0 175 79 68.9% Observed i Yi = 1 76 174 69.6% Overall Percentage 69.25%

The final model appropriately fits the data, as the the Hosmer and Lemeshow goodness-of-fit is greater than 0.05 and the value of joint signifiance test is Chi-Square = 112.997 with p-value = 0.000. Final Logit model for is represented by Equation (2) Prob logit(Prob) = log = 3.423 + 0.371(D2) + 0.264(D10) + 0.481(S01) + 0.613(S03) + 0.586(S12) (1 Prob) − − (2) + 0.580(A34) + 0.892(A35) + 0.550(De6) + 1.112(De8) 0.517(S1) 0.835(De7) − − Energies 2020, 13, 732 18 of 27

5. Conclusions, Discussion and Recommendations

5.1. Conclusions and Discussion Turkey is facing growing demand for electricity. A trend, which is projected to continue in the coming years. Increasing the energy efficiency of the transmission grids, by upgrading to SG, is one of the feasible solutions that has been looked upon by the Turkish government, apart from moving towards renewables. A national rollout of SM has been initiated with CLK Enerji, Turkey’s largest electricity distribution and retain group, undertaking the replacement of current electricity meters with SM in 11 provinces. The consumers’ role is vital for the success of this technological upgrade. Studies in the literature indicate that enabling consumers the access to the information about their electricity consumption is not enough to guarantee benefits such as energy and money saving. The savings can be possible only if the consumers are willing to accept SM together with SMP, and engage in monitoring their usage of energy. Complete knowledge about SM, its features and benefits, are more likely to draw acceptance and engagement from the consumers. To reach out to consumers and disseminate the knowledge about SM, a range of communication channels would have to be used, as different groups of consumer prefer different communication channels. Our study addresses both these factors, determinants of knowledge regarding SM and the communication channels preferred by the consumers in Turkey.

5.2. Determinants of Knowledge about SM Our study shows that consumers’ knowledge about SM increases with age. Knowledge about SM was found to be higher among the aged consumers as compared to the young ones, who according to the literature are more tech-savvy, also referred to as early adopters of technology and more active on social media. This shows that the dissemination of information regarding SM is not adequate on social media platforms as a large majority of users on social media are of younger age groups. At the same time consumers active on social media platforms, like Facebook and LinkedIn, were more likely to have knowledge about SM, as compared to those who were not active on them. These young social media users can act, as so called influencers, to facilitate the propagation of information regarding SM enhancing the outcomes of the dissemination campaigns. Moreover, consumers who rely on government websites as the source of information for electricity are more likely to have knowledge about SM whereas those relying on TV news are found to be less probable to have knowledge. People, with a more pro-environmental attitude, such as a willingness to invest more for energy saving appliances and monitoring energy consumption were also found to be more probable to have knowledge about SM. Finally, the knowledge of what an SM is also correlates positively with the consumers’ willingness to install SM if the upgrade is for free and if their peers have already installed SM. It is interesting to point out that the social influence has an important role to play for the acceptance of SM in cases where the consumers do not have knowledge about SM. This is evident through the negative correlation of the knowledge about SM with the willingness to accept SM if one of the friends, relatives or neighbors recommends it. The fact that the level of education (D4) was not significant for the knowledge about SM points out to the lack of attention in the educational curriculum regarding the socio-economic aspects of SM. For older age groups, above 25, the insignificance of the educational level for knowledge about SM can be justifiable, as when they would have been in school or in a higher educational institution, SM was not a popular concept. But, younger respondents were still going through the educational system when SM became a popular concept over a decade ago. Hence there is a need for the educational system, in Turkey, to address SM and related topics for the younger generation.

5.3. Communication Channels Our study, conducted among social media users, shows that using a wide spectrum of communication channels is quite important. For instance “TV News” was found to be the top source Energies 2020, 13, 732 19 of 27 of information regarding electricity among the consumers. But, at the same time it has a negative relation with knowledge about SM, which indicates that consumers who rely more on TV news for information regarding electricity are not receiving knowledge about SM through it. The preferences regarding the communication channels to receive more information regarding SM also shows wide range of choices of both conventional and digital platforms including social media. Due to the high importance of social inﬂuence in acceptance of SM, social media platforms would be highly effective to create snowball effect for spreading knowledge about SM.

5.4. Recommendations for Social Media Management to Enhance Consumer Acceptance of SM Based on the obtained results we recommend that the following steps would aid in enhancing consumer knowledge and acceptance of SM: A huge number of consumers are missing the primary information about SM, which requires an • effective awareness campaign. Although there are some media outlets who have published articles about the SM rollout, it has yet to reach critical mass. Energy companies have posted content regarding SM on their websites, but there is a need to reach out to consumers through various communication channels, especially social media, disseminating the published information. It would be of great value to initially target the social media users in Turkey, as they are in considerable numbers and provide an opportunity to create a snowball effect for positive e-word of mouth. Initiating small campaigns through local influencers, user stories, use cases, the benefits of SM and especially addressing the fears/concerns discussed in Section 4.7 would prove to be effective. In continuation with the previous point, it would be valuable to create online social communities • which would facilitate discussion and interaction among the consumer regarding the use of SM. In the study we found that social influence can play a role in increasing the acceptance of SM among consumers. Moreover, such online social communities also provide an opportunity for the energy companies to create gamification campaigns which would engage the consumers further. In some cases we found that lack of proper knowledge or incomplete knowledge was one of the • reasons for false fears, which requires swift rectification. Creating a standard information package for users, based on their preference of different types of communication channels, with information about SM basics at first and then the advanced features, is recommended. Consumers active on different communication channels are in sync with the type of content propagated through that communication channel. For instance, users preferring to get information through videos are more active on YouTube as compared to a text blog and vice-versa in case of users preferring to read text instead of watching videos. An information package for different channels would ensure that consumers with different attitudes towards obtaining information would be addressed. Based on the results of this study as well as the previous one [34], the information containing the following topics is recommended:

– Basic knowledge about SM, its functions, myths, long term and short term impacts, potential benefits: financial, social, environmental and economic ones. – Usefulness of SM: monitoring of energy use, ability to remotely control energy usage and getting real time information. – Addressing fears and concerns: security of personal data, safety features, health issues, accuracy in billing and others. – Involving the social factors: interactions through experts with the support of social influencers and current users of SM, consumer feedbacks and experiences or assurances.

6. Limitations of the Study and Future Work The present study has certain limitation, although vigorous efforts were made to broaden the scope of research, which point to future research avenues. The analyzed sample introduces limitations in Energies 2020, 13, 732 20 of 27 terms of geographical context and the non-random sampling methods used. Hence it is recommended to replicate this study in different samples, for instance independent studies in different regions of Turkey, to obtain a more specific regional understanding. Respondents for this study were located from almost all regions of Turkey, but we did not take into consideration the regional effect on the responses. This can prove to be important, as Turkey’s transition plan includes getting region specific requirements. Secondly, the study can be replicated in other countries where SM rollout is at an initial phase or is about to commence. This particular study is a part of our larger project, under which similar studies have already been conducted in five countries and we also aim to carry out further similar studies in other countries. We also feel that testing the effectiveness of recommendations for SM diffusion, outlined in this study, could be one of the future steps. The study also lays grounds for further research on creation and testing of online social communication for diffusion of SM knowledge. Similar studies for understanding consumer attitudes and opinions for other innovations/smart technologies can also be undertaken.

Author Contributions: Y.C. conceived and designed the survey; A.K.-P. reviewed the design of the survey; B.O. carried out translations (English–Turkish and vice-versa) of the questionnaire; Y.C. created and managed the online questionnaire; B.O. collected the data through the online questionnaire; Y.C. analyzed the data; A.K.-P. and B.O. reviewed the literature; A.K.-P. and Y.C. drafted and edited the paper; A.K.-P. and Y.C. revised the paper. All authors have read and agreed to the published version of the manuscript. Funding: This work was supported by the Faculty of Computer Science and Management, Wrocław University of Science and Technology from funds of the Ministry of Science and Higher Education subsidy in the part devoted to conducting research activities in 2019 and by the National Science Center (NCN, Polska) by grant no. 2016/23/B/HS4/00650. Acknowledgments: We would like to thank the editors and reviewers for their constructive remarks and suggestions. We want also to thank Marta Pytel for English language proofreading. Conﬂicts of Interest: The authors declare no conﬂict of interest.

Abbreviations The following abbreviations are used in this manuscript:

SG smart grids SM electricity smart meters SMP smart metering platform (SM information systems) TL Turkish Lira DSM/DR Demand Side Management and Demand Response tools DoI Diffusion of innovation model

Appendix A

Table A1. Deﬁnitions of the variables, coding and description (N = 504).

Variable Code Description Group A (Variables of Phase 1) Demographics D1–D10, D81 Gender D1 1 = male, 2 = female Age D2 6 categories (ordinal) Relationship status D3 5 categories (nominal) Highest Educational Qualiﬁcation D4 6 categories (ordinal) Occupation/Employment D5 6 categories (nominal) Monthly Household Income (in TL per month) D6 12 categories (ordinal) Range of electricity bill (in TL per month) D7 5 categories (ordinal) Total members in the household D8 6 categories (ordinal) Number of children D81 5 categories (ordinal) Type of house D9 4 categories (nominal) Place of living D10 5 categories (ordinal) Energies 2020, 13, 732 21 of 27

Table A1. Cont.

Variable Code Description Belongings of smart devices and personal assets) B1–B7 House B1 (2) yes/(1) no, but I plan to buy it within one year/(0) no, and I do not plan to buy it. Flat or Apartment B2 Laptop B3 Wifi/Internet connection home B4 Home appliances that can connect to the internet B5 Electric vehicle B6 Any smart technologies that enable monitoring and B7 control of energy consumption at house Behaviour towards buying new technology A1–A2 Upgrading electronic home appliances with new A1 (1) I buy once in more than three versions years/(2) I buy once every three years/(3) I buy once every two years/(4) I buy once in a year/(5) I buy twice per year/(6) I buy thrice or more per year Buying new mobile phone to get latest technology A2 Behaviour towards environment and energy A31–A39 saving Followed any organization(s) or profile(s) on social A31 media promoting energy saving Performed internet search about eco-friendly ways A32 of living Re-used grocery bags A33 (1) yes/(0) no Invested in energy saving appliance(s) for A34 household Regular monitored of energy consumption at A35 household Segregated garbage A36 Ever returned home to check whether all home A37 appliance(s) or light(s) are turned off Ever paid more for buying more energy efficient A38 appliance Ever picked up trash left by somebody elese, while A39 being outdoor Renewable energy sources installed at the R1 (1) yes/(0) no household Social media platforms commonly used S01–S08 Facebook S01 Facebook Messenger S02 LinkedIn S02 Twitter S03 (1) yes/(0) no WhatsApp S04 Youtube S05 Instagram S02 Snap Chat S02 Source of information regarding electricity S1–S15 (1) Yes/(0) No (Note: variables are listed in (prices, new offers, etc.) the table notes) Preferences regarding SM platforms P1–P4 Getting more details on electricity usage is P1 desirable (1) yes/(0) no Getting real time information of electricity usage P2 would be useful Prefer to be able to remotely turn on or off the P3 electricity supply Prefer to have fluctuating unit rates of electricity P4 usage Energies 2020, 13, 732 22 of 27

Table A1. Cont.

Variable Code Description Willingness to install SM De1–De4, De6–De8 Willing if SM could help save money De1 Willing if SM could help save money, but possible De2 have adverse effect on health Willing if SM could have save money, but energy D 3 e (1) yes/(0) no companies would have access to electricity usage data Willing if company representative visits home and De4 expalin all details Willing to install if upgrade to SM is free De6 Willing to install if one of the De7 friends/relatives/neighbours recommends it Willing to install if one of the De8 friends/relatives/neighbours installs SM at their house Concerns about SM usage F1,F4 Data privacy concerns F1 (1) yes/(0) no Fluctuations in unit rate of electricity would cause F4 additional stress Group B (Conditional Variable for Phase 2) Knows what is a SM K1 (1) yes/(0) no Group B (Variables of Phase 2—For K1 = 1) Knowledge about SM K2–K4 Has SM installed at home K2 In process of installing SM at home K3 (1) yes/(0) no Plans to install SM at home K4 Source of information regarding SM I1, I2, I31–I45 Internet, other than search engine I1 Government’s rollout programme I2 (1) yes/(0) no Other sources (variables are listed in the table notes) I31–I45 Social inﬂuence W1 (1) yes/(0) no Willing to install SM even if payment is required De5 (1) yes/(0) no for upgrade Preferences regarding the role of the government G1–G3 in SM enrollment Government should make it mandatory for all to G1 have SM (1) yes/(0) no Government should give an option to decline G2 installation of SM Would protest if government makes it mandatory G3 to install SM Concerns about SM usage F2, F3 Billing through SM could be inaccurate F2 (1) yes/(0) no SM could have adverse effects on health F3 Willingness to have one’s home to be equipped X1 (1) yes/(0) no with SM Group C (Conditional Variable for Phase 3) Willingness to search or collect more information Q1 (1) yes/(0) no regarding SM Group C (Variables of Phase 3—For Q1 = 1) Source of Information preferred to search or Q21–Q35 (1) yes/(0) no (Note: variables are listed in collect more information regarding SM the table notes) Note: TV News (S1, I31, Q21); Radio (S2, I32, Q22); Newspaper (S32, I33, Q23); Friends, relatives, colleagues (S4, I34, Q24); Facebook (S5, I35, Q25); Facebook Messenger (S6, I36, Q26); Twitter (S7, I37, Q27); WhatsApp (S8, I38, Q28); LinkedIn (S9, I39, Q29); YouTube (S10 I40, Q30); Energy Companies (S11, I41, Q31); Ofﬁcial government websites (S12, I42, Q32); Workshops/educational campaigns (S13, I43, Q33); Telephone/SMS (S14, I44, Q34); Search engines (S15, I45, Q35). Energies 2020, 13, 732 23 of 27

Table A2. Results of initial Logit Regression Model for Dependent variable: K1 (Standard errors based

on Hessian) for determination of Yi.

Variable Coefﬁcient Std. Error z p-Value const 4.34585 1.46679 2.963 0.0030 − − D1 0.0931032 0.255792 0.3640 0.7159 D2 0.267239 0.165476 1.615 0.1063 D3 0.279179 0.185745 1.503 0.1328 D4 0.0955307 0.175709 0.5437 0.5867 D5 0.0136375 0.0664223 0.2053 0.8373 D6 0.00488104 0.0727889 0.06706 0.9465 D7 0.0973837 0.170998 0.5695 0.5690 − − D8 0.109854 0.102405 1.073 0.2834 D81 0.0657327 0.191642 0.3430 0.7316 − − D9 0.0210567 0.197214 0.1068 0.9150 D10 0.310426 0.138547 2.241 0.0251 S01 0.579175 0.290663 1.993 0.0463 S02 0.0210224 0.333660 0.06301 0.9498 S03 0.680142 0.318111 2.138 0.0325 S04 0.525183 0.304589 1.724 0.0847 − − S05 0.234781 0.468434 0.5012 0.6162 − − S06 0.144067 0.293251 0.4913 0.6232 − − S07 0.111510 0.330682 0.3372 0.7360 − − S08 0.142987 0.344899 0.4146 0.6785 B1 0.0700210 0.153048 0.4575 0.6473 B2 0.0715037 0.136685 0.5231 0.6009 − − B3 0.183646 0.171654 1.070 0.2847 B4 0.0629511 0.243642 0.2584 0.7961 − − B5 0.248945 0.186826 1.332 0.1827 B6 0.212604 0.210233 1.011 0.3119 B7 0.127893 0.165379 0.7733 0.4393 S1 0.577253 0.274246 2.105 0.0353 − − S2 0.0121528 0.351380 0.03459 0.9724 S3 0.178391 0.273498 0.6523 0.5142 S4 0.0291626 0.245521 0.1188 0.9055 S5 0.0659728 0.329382 0.2003 0.8413 − − S6 0.201088 0.685628 0.2933 0.7693 S7 0.164173 0.325877 0.5038 0.6144 S8 0.542389 0.322082 1.684 0.0922 − − S9 1.13231 0.697245 1.624 0.1044 S10 0.0283239 0.332303 0.08524 0.9321 − − S11 0.0151670 0.360266 0.04210 0.9664 − − S12 0.521208 0.296452 1.758 0.0787 S13 0.249130 0.950570 0.2621 0.7933 S14 0.278293 0.283591 0.9813 0.3264 S15 0.784459 1.34753 0.5821 0.5605 A1 0.198726 0.156349 1.271 0.2037 − − A2 0.0856742 0.123348 0.6946 0.4873 A31 0.300665 0.274168 1.097 0.2728 A32 0.514857 0.269346 1.912 0.0559 − − A33 0.171275 0.370348 0.4625 0.6437 − − A34 0.696631 0.249602 2.791 0.0053 A35 1.10592 0.278474 3.971 0.0001 A36 0.401872 0.254818 1.577 0.1148 A37 0.231040 0.292856 0.7889 0.4302 − − A38 0.222811 0.258386 0.8623 0.3885 − − A39 0.293846 0.333667 0.8807 0.3785 − − R1 0.216237 0.285061 0.7586 0.4481 P1 0.303739 0.308630 0.9842 0.3250 P2 0.390941 0.320582 1.219 0.2227 − − P3 0.0158712 0.335997 0.04724 0.9623 P4 0.0849537 0.297580 0.2855 0.7753 Energies 2020, 13, 732 24 of 27

Table A2. Cont.

Variable Coefﬁcient Std. Error z p-Value F1 0.0296559 0.235794 0.1258 0.8999 F4 0.190290 0.237482 0.8013 0.4230 De1 0.436749 0.364867 1.197 0.2313 − − De2 0.968640 0.447898 2.163 0.0306 − − De3 0.306588 0.248226 1.235 0.2168 De4 0.396967 0.283658 1.399 0.1617 De6 0.591198 0.278186 2.125 0.0336 De7 1.11000 0.308945 3.593 0.0003 − − De8 1.33244 0.293184 4.545 0.0000 Q1 0.189544 0.251113 0.7548 0.4504 − − Mean dependent var 0.496032; S.D. dependent var 0.500481; McFadden R2 0.217410; Adjusted R2 0.022752; Log-likelihood 273.3825; Akaike criterion 682.7650; Schwarz criterion 969.9002; Hannan–Quinn 795.3980. −

References

1. Ellabban, O.; Abu-Rub, H. Smart grid customers’ acceptance and engagement: An overview. Renew. Sustain. Energy Rev. 2016, 65, 1285–1298. [CrossRef] 2. Biresselioglu, M.E.; Nilsen, M.; Demir, M.H.; Royrvik, J.; Koksvik, G. Examining the barriers and motivators affecting european decision makers in the development of smart and green energy technologies. J. Clean. Prod. 2018, 198, 417–429. [CrossRef] 3. Crispim, J.; Braz, J.; Castro, R.; Esteves J. Smart grids in the eu with smart regulation: Experiences from the UK, Italy and Portugal. Util. Policy 2014, 31, 85–93. [CrossRef] 4. Zhou, S.; Brown, M.A. Smart meter deployment in Europe: A comparative case study on the impacts of national policy schemes. J. Clean. Prod. 2017, 144, 22–32. [CrossRef] 5. Sial, A.; Singh, A.; Mahanti, A. Detecting anomalous energy consumption using contextual analysis of smart meter data. Wirel. Netw. 2019.[CrossRef] 6. Thomson, W.; Mahanti, A.; Gong, M. Uploader Motivations and Consumer Dynamics in the One-Click File Hosting Ecosystem. In Proceedings of the 2018 IEEE International Conference on Communications, (ICC), Kansas City, MO, USA, 20–24 May 2018; doi:10.1109/ICC.2018.8422248. [CrossRef] 7. Gans, W.; Alberini, A.; Longo, A. Smart meter devices and the effect of feedback on residential electricity consumption: Evidence from a natural experiment in Northern Ireland. Energy Econ. 2013, 36, 729–743. [CrossRef] 8. Beemsterboer, S.; Sengers, F.; Verbong, G.P.J. Smart grids or smart users? involving users in developing a low carbon electricity economy. Energy Policy 2013, 52, 115–125. 9. Burchell, K.; Rettie, R.; Roberts, T.C. Householder engagement with energy consumption feedback: The role of community action and communications. Energy Policy 2016, 88, 178–186. [CrossRef] 10. Schleich, J.; Faure, C.; Klobasa, M. Persistence of the effects of providing feedback alongside smart metering devices on household electricity demand. Energy Policy 2017, 107, 225–233. [CrossRef] 11. Krishnamutri, T.; Schwartz, D.; Davis, A.; Fischoff, B.; de Bruin, W.B.; Lave, L.; Wang, J. Preparing for smart grid technologies: A behavioral decision research approach to understanding consumer expectations about smart meters. Energy Policy 2012, 41, 790—797. [CrossRef] 12. Kabalci, Y. A survey on smart metering and smart grid communication. Renew. Sustain. Energy Rev. 2016, 57, 302–318. [CrossRef] 13. Avancini, D.B.; Rodriques, J.J.P.C.; Martins, S.G.B.; Rabelo, R.A.L.; Al-Mahtadi, J.; Solic, P. Energy meters evolution in smart grids: A review. J. Clean. Prod. 2019, 217, 702–715. [CrossRef] 14. Wolsink, M. Distributed generation for sustainable energy as a common pool resource: Social acceptance in rural setting of smart (micro-) grid conﬁgurations. In New Rural Spaces: Towards Renewable Energies, Multifunctional Farming, and Sustainable Tourism; Frantal, B., Martiant, S., Eds.; Publisher: Brno, Czech Republic, 2014; pp. 36–47. 15. Chen, C.; Xu, X.; Arpan, L. Between the technology acceptance model and sustainable energy technology acceptance model: Investigating smart meter acceptance in the United States. Energy Res. Soc. Sci. 2017, 25, 93–104. Energies 2020, 13, 732 25 of 27

16. Kahma, N.; Matschoss, K. The rejection of innovations? rethinking technology diffusion and the non-use of smart energy services in Finland. Energy Resour. Soc. Sci. 2017, 34, 27–36. [CrossRef] 17. Kowalska-Pyzalska, A.; Byrka, K. Determinants of the willingness to energy monitoring by residential consumers: A case study in the city of Wroclaw in Poland. Energies 2019, 12, 907. [CrossRef] 18. Nachreiner, M.; Mack, B.; Matthies, E.; Tampe-Mai, K. An analysis of smart metering information systems: A psychological model of self-regulated behavioral change. Energy Res. Soc. Sci. 2015, 9, 85–97. [CrossRef] 19. Droge, C.; Stanko, M.A.; Pollitte, W.A. Lead users and early adopters on the web: The role of new technology product blogs. J. Prod. Innov. Manag. 2010, 27, 66–82. [CrossRef] 20. Alsabbagh, M. Public perception toward residential solar panels in Bahrain. Energy Rep. 2019, 5, 253–261. [CrossRef] 21. Sohn, K.; Kwon, O. Technology acceptance theories and factors influencing artificial intelligence-based intelligent products. Telemat. Inform. 2020, 47, 101324. [CrossRef] 22. Park, C.; Kim, H.; Yong, T. Dynamic characteristics of smart grid technology acceptance. Energy Procedia 2017, 128, 187–193. [CrossRef] 23. Mani, S.; Dhingra, T. Diffusion of innovation model of consumer behaviour—Ideas to accelerate adoption of renewable energy sources by consumer communities in India. Renew. Energy 2012, 39, 162–165. [CrossRef] 24. Kowalska-Pyzalska, A. Do consumers want to pay for green electricity? A case study from Poland. Sustainability 2019, 11, 1310. [CrossRef] 25. Ozaki, R. Adopting sustainable innovation: What makes consumers sign up to green electricity? Bus. Strategy Environ. 2011, 20, 1–17. [CrossRef] 26. Gerpott, T.J.; Mahmudova, I. Determinants of green electricity adoption among residential customers in Germany. Int. J. Consum. Stud. 2010, 34, 464–473. [CrossRef] 27. Rogers, E.M. Diffusion of Innovations, 5th ed.; Free Press: New York, NY, USA, 2003. [CrossRef] 28. Peres, R.; Muller, E.; Mahajan, V. Innovation diffusion and new product growth models: A critical review and research directions. Int. J. Res. Mark. 2010, 27, 91–106. 29. Kowalska-Pyzalska, A. An empirical analysis of green electricity adoption among residential consumers in Poland. Sustainability 2018, 10, 2281. 30. Iqtiyanillham, N.; Hasanuzzaman, M.; Hosenuzzaman, M. European smart grid prospects, policies, and challenges. Renew. Sustain. Energy Rev. 2017, 67, 776–790. 31. Park, C.K.; Kim, H.J.; Kim, Y.S. A Study of Factors Enhancing Smart Grid Consumer Engagement. Energy Policy 2014, 7, 211–218. 32. Bugden, D.; Stedman, R. A synthetic view of acceptance and engagement with smart meters in the United States. Energy Res. Soc. Sci. 2019, 47, 137–145. 33. AFMERCADOS EMI. Turkey Smart Grid 2023 Vision and Strategy Roadmap Summary Report. 2014. Available online: yearwww.smartgridturkey.org (accessed on 10 November 2019). [CrossRef] 34. Chawla Y.; Kowalska-Pyzalska A. Public awareness and consumer acceptance of smart meters among Polish social media users. Energies 2019, 12, 2759. [CrossRef] 35. Claudy, M.C.; Michelsen, C.; O’Driscoll, A.; Mullen, M.R. Consumer awareness in the adoption of microgeneration technologies. An empirical investigation in the Republic of Ireland. Renew. Sustain. Energy Rev. 2010, 14, 2154–2160. [CrossRef] 36. Chawla, Y.; Kowalska-Pyzalska, A.; Skowronska-Szmer A. Perspectives of Smart Meters’ Roll-out in India: An Empirical Analysis of Consumers’ Awareness and Preferences; Working Paper, Hugo Steinhaus Center, Wroclaw University of Technology HSC Research Reports HSC/19/03; 2020. Available online: https://ideas/repec/org/ p/wuu/wpaper/hsc1903.html (accessed on 7 January 2020). [CrossRef] 37. Buchanan, K.; Russo, R.; Anderson B. Feeding back about eco-feedback: How do consumers use and respond to energy monitors? Energy Policy 2014, 73, 138–146. [CrossRef] 38. Fisher, C. Feedback on household electricity consumption: A tool for saving energy? Energy Effic. 2008, 1, 79–104. 39. Good, N.; Ellis, K.A.; Mancarella, P. Review and classification of barriers and enablers of demand response in the smart grid. Renew. Sustain. Energy Rev. 2017, 16, 57–72. [CrossRef] 40. Foulds, C.; Robison, R.A.V.; Macrorie, R. Energy monitoring as a practice: Investigating use of the imeasure online energy feedback tool. Energy Policy 2017, 104, 194–202. [CrossRef] Energies 2020, 13, 732 26 of 27

41. Ma, G.; Lin, J.; Li, N. Longitudinal assessment of the behavior-changing effect of app-based eco-feedback in residential buildings. Energy Build. 2018, 159, 486–494. 42. McMichael, M.; Shipworth, D. The value of social networks in the diffusion of energy-efficiency innovations in UK households. Energy Policy 2013, 53, 159–168. [CrossRef] 43. Chawla, Y.; Kowalska-Pyzalska, A.; Silveira, P. Marketing and Communication Channels for Diffusion of Smart Meters in Portugal; Working Paper (No. HSC/19/05); Hugo Steinhaus Center, Wroclaw University of Technology. Available online: https://ideas/repec/org/p/wuu/wpaper/hsc1905.html (accessed on 7 January 2020). [CrossRef] 44. Ravazi, R.; Gharipour, A.; Fleury, M.; Ikpe-Justice, A. Occupancy Detection of Residential Buildings Using Smart Meter Data: A Large-Scale Study. Energy Build. 2019, 183, 195–208. [CrossRef] 45. Hess, D.J. Smart meters and public acceptance: Comparative analysis and governance implications. Heal. Risk Soc. 2014, 16, 243–258. [CrossRef] 46. Bigerna, S.; Bollino, C.A.; Micheli, S. Socio-economic acceptability for smart grid development—A comprehensive review. J. Clean. Prod. 2016, 6131, 399–409. [CrossRef] 47. Gerpott, T.J.; Paukert, M. Determinnants of willingness to pay for smart meters: An empirical analysis of household customers in Germany. Energy Policy 2013, 61, 483–495. [CrossRef] 48. Kowalska-Pyzalska, A. What makes consumers adopt to innovative energy sources in the energy market? A review of incentives and barriers. Renew. Sustain. Energy Rev. 2018, 82, 3570–3581. 49. Podgornik, A.; Sucic, B.; Blazic, B. Effects of customized consumption feedback on energy efficient behavior in low-income households. J. Clean. Prod. 2016, 130, 25–34. 50. Weron, T.; Kowalska-Pyzalska, A.; Weron, R. The role of educational trainings in the diffusion of smart metering platforms: An agent-based modeling approach. Phys. A Stat. Mech. Its Appl. 2018, 505, 591–600. [CrossRef] 51. Raimi, K.T.; Carrico, A.R. Understanding and beliefs about smart energy technology. Energy Res. Soc. Sci. 2016, 12, 68–74. [CrossRef] 52. Khan, M.Z.; Miankhel, A.K.; Nawaz, A. Information & communication technology and ‘individual’: Prospects & concerns. Glob. J. Comput. Sci. Technol. 2013, 13, 1–10 53. Ma, L.; Lee, C.S.; Goh, D.H.L. Understanding news sharing in social media: An explanation from the diffusion of innovation theory. Online Inf. Rev. 2014, 38, 598–615. [CrossRef] 54. Colak, I.; Bayindir, R.; Fulli, G.; Tekin, I.; Demirtas, K.; Covrig, C.-F. Smart grid opportunities and applications in Turkey. Renew. Sustain. Energy Rev. 2014, 33, 344–352. [CrossRef] 55. Elma O.; Selamo˘gullari, U.S. An overview of demand response applications under smart grid concept. In Proceedings of the 4th International Conference on Clectrical and Electronic Engineering (ICEEE), Ankara, Turkey, 8–10 April 2017; pp. 104–107. [CrossRef] 56. Atasoy, T.; Akinc, H.E.; Erol, E.; Ercin, O.; Gurec, O.; Benli O. Challenges and opportunities towards smart grid in Turkey: Distribution system operator perspective. In Proceedings of the IEEE PES Innovative Smart Grid Technologies Rope, Istanbul, Turkey, 12–15 October 2014; pp. 1–6. [CrossRef] 57. Gungor, V.C.; Sahin, D.; Kocak, T.; Ergüt, S. Smart Grid Communications and Networking; Türk Telekom Technic Report, 11316-01; Türk Telekom: Altında˘g,Turkey, 2011. 58. Demirci, T.; Kalaycıoglu, A.; Küçük, D.; Salor, Ö.; Güder, M.; Pakhuylu, S.; Atalık, T.; Inan, T.; Çadırcı, I.; Akkaya, Y.; et al. Nationwide real-time monitoring system for electrical quantities and power quality of the electricity transmission system. IET Gener. Transm. Distrib. 2011, 5, 540–550. [CrossRef] 59. Tumbaz, M.N.M.; Mogulkoc, H.T. Profiling energy efficiency tendency: A case for Turkish households. Energy Policy 2018, 119, 441–448. [CrossRef] 60. Malinen, S.; Ojala, J. Maintaining the instant connection—Social media practices of smartphone Users. In From Research to Practice in the Design of Cooperative Systems: Results and Open Challenges; Springer: London, UK, 2012; pp. 197–211. [CrossRef] 61. Ahmad, W.M.; Mourshed, M.; Mundow, D.; Sisinni, M.; Rezgui, Y. Building energy metering and environmental monitoring–a state-of-the-art review and directions for future research. Energy Build. 2016, 120, 85–102. 62. Barringer, F. New Electricity Meters Stir Fears. The New York Times, 2011; p. 30. Available online: https://www. nytimes.com/2011/01/31/science/earth/31meters.html (accessed on 20 December 2019). Energies 2020, 13, 732 27 of 27

63. Sullivan, C.; Kahn, D. Smart Grid: Calif. Agency Mulls ‘opt out’or Wired Substitutes as Fallout Persists. E&E News, January 2011. Available online: https://www.eenews.net/stories/1059944053 (accessed on 7 January 2020). 64. Opris, I.; Caracasian, L. The relation between smart meters and electricity consumers. In Proceedings of the 2013 12th International Conference on Environment and Electrical Engineering, Wroclaw, Poland, 5–8 May 2013; pp. 325–329. [CrossRef] 65. Bauer, P. A sequential elimination procedure for choosing the best population (s) based on multiple testing. J. Stat. Plan. Inference 1989, 21, 245–252. [CrossRef]

c 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access

article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). Paper 2

Y. Chawla, A. Kowalska-Pyzalska, P. Silveira Marketing and communications channels for diffusion of smart meters in Portugal Marketing and communications channels for diﬀusion of electricity smart meters in Portugal

Yash Chawlaa, Anna Kowalska-Pyzalskaa,, Paulo Duarte Silveirab aDepartment of Operations Research, Faculty of Computer Science and Management, Wrocław University of Science and Technology, 50-370 Wrocław, Poland bInstituto Polit´ecnicode Set´ubal,College of Business and Administration, Department of Marketing and Logistics, Portugal

Abstract National roll-outs of electricity smart meters (ESM) have been undertaken in most of the Euro- pean countries. The exchange of traditional meters into smart ones is a part of power system transmission into so called smart grids. In these smart grids, the communication and sharing of information happens in real-time and all market players, such as energy suppliers, sellers and consumers, play an important role. As the literature reveals, the successful deployment of ESM requires consumers’ awareness and engagement. That is why, within this paper, we investigate the impact of consumers’ knowledge on what ESM is, as well as the role of marketing platforms: both traditional (i.e. TV or radio) and modern ones (i.e. social media) in ESM diffusion. Based on the on-line survey conducted in Portugal (N=518), we provide some policy and practical recommendations for energy companies and local authorities regarding the effective usage of marketing platforms and content. Keywords: smart metering; knowledge; marketing platforms; communication channels; social media; diffusion; on-line questionnaire

1. Introduction The recent industrial revolution, Industry 4.0, has led to the introduction of smart technologies in various ﬁelds of our lives. It is also present in the energy sector, where, for several years, practi- tioners and researchers have been investigating various intelligent technologies in the generation, distribution and consumption of energy. They aim to combine modern advanced communication and information technologies to enhance and optimize the interaction between all players of the electricity market: producers, suppliers, sellers, and consumers. In coming times, electricity will become a technology that is visible and would require attention and decision-making from the consumers (Kowalski and Matusiak, 2019). This new approach, called smart grid (SG), includes the broad implementation of electricity smart meters (ESM), which are electronic devices that enable

Email addresses: [email protected] (Yash Chawla), [email protected] (Anna Kowalska-Pyzalska ), [email protected] (Paulo Duarte Silveira) Preprint submitted to Telematics and Informatics This version: March 13, 2020 automatic collection of consumers’ energy consumption data and share this information with the electricity supplier for monitoring and billing purposes, as well as with the consumers themselves. This demand side response management can help in energy saving (Soroczynski´ and Szkutnik, 2015). The broad implementation of ESM in Europe is induced among others by the EU direc- tives concerning common rules for the internal market for electricity and gas (2009/72/EC and 2009/73/EC) and the EU directive on energy efficiency (2012/27/EC). These regulations require EU Member States to ensure the implementation of ESM in order to enable active participation of consumers in the energy market. According to Biresselioglu et al. (2018); Zhou and Brown (2017), ESM would allow consumers to take advantage of the benefits of the progressive digitalisation of the energy market via several different functionalities. For example, consumers would be able to access dynamic electricity tariffs, such as real-time tariffs, which belong to one of the most important demand side management/ demand response tools (DSM/DR). The regulatory decision regarding ESM implementation is done on a national level based on the assessment of long-term costs and benefits. If the assessment is positive, then at least 80% of households should be equipped with smart metering systems by 2020 (Crispim et al., 2014). It is predicted that smart metering and smart grid roll-outs can reduce emissions in the EU and annual household energy consumption by even 9%1. Although an ESM roll-out has indisputable advantages, its implementation is in different stages, depending on the location (Zhou and Brown, 2017; Avancini et al., 2019). Whereas, in many of the European countries, such as Denmark, Sweden, Finland, Estonia or Spain, the ESM roll-out is already finalized, and in countries such as Norway, Italy or UK, it is at an advanced stage (Sovacool et al., 2017; Zhou and Brown, 2017), other countries, such as Germany, the Czech Republic, Greece and Ireland, represent a lower commitment level towards ESM deployment (European Commission DG Energy, 2018). In those countries, the governments have usually not formally decided to have a national ESM roll-out, mainly because of negative results of cost-benefit analysis (CBA). According to the Agency for the Cooperation of Energy Regulators (ACER), in late 2018, only 37% of EU consumers were equipped with ESM, which is a very weak result. Portugal belongs to the group of countries, where a nationwide roll-out of ESM has been recently approved by the government, but the Portuguese government has not decided to take any formal commitment for the target to reach in terms of smart metering deployment (European Commission DG Energy, 2018). Currently, only one third of Portuguese citizens are equipped with ESM. At the same time, there are many pilot projects underway across the country. The implementation of those projects, together with the recent energy market liberalization, has increased the awareness level of the consumers and their attention to energy savings (Ghazvini et al., 2019). Joao Torres - the CEO of EDP Distribution, the main distribution system operator in Portugal, clearly stated in January 2019 that: “(electricity) smart metering is really important because it is the first step to get consumer engagement”2. Taking all of that into consideration, it can be pro-

1https://ec.europa.eu/energy/en/topics/markets-and-consumers/smart-grids-and-meters/overview (accessed September, 6th 2019) 2for more details see https://www.euractiv.com/section/energy/news/smart-meter-woes-hold-back-digitalisation- of-eu-power-sector (accessed September, 10th 2019) 2 jected that the Portuguese power system is going to experience great changes in the coming years because of its further digitalisation. It may also have a significant influence on the consumers’ electricity consumption unless they stay uninterested and disengaged. The current digitalisation of the Portuguese power system was one of the reasons that prompted us to carry out this study in Portugal. Moreover, there is a gap observed in the literature about marketing tools used for the diffusion of information and awareness regarding ESM, especially among social media users. As already revealed in similar studies conducted in Poland (Chawla and Kowalska-Pyzalska, 2019), in India (Chawla et al., 2020), or in Indonesia (Chawla et al., 2019), social media are often neither used to promote energy efficiency among consumers nor are used to explain the benefits of the smart grid approach to the power system. Secondly, there is a lack of effective communication between electricity distribution companies and their customers in terms of sharing the information about the advantages and opportunities connected with ESM roll-outs. In order to enhance the ESM deployment, some regulatory support is needed. But this is not enough. The society will benefit from this enrollment, only if consumers learn how to use the information provided by ESM. In particular, how to monitor energy consumption based on the information provided by the enabling technology, such as smart metering information systems (platforms) or in-home displays (Kowalska-Pyzalska and Byrka, 2019; Foulds et al., 2017; Ma et al., 2018; Schleich et al., 2017). The first step in this process is connected with the increase of consumers’ awareness and engagement regarding energy efficiency issues (Verbong et al., 2013; Ellabban and Abu-Rub, 2016; Burchell et al., 2016; Akroush et al., 2019; Gans et al., 2013). Within our survey we want to explore how to reach the residential consumers, represented by the social media users, to raise consumers’ awareness of ESM. We focus on social media users, who, according to the literature (Droge et al., 2010; Kumar Verma et al., 2017; Barrios-O’Neill and Schuitema, 2016; Bentoa et al., 2018) are a social group, which is perceived to be more open- minded towards innovations and IT-based solutions than the rest of the society. Further, we believe that reasonable usage of traditional and modern communication channels and marketing platforms could be very useful in ESM promotion and diffusion in the energy market. Hence, the aim of this paper is threefold. First, we want to investigate the attitudes, preferences and fears among social media users towards ESM. Second, we want to verify which socio-economic and attitudinal variables influence the knowledge about ESM. Third, we want to explore the communication channels, marketing platforms and content that could be used by the energy companies and the government to enhance the outcomes of ESM roll-outs. The survey is based on the example of Portugal - the country where the ESM is not finalized yet and still needs initiative to ensure the effective transition of the power system into the smart grid. This paper contributes, not only to the scientific literature through the findings and analysis of this study, but it also offers practical recommendations regarding marketing platforms and content, which can be useful for energy companies and local authorities for the facilitation of ESM diffusion in Portugal and in the other countries. The structure of the article is as follows. In Section 2, the ESM deployment, in terms of political will and consumers’ engagement, with respect to the Portuguese energy market, is elaborated. Next, in Section 3 we discuss the marketing trends in the context of the energy market in general and ESM diffusion in particular. In Section 4, the results of the study are presented and discussed. 3 This Section contains the methods of data collection, framework of the survey, and the analysis of the data. Finally, in Section 5 we provide a broad discussion and elaborate on the marketing platforms and their content dividing them into traditional platforms and those based on social media. The conclusions of this study are followed by the limitations and the future scope.

2. ESM deployment in Europe: literature review Smart and green technologies are becoming increasingly popular all over the world. There are many reasons that induce the broader and faster diffusion of those goods in the energy markets. Let us just mention the urgent need of societal and energy transitions of the power systems from the traditional ones - based on fossil fuels with passive consumers into the smart grids, where modern information and communication technologies play a great role in sharing the information in real-time between all market players: energy producers, sellers, distributors and end-users, that is, consumers. Nowadays, consumers are encouraged to play an active role in the energy market. Within the smart or micro grid approach, consumers may become producers of electricity and heat, by installing their own energy generators in their households. The introduction of electrical smart meters, combined with other enabling technologies, such as DSM/DR tools or smart metering information platforms (SMP), give consumers the access and better control over their energy consumption (Kowalska-Pyzalska, 2019; Chawla and Kowalska-Pyzalska, 2019; Biresselioglu et al., 2018; Bellido et al., 2018). Without any doubt, many of the EU countries are world leaders in the diffusion of those innovative, smart and green technologies in the energy markets. There are many initiatives, such as the European Electricity Grid Initiative (EEGI) and European Strategic Energy Technology Plan (SETplan), that encourage the sustainable transition of the power systems in terms of its economic, ecological, technical, and social aspects in the coming decades (Biresselioglu et al., 2018). The literature broadly emphasizes that implementation of the smart grid approach is, not only a matter of modernization and digitalization of the electricity grid, but firstly it requires some new business models and practices and some appropriate legal regulations that would enable and motivate some behavioral change on the consumers end and, hence, would lead to social acceptance of this revolution (Chawla and Kowalska-Pyzalska, 2019; Kowalska-Pyzalska, 2018; Biresselioglu et al., 2018). Currently, one of the main reasons for the slow ESM diffusion is the consumers’ reluctance and lack of engagement, but also the lack of legal regulations and arrangements provided by the national authorities. As emphasized by the CEO of the Portuguese EDP - Joao Torres, although reminding that consumer empowerment and dynamic pricing of electricity were among the key objectives of EU clean energy laws adopted in 2018, the speed of ESM deployment is too slow. ESM and dynamic electricity tariffs are very important for households as they enable consumers to take active control over their electricity consumption. The advantages of smart metering for the future use of DSM/DR tools, such as for example dynamic tariffs are emphasized in the literature, see for example (Doostizadeh and Ghasemi, 2012; Aghaei and Alizadeh, 2013). Although there are countries in the EU that have already finalized their ESM roll-out3, in many of the EU countries, the

3Sweden is the uncontested EU leader in smart metering penetration (100%), followed by Finland (99%), Estonia 4 billing is still done on quarterly or even yearly forecasts rather than on the basis of real electricity consumption. Installation of ESM would allow consumers better control over their electricity bills and could encourage them to monitor and save on energy consumption (Kowalska-Pyzalska and Byrka, 2019). According to ACER, many of the consumers are still very sceptical and reluctant to adopting ESM. For example, in some of the French cities, the residents may legally refuse entry to the installations teams of the local distribution system operator. There are still EU countries, including Germany, Croatia, Cyprus, the Czech Republic, Greece and Ireland, where the national governments are sceptical about ESM deployment. The European Commission does not encourage the reluctant countries to change their minds and decide for ESM enrollment by means of some additional measures or regulations, even if their decline will not allow to achieve some ambitious goals regarding increases in energy efficiency. At the same time, the European electricity industry is in favor of further digitalisation as they observe great benefits due to smart metering. According to Eurelectric, by 2030, more than half of all electricity in Europe is expected to come from renewable energy, including solar PV systems installed on people’s rooftops. More than 40 million electric cars are also expected to appear on the European roads by that date, according to EU estimates. Meanwhile, the number of electric heat pumps, batteries and other grid-connected smart devices are expected to rise steeply. It is predicted that all this equipment will require smart meters to function properly4. It must be, however, underlined that the digitalisation and rapid increase of grid-connected smart devices is not enough to provide sustainable development of the power system. Without regulatory support, ensuring interoperability of ESM systems and their concentration on the end-users of electricity and connectivity with other consumer energy management systems, the diffusion of ESM will not be effective and successful (Bellido et al., 2018; Biresselioglu et al., 2018; Zhou and Brown, 2017; Avancini et al., 2019; Nizetic et al., 2019; Park et al., 2018).

2.1. ESM in Portuguese Energy Market The Portuguese energy market has been recently liberalized and, since 2017, residential consumers may choose the energy company from which they buy electricity (Ghazvini et al., 2019; Miguel et al., 2018). Currently there are two main national legislation regulating electricity and gas smart metering deployment in Portugal: ‘Decreto-Lei n 215-A/2012’ (October 8) and ‘Decreto-Lei n 231/2012’ (October 26), and ‘Portaria n 231/2013’ (July 22). A few years ago Portugal belonged to the EU countries which have not decided in favour of a large-scale ESM roll-out and 80% target penetration rate by 2020, as recommended by the European Commission (Lopes et al., 2016). One of the main reasons why Portugal did not want to join the European smart meters deployment, was the fact that the initial cost-beneﬁt analysis (CBA) in 2013 has resulted as non-conclusive5. A few years later, in 2018, the CBA results changed into positive giving a green light for ESM deployment (European Commission DG Energy, 2018). In particular the estimated normalised costs per

(98.2%), Spain (91.7%) and Denmark (80%). 4https://www.euractiv.com/section/energy/news/smart-meter-woes-hold-back-digitalisation-of-eu-power-sector (accessed, September, 10th 2019) 5According to EUs’ Joint Research Centre in 2017 in seven countries (Belgium, Czech Republic, Germany, Latvia, Lithuania, Portugal and Slovakia), the CBAs for large-scale roll-out by 2020 were negative or inconclusive. 5 a smart metering point were equal to 333.30 Euro/meter in 2018 (versus 99 Euro/meter in 2013) and normalised benefits 466.70 Euro/meter (versus 202 Euro/meter in 2013). Currently Portugal, together with Latvia and Slovenia belong to the group of countries that have not decided for taking any formal commitment for the target to reach in terms of smart metering deployment. As the final report about Smart Metering Deployment among EU-28 Member States summarizes (European Commission DG Energy, 2018), in case of Portugal, a simultaneous decrease in costs and increase in benefits has triggered a wave of new commitment for smart metering, even though the full roll out will require more than the business as usual to be completed within the next 10 years. The approximate ESM penetration in Portugal has already reached around 25%, in comparison to 10% in 2017 (European Commission DG Energy, 2018) . If such a fast rate of change continues, Portugal will be able to achieve even 80% rate of ESM penetration in 2022-2023 (European Commission DG Energy, 2018). Portugal can show off with its several pilot programs that have already been carried out in various Portuguese regions and cities. One of the most well-known large-scale smart grids projects is called InovGrid. It has been run by EDP - Portuguese transmission and distribution system operator. This is a very broad project which includes several issues, such as: support of dispersed generation, on-demand management, customer self-service, renewable energy sources, electric vehicles, cyber security, data privacy and others. Additionally, as a part of the project, in a small Portuguese city Evora´ in 2009, approximately 35,000 ESM were installed. Since then, this number was expected to climb to six million by the end of 20196. Today, InovGrid is delivering impressive results, including improved energy efficiency with significant consumption reductions, faster response times and improved service quality, increased knowledge of customer and grid behaviors, and easier integration of renewable energy sources and other emerging technologies. Its scale and success led to its selection by the European Union Commission and Eurelectric as a leading smart grids reference project in Europe. Energy box is an example of another, recently conducted, project developing a demand responsive energy management system to be used to control, manage and optimize smart grid technologies and home electricity use7. Both innovative projects share the knowledge about smart grids and smart metering among consumers.

3. Marketing in the context of energy market Marketing any product or service presents a number of challenges, especially in the current era, where going digital is the new normal. More and more businesses prefer digital marketing, especially social media, in place of traditional marketing, such as direct sales, TV, radio, mail, print advertisements in newspapers or magazines, and printed materials like billboards, posters, catalogs or brochures (Das and Lall, 2016). Globally digital media has emerged as a cost effective medium which caters to the marketing and strategy of firms, especially in terms of engaging customers, building and managing customer relations and communication in a wide variety of fields (Filo et al., 2015; Saxena and Khanna, 2013). It provides a two-way or multi-way communication

6for more details see https://www.iea-isgan.org/cee-case09-portugal/ (accessed 7th July 2019) 7 for more details see https://www.uc.pt/en/org/inescc/Projects/projects/EnergyBox (accessed 23rd August 2019) 6 channel, as compared to only one way communication through the traditional marketing channels (Hanna et al., 2011; Abidin et al., 2010), which enables businesses to listen to consumers. Digital media also plays a very important role for increasing customer awareness and knowledge. They are used to share customer experience, to get to know their preferences and to accept certain products or services (Duffett, 2015). Understanding consumer expectations and preferences helps businesses to adopt appropriate strategies to enhance the diffusion of products or services in the market (Fiore et al., 2017). The majority of the studies in the literature state that digital media, especially social media, increases the impact and prevalence of word of mouth as compared to traditional mediums (Alalwan et al., 2017). This is due to the ease with which the consumers are able to share their own pleasant or unpleasant experiences, thoughts, as well as recommendations about brands, products or services to a large number of users on social media platforms (Leong et al., 2019; Hudson et al., 2016; Priyanka, 2013). This results in higher involvement of the customer in the market. Although digital media is highly effective, specialists suggest that businesses should not abandon traditional methods and to try to blend digital and traditional channels to reach their customers (Todor, 2016). For energy markets, the importance of marketing and understanding the prospective consumer for diffusion of new technologies, as well as sustainable development, was highlighted by the researchers since the later part of the 20th century (Nakarado, 1996). This is even more important because of the changing dynamics of the electricity market (Parag and Sovacool, 2016) with the integration of Internet of Things (IoT). In energy markets, a social marketing mix of both traditional and digital media is prevalent (Sheau-Ting et al., 2013), with more focus on the traditional marketing tools. Although, in recent years, more energy companies have started increasing their presence on digital media, especially social media, in order to increase their reach to the consumers, but still it has not reached the critical masses. Barrios-O’Neill and Schuitema (2016) found that using strategic social marketing mix through diverse, networked platforms, which is meaningful in the contemporary social and technological context, is most likely to influence consumer behaviours. Engaging the customers also increases their awareness about new technologies, which has a significant effect on consumers’ purchase intentions towards more energy efficient devices (Akroush et al., 2019; Heikkinen et al., 2012). This type of interaction and consumer engagement is likely to become an integral part of future energy delivery systems, due to which digital marketing in the energy sector becomes an important area of investigation (Barrios-O’Neill and Schuitema, 2016). There have been a few studies regarding marketing strategies and tools for the energy sector. Bogdal (2013) studied communication management in the public sector for energy markets, where he highlighted the need for updating strategies and tools used for diffusing information to the consumers. Streimikiene and Vveinhardt (2015) found that a community based marketing approach is effective for saving energy and one of the hurdles for energy saving includes a lack of information with the consumers. Hille et al. (2019) studied various programs aimed towards promoting electricity saving and proposed that the marketing and communication programs should be tailored for specific target groups. Gong et al. (2019) explored various marketing strategies adopted by the energy companies and concluded that there is a need for energy companies to put more effort in marketing as well as to broaden the communication channels. To cater to these challenges dis- covered by the researchers, the understanding of various consumer preferences and habits would prove to be beneficial in devising an effective marketing strategy and content. 7 3.1. Marketing tools used by Portuguese energy suppliers in ESM roll-out The marketing tools used in the energy market can be discussed with the example of the ESM roll-out in Portugal. As already mentioned, the ESM roll-out has been conducted by the Portuguese main distribution system operator - EDP. The company had highlighted the importance of the role of customer engagement on energy efficiency and its impact on the value of a pilot project, Inovgrid, conducted in Evora, Portugal8. EDP has used a mix of initiatives for raising awareness about this project. First of all, they reached out to key stakeholders and groups of customers through activities like the Inovcity showroom, the Energy Bus, the organization of conferences and events, articles in the local press, several public sessions and so on. However, these initiatives had covered only a part of the population. Before making the upgrade to ESM in each household, EDP notifies the customer through a letter. Figure 1 shows the example of a letter sent by the energy supplier - EDP to the Portuguese customer9.

Figure 1: The example of the content of the letter sent by the energy provider to the customer.

Although EDP gives detailed information on their website about smart meters and smart grids, the letter does not mention these terminologies. It consists of a simple notiﬁcation of upgrading the electric meter to a more technologically advanced one. It lacks the primary information, such

8for details see https://www.iea-isgan.org/cee-case09-portugal/ 9Originally, the letter was in the Portuguese language, which has been translated into English for representation, ensuring the contents and meaning remains the same. Apart from the website address, all the contact information has been made anonymous to conserve the privacy of the user 8 as the terminologies, and it does not explain the benefits and opportunities available via ESM. The letter does not answer any of the following questions: What is a difference between a smart and traditional electricity meter? How is the smart meter more technologically advanced? Is the implementation of ESM obligatory? Who pays for it? Is the cost of replacement borne by the company or is there government funding? What is the timeline for the change of the meters? Will the installation of ESM have any effect on the current billing methods? Will ESM have any effect one’s privacy and safety of the personal data? Hence, only the consumers who made an effort to log on to EDP’s website to get more information, would have acquainted themselves with ESM. For the rest, it would be business as usual. By means of this study, we want to highlight consumer preferences for receiving such information and also the preferred communication channels in Portugal, which would be a strong basis for the government or energy companies to prepare their marketing content, as well as to choose the proper communication channels.

4. Study, results and discussion 4.1. Data collection and the sample An empirical quantitative study was conducted to gather primary data, via a self-administrated online questionnaire directed to a target population of individuals living in Portugal in May - June 2019 (regardless of their nationality), responsible or co-responsible for the household energy decisions. Two sampling techniques were used in sequenced phases. First, a convenience sample was selected and contacted. Respondents were recruited via social media general posts (via Face- book and Linkedin) and private messages (via Linkedin and WhatsApp), as well as direct email messages. During the second phase, snowball sampling was used, requiring the respondents of the ﬁrst phase to forward the study invitation. The techniques used to recruit the respondents didn’t allow us to control the type of people who responded to the questionnaire. We were able to verify that the respondents belonged to our target audience, social media users in Portugal above the age of 18, based on their responses to usage of social media channels and age variables. The user sessions on the web page that hosted the questionnaire was tracked by Google Analytics (GA), to know the response rate. This did not aﬀect the anonymity of the respondents, as the IP addresses of the respondents were not available to the authors through GA. 932 user sessions were recorded during the period of dissemination of the questionnaire, resulting in 518 valid records being collected. The average time spent by users who submitted the completed questionnaire was 5 minutes and 37 seconds.

4.2. Framework of the survey The questionnaire was divided into two parts: the first part being dedicated to all respondents and second directed only to those respondents who knew what electricity smart meter was (”Yes” to question K1). In the first part, all of the respondents were asked about their demographic attributes, such as gender, age and household’s income (D1-D10). Then, they were asked about their belongings (B1-B6), social media platforms used by them on a regular basis (S01-S08) and the most common sources through which they get information about electricity and the energy market (S1-S14). They were also asked about their consumers’ preferences (P1-P6) and concerns (F1, F4). They were inquired about the potential usage of renewable energy sources in their 9 households (R1) and whether they monitor their energy usage (A1). Finally their willingness to install ESM under various conditions (De De , except De ) were investigated. 1 − 8 5 In the second part of the questionnaire, the respondents who declared to know what ESM was, were additionally asked about their information sources for ESM (I1, I2, I31-I45), preferences regarding the government’s role in ESM roll out (G1-G3), concerns about ESM (F2,F3), social influence in the context of ESM (W1), knowledge whether they already have ESM installed or/and their willingness to have it installed in the future (K2-K4, X1), and, finally, their willingness to install ESM if they had to pay for it (De5). The variables included in the survey were motivated by the literature and similar studies regarding consumers’ acceptance and preferences towards ESM and other enabling technologies (see, for example (Gerpott and Paukert, 2013; Krishnamutri et al., 2012; Kowalska-Pyzalska and Byrka, 2019; Chawla and Kowalska-Pyzalska, 2019; Kahma and Matschoss, 2017; Paetz et al., 2012)). According to the literature, apart from social and economic attributes, consumers’ knowledge, awareness and preferences, as well as social influence, may play a role in ESM diffusion. Figure 2 presents the survey’s framework. The definitions of the variables and their coding are described in Table A.2.

Figure 2: The framework of the survey (N=518).

10 4.3. Initial analysis of the data set, preferences, attitudes and fears regarding ESM The demographic variables are presented in Table 1. The majority of respondents were male, middle-aged, married, with a bachelor degree completed, working in the private sector and living in a small or a big city. Our data can be considered as representative, as the demographics are similar to that of the social media users in Portugal10, which were the target participants in this study. Moreover, 65% of Portuguese population is active on social media, as of 2019. Hence the data is also representative for majority of Portuguese population in general. Most of the respondents possessed some basic smart devices, such as a smart TV (64%) or other home appliances connected to Internet (70%). 97.7% of respondents had a wi-ﬁ/ Internet connection at home. Around 40% of respondents declared to monitor their energy consumption. Only 12.5% had some renewable energy sources installed at their household.

Table 1: Frequencies of the demographic variables (D1-D10), N = 518

Variable Frequency Gender (D1) female 47.9% male 52.1% Age (D2) 18-25 years old 11.5% 26-35 years old 19.8% 36-45 years old 37.8% 46-55 years old 24.0% over 56 years old 7.3% Martial status (D3) single 32.2% married 52.2% divorced/separated 8.8 % in a relationship 6.4% widowed 0.4% Education (D4) no formal education 0.4% high school pass 6.6% bachelor completed 64.4% masters completed 18.2% PhD completed 10.4% Occupation/ Employment (D5) private sector 61.6% public sector 21.7% student in college/ university 4.4% others 12.3% Household’s income (in Euro per month) (D6) less than 830 3.9% 831 to 1125 7.3% 1126 - 1580 11.8% 1581 to 2290 13.1% 2291 to 2700 7.5% 2701 to 3300 13.3% 3301 to 4165 13.7% 4166 to 6156 12.4% more than 6166 11.9% prefer not to say 5% Electricity bill (in Euro per month) (D7) 0 to 25 7.1% Continued on next page

10for more details see: https://datareportal.com/reports/digital-2019-portugal 11 Table 1 – continued from previous page Variable Frequency 26 to 50 29.7% 51 to 76 30.5% 75 to 100 18.1% more than 100 11.6% prefer not to say 3.0% Household size (D8) M=2.75, SD=1.29 (where the integer number indicates the number of family members) Number of children (D81) M=1.01, SD=1.05 (where the integer number indicates the number of family members) Type of a house (D9) apartment/ flat (in a 4 stored building) 37.3% apartment/ flat (in a more than 4 stored building) 34.2% house (only ground floor) 6.9% house (ground and upper floor) 21.6% Place of a living (D10) village 22.6% city 75.1% others 2.3%

The distribution of the respondents’ preferences (P1-P6) and concerns related to data privacy (F1) and potential stress caused by fluctuations of electricity prices (F4) are presented in the Figure 3. Generally, respondents would have liked to know how to use electricity in a more efficient way. The majority of them (85%) would have liked to have real time information about their energy consumption. They believed that having access to such information would allow them to reduce their energy consumption. More than 70% of the respondents would have liked to have the possibility to control their electricity supply through mobile applications. 73.2% were interested in having a dynamic electricity tariff (e.g. real-time tariff) which would allow them to shift the demand when the electricity is cheaper.

4.4. Logit model of having knowledge about ESM To evaluate the impact of the socio-economic and attitudinal variables on the knowledge what ESM is (K1), a binary logistic regression model has been used. For this reason a binary variable Yi is constructed, which takes two values: Yi = 1 when an i-th individual reports to have knowledge what ESM is (K1=1; N1=181) and Yi = 0 when a respondent does not know (K1=0; N2=205) or is not sure what ESM is (K1=0.5; N3=128)11 The logistic regression model enables to condition the probability of having knowledge what ESM is on a set of exogenous variables. The model is aimed to reveal and estimate, which variables have an impact of having knowledge of ESM, which in turn may be signiﬁcant for deciding about marketing tools and communication channels by the energy companies. A probability of having knowledge about ESM, is assumed to depend on a set of variables, Xi, which includes a constant, demographics (D1-D10), belongings (B1-B6, R1), social media platforms (S01-S08), sources of information about the electricity market (S1- S14), energy monitoring behavior (A1), preferences and fears towards ESM (P1-P6, F1, F4) and willingness to install ESM under various conditions (D 1 D 8, excluding D 5). The model takes e − e e the following form:

11The total number of N1+N2+N3=514, as 4 responses were unclear and were not included in the further analysis. 12 Figure 3: and concerns regarding electricity (N=518).

exiβ Prob(Yi = 1) = (1) (1 + exiβ) where β is a vector of the model coefficients. In order to limit the number of the insignificant variables, we have used the stepwise selection method, as in the works of (Ntanos et al., 2018; Kowalska-Pyzalska, 2019). Table 4.4 presents the final model with its coefficients, their standard deviations and marginal effects. The latter can be used to approximate the change in probability of having knowledge what ESM is by an increase of the corresponding variable by one. The final model prediction accuracy is quite high, as even 72% of the respondents were correctly classified by the model in terms of having or not having knowledge about ESM. The model Log-likelihood ratio equals 590.01, Nagelkerke R square: 0.17, and Chi-square 66.81(9) with p=0.000, indicating that the model is well specified.

The ﬁnal logistic regression model for the estimation of Yi.

13 Coeﬃcient (β) Standard error Marginal eﬀects const -0,174 0,634 0,840 D1 0,793 ∗∗ 0,206 0,453 − ∗ − D2 0,253∗∗ 0,098 1,288 D4 0,281∗ 0,136 1,324 D9 0,197 0,091 0,821 − ∗ S11 0,555∗∗ 0,210 1,742 A1 0,445∗∗ 0,206 1,560 P1 0,881∗∗ 0,307 0,415 P6 0,728∗∗ 0,294 0,483 ∗ De4 0,906∗ 0,317 2,475

Note: The statistical significance of the results is coded as follows: ***p < 0.001, **p < 0.01, *p < 0.05 (two-tailed test). In the final model all the remaining explanatory variables: gender (D1), age (D2), education (D4), type of residence (D9), as well as information about the electricity market directly from the energy companies (S11), monitoring of energy usage (A1) and the desire to get more information about one’s electricity use (P1) and to have dynamic electricity tariff (P6), as well as the willingness to install ESM if a representative of the energy company provides information about the advantages of ESM (De4) are statistically significant. The results indicate that the male respondents, living rather in apartments or flats than in houses, are more probable to have knowledge what ESM is. Also the increase of age (by one category e.g. from 18-25 years old to 26-35 years old) and the increase of education, leaving all the other explanatory variables unchanged, increases the probability of having knowledge a lot. Knowledge about ESM is also positively influenced by regular monitoring of energy and collecting the information about electricity market from the energy company. The probability of having knowledge increases also significantly if the representative of the energy supplier would visit the respondent and present the benefits of having ESM. Finally, the probability of having knowledge is positively influenced by the desire to learn more about the one’s electricity consumption and the willingness to have dynamic electricity tariffs which may lead to decrease of energy consumption.

4.5. Marketing Platforms and Content To share the information about electricity in general, and smart metering or DSM/DR tools in particular, energy suppliers need to use some marketing platforms. In our study, we examined not only traditional marketing platforms, such as TV or radio (S1-S3 / I31-I33), but also social media, such as Facebook or LinkedIn (variables S5-S10 /I35-I40), social inﬂuence (S4 / I34), information from energy companies and oﬃcial governmental portals (S11-S12 / I41-I42), workshops and seminars (S13 / I43), and telephone or text messages (S14 / I44). First, we examined how often those platforms were used (see Figures 4 and 5). The upper panel of Figure 4 compares the distribution of the usage of social media platforms in a daily routine between those respondents who declared being familiar with ESM (K1=1; N=181) with those who do not know or are not sure what ESM is (K1=0 and K=0.5, N=333). In both groups most of the respondents admitted to using social media platforms regularly, such as: Face- book, Facebook Messenger, LinkedIn, WhatsApp and Instagram. Youtube, Twitter and SnapChat were used more rarely. 14 Figure 4: Upper panel: Usage of various social media platforms S01-S08 (N=514). Bottom panel: Sources of information regarding electricity S1-S14 dependent on the knowledge what ESM is (K1).

In order to illustrate the impact of communication channels on consumers’ knowledge about ESM, we have also compared the sources of information about electricity market (S1-S14) dependent on the knowledge what ESM is (K1), see bottom panel of Figure 4. The analysis shows that respondents who declared to know what ESM is, collect information about electricity market mostly from traditional sources of information, such as: TV news, radio, newspaper, but also from their social network and Facebook. This group of respondents collect information also directly from energy companies and from official government web-pages. At the same time the respondents who are not familiar with ESM also possess information mostly from TV news, and then radio, newspaper, from their peers and Facebook and directly from the energy companies. Gener- ally, we have not observed large differences between both groups of respondents in terms of their communication channels. Among respondents who declared to have knowledge what ESM is, we have also verified which sources of information are used to find some information about ESM. As presented in Figure 5 mostly TV news, energy companies, and then newspaper feeds and social network belong to the main sources of such information. Further, as the diagrams in the bottom panel of the Figure 5 present, only around 40% of the respondents had ever searched online for ESM. Even less respondents (only 12%) had heard about the government’s program of ESM enrollment in Portugal. Then, we tested the hypotheses about the relation between marketing platform and various

15 Figure 5: Sources if information regarding electricity smart meters (N1=181). parameters by means of student’s t-tests for independent samples and different variances. In particular, we checked whether there was any statistically significant relation (positive or negative) between the marketing platforms and: the consumers preferences (P1-P6), fears and concerns (F1-F4), attitudes towards the role of the government in ESM roll-out (G1-G3), consumers’ willingness to install ESM under various conditions (D 1 D 8), monitoring of energy consumption e − e (A1) and possession of renewable energy sources in one’s household (R1), as well as possessing knowledge about ESM, and having ESM or willingness to have ESM in the future (K1-K4). First, the test was used to identify variables with significant negative correlations, which indicated the scope for improving promotions on respective communication channels. For example, a significant negative correlation between Facebook (a communication channel) and renewable energy installation at household, would indicate that there is scope to promote renewable energy among Facebook users. Thereafter, the test was used to identify variables with positive correlations, which indicated that the the communication channel is performing well. For example, a significant positive correlation between TV news (a communication channel) and renewable energy installation at household, would indicate that TV news is currently effective in promoting renewable energy among TV news audience. Analysis reveals that the importance of using conventional marketing channels for the energy markets is still quite high. Communications through energy companies, official government websites and educational programs or workshops are still the most common sources through which the users have been receiving information regarding the electricity policies and developments, as well

16 as ESM. For social media platforms, it is very evident that there is large scope of improvement. Social media platforms are proving to be eﬀective in terms of propagating the diﬀusion of new innovation (Leong et al., 2019; Hudson et al., 2016; Priyanka, 2013), but analysis in this study showed that people, who were aware about ESM, claimed to have negligible information on social media. In contrast, the same users reveled high preference towards using various social media platforms. Based on the results, in Tables A.3, A.4 and A.5, detailed recommendations have been drawn and discussed in the subsection 5.2.

5. Conclusions and Recommendations The analysis provided in this paper confirms that electricity is an abstract commodity for most people who do not usually engage much in energy conservation unless they are motivated by financial, environmental, or social incentives. In terms of ESM roll-outs, the literature, as well as our study, clearly indicates that social aspects cannot be neglected in the process of launching new technologies because societal resistance may slow down or even stop the deployment of the novel technology (see, for example (Zhou and Brown, 2017; Kowalska-Pyzalska et al., 2014; Biresselioglu et al., 2018)). Knowledge of what ESM is, increases the probability of willingness to have such equipment in one’s household and reduces the level of uncertainty regarding the potential advantages and disadvantages of ESM. We found that, in terms of having awareness about ESM in Portugal, only about 35% of the respondents knew what an ESM was, which is very low score. On the other hand over 62.5% of the respondents were interested for getting more information regarding ESM, including 19% respondents who knew what an ESM was. Thus energy companies must take more efforts to reach out to the consumers and make them aware regarding ESM and its details. Analysis of the the attitudes, preferences and fears, revealed that consumers prefer to have facilities that an ESM could provide. They would like to get more detailed and real time information about electricity consumption, would like to be able to remotely control their electricity supply, are positive towards fluctuating tariff and express confidence that they would be able to act more efficiently by making more informed decisions with access to such information. Privacy concerns were found to be an important issue, but over 61% of total respondents and 42.8% respondents who expressed privacy concerns were willing to accept ESM despite possibilities of energy companies having access to energy usage data. One of the reasons for this is the fact that the respondents are social media users, who are more aware about the data privacy issues as compared to non-users of social media. Moreover, interacting with ESM through in-home displays or a mobile application would be more or less similar to interacting with peers on social media. Hence these users are more willing towards acceptance of smart meters. Social influence is also expected to have an impact on the willingness to accept ESM. In the context of this study, social influence has an emphasized effect because of the power possessed by social media users to influence other social media users, which form even 65% of Portuguese population. 5.1. Impact of knowledge Knowledge about ESM was found to be higher among consumers who were interested or were keen to engage with the details of their energy consumption, where as there were no factors indicating interest from passive consumers to engage. This conclusion was drawn based on the 17 significant variables such as attitude of regularly monitoring energy consumption, preference to have dynamic tariff for electricity throughout the day, and desire to know get more details about electricity consumption. These factors combined with the statistically significant source of information as energy companies, shows the effect of the passive marketing of ESM by the energy companies. The example of the letter received by one of the co-authors, shown in Figure 1, is an evidence of energy companies’ passiveness. The letter doesn’t even refer to the phrase ”smart meter” or ”electricity smart meter” but just calls it ”new and technologically more evolved meter”. Although the options, to get more information regarding the upgrade for instance learning more through the website, talking to the company representative to clarify doubt or scheduling a technician visit and so on are given in the letter, only the interested consumers would engage in this scenario. This is consistent with our findings regarding the factors influencing knowledge among consumers. More attractive content with upfront explanation of benefits, and use of wider variety of communication channels should be the way forward. Due to the peculiar nature of the energy markets, a marketing mix of various types of content, as well as conventional channels and social media platforms, would be effective. Based on the analysis of the various sources of information indicated by the respondents in this study and their responses to other variables we drawn out recommendations of marketing content that can be used by the energy companies on various channels. Even though the study was conducted among social media users, the importance of using conventional marketing channels for the energy markets was still found to be high.

5.2. Recommendations for marketing content on various communication channels for raising awareness of ESM in Portugal Below we present the recommendation for marketing content, first through the conventional platforms, such as TV news or radio, and second through social media, such as Facebook or Twitter. Our recommendations can be useful not only for the energy companies who want to find an effective way to reach out the target consumer segment, but also for the national and local authorities interested in rising consumers’ awareness and engagement towards energy efficiency issues.

5.2.1. Marketing through Conventional Platforms TV News: It was found to be the most popular channel as the source of information for • electricity policies in general and ESM in particular. Based on the analysis, content showing benefits about energy consumption monitoring, effects on electricity prices, link of ESM to renewable energy systems, safety in terms of health as well as privacy, return on investment (ROI) or cost effectiveness of ESM and increased savings, would attract the consumer who prefer this platform.

Radio: It was found to be among the top platforms through which information regarding • electricity policies in general was received, but in particular for ESM, it was rated relatively lower. On radio, content addressing no stress or decrease in stress due to the use of ESM, benefits towards monitoring consumption, effects on savings & billing, safety in terms of health and accuracy of measurements would prove to be beneficial. There was quite a high significance of social influence for this platform, which would indicate that

18 campaigns through radio would be more eﬀective if people with high social inﬂuence were active participants of the campaign.

Newspaper: It was among the top platforms both, in terms of information regarding elec- • tricity policies in general and ESM in particular. Effective content for newspaper, including the ones for TV news, in addition to the ones that stress on benefits, such as remote access of electricity data or devices and government plans or policies, would prove to be effective.

Friends, Colleagues and Relatives: Word of mouth was one the most inﬂuential mediums • for the adoption of new technology, but its propagation is now more dependent on digital media (Alalwan et al., 2017). Through the analysis, it was observed that consumers are more prone to discuss the monitoring of energy consumption, eﬀects on health & personal information and their tendency to protest against ESM being made mandatory. To have positive outcomes through this platform, ensuring the availability of this information through the rest of the platforms would be required.

Energy Companies: It is the second most common source of information for electricity • policies, as well as ESM. The effect of information from energy companies would be highly actionable as they are the direct sellers and their openness towards making consumers aware, by providing information about the benefits and options, would have a convincing effect on the consumers. They need to put extra effort to provide, not just notifications, as we observed with the EDP letter in this study (please see Figure 1), but more detailed information through conventional channels. In the context of content, reminding assurances for the consumers and involving them through feedback, would attract more consumers to adopt new technologies offered by the energy companies. There is also a need make the consumers aware about the availability of personalized information, through the website, mobile application or in-home displays, as well as making the same available to them. Increasing face-to-face interactions with company representatives will also be an important trigger for the acceptance of ESM.

Workshops Educational Programs: This was one of the least common sources of infor- • mation for electricity policies, as well as ESM, mentioned by the respondents. One of the reasons for this was possibly the lack of suitable number of such programs to reach to the masses or lack of proper content delivery (Hess, 2014), which was not in the scope of this research. Based on the analysis in this study, we can recommend that content, such as benefits of ESM in reducing energy wastage, its effect on cost & billing, addressing health and privacy concerns and cost effectiveness or ROI of ESM, would make such programs more effective. These programs were also linked to increasing social influence and interaction between friends, relatives and colleagues, which have been found to be important for increasing the diffusion of ESM.

Official Government Websites: Detailed information about various aspects of electricity • in general and ESM in particular are available on these websites, but difficult to find. These websites need to be updated, so as to enable the consumers to find the information they are

19 seeking more conveniently and also addressing the preferences, fears and decision making factors of the consumers addressed in this whole sub-section 5.2.1.

Telephone / SMS: This platform was found to be in the bottom half in terms of a signiﬁcant • source of information about electricity in general and ESM in particular. This platform was well suited for notiﬁcations, updates with web-links and interactions with representatives of energy companies. Content for marketing through this platform would include areas in all other conventional platforms mentioned in the points prior to this and, additionally, the availability of redressing to consumer queries as well as complaints.

5.2.2. Marketing through Social Media Facebook: It is the world’s largest social network, with over 2 billion users, but among the • respondents in this study, it was second most preferred platform. It was also not indicated as a source through which the respondents received information regarding electricity or ESM. Promoting content, such as availability of remote access of electricity usage and control through ESM, cost effectiveness of ESM and effectiveness in terms of reducing wastage of electricity, would be effective.

Facebook Messenger and WhatsApp: Both these platforms had similar functions and the • same parent company. WhatsApp was the most popular social network being used by the respondents of this study but majority the of the respondents indicated that they did not receive information regarding ESM through these two platforms. The significance of social influence for making a decision regarding the installation of ESM was high amongst these users, which indicated that, on receiving information regarding ESM through What- sApp or Facebook Messenger, the users were more likely to take positive steps towards the installation of ESM. The users of these platforms preferred to have information regarding electricity consumption, wanted to reduce energy wastage, wished to have remote access to devices and cared about information privacy and the effects on health. Marketing content addressing these themes would be appealing to the users on these platforms.

LinkedIn: It was the most popular social media platform among the respondents for getting • the source of information regarding electricity in general and ESM in particular. In addition to the content recommended for Facebook, elaborating government plans and policies would be useful. LinkedIn also had a high social influence significance, which suggested that these consumers would be able to gather influence among their peers. Hence, increasing the output manifolds.

Twitter: Twitter played an important role to empower people for partaking in sharing what • continues to happen around the world. It was shown to be quite useful for the diffusion of innovation, due to its hashtag function (Chang, 2010). From our analysis, content related to savings, due to the installation of ESM, would be effective in attracting consumers. Respon- dents, who preferred Twitter as one of the communication platforms, were also the ones who expressed interest in installing ESM. Hence, the use of Twitter would amplify the outcomes of the marketing campaign. 20 YouTube: It is another social media platform which was quite commonly preferred by the • respondents, but they indicated not receiving information regarding electricity in general and ESM in particular through it. These users showed interest towards monitoring energy consumption, reducing energy wastage, receiving updates regarding electricity pricing and cared about savings and health effects. Content regarding these topics would attract these users towards the acceptance of ESM.

Instagram: This photo sharing platform was the 4th most popular among the respondents in • this study. The preference and interests of users were similar to that of Facebook, therefore, it can be suggested that the content recommended for Facebook would also be suitable for Instagram.

SnapChat: This is a very peculiar platform, and was found to be the least popular among • the respondents in this study. Although the statistical signiﬁcance of SnapChat was quite high with some variables, which can be see in Table A.5, the number of users were too low to recommend enacting separate eﬀorts for marketing on this platform. Content created for Facebook Messenger and WhatsApp would be suitable for this platform at large, if this was included in the marketing strategy.

Managers should interpret these recommendation in conjunction with the impact of knowledge and also the goals they wish to achieve. The recommendations describe the content material and content type (images, videos, text, web links and so on) based on the properties of the communication channel. Particular care should be taken while using the same content type for the same content material to ensure that intensity of marketing is not too high. We also recommend that managers use social media measures, already available in the literature (for example see: Chodak et al. (2019); Chawla and Chodak (2020)), to plan and measure the eﬀectiveness of their campaigns.

6. Limitations and Future Scope of Research This study has limitations that point to future works and research avenues. The sample analyzed introduces limitations, due to the geographical context and to the non-random sampling methods used. Therefore, it is recommended to replicate the study in different samples. This study also opens new horizons for research, such as analyzing the particular effectiveness of each of the marketing platforms for smart meter diffusion.

Acknowledgments This work was supported by the National Science Center (NCN, Poland) by grant no. 2016/23/B/HS4/00650 and partly by the Faculty of Computer Science and Management, Wrocław University of Science and Technology from funds of the Ministry of Science and Higher Education subsidy in the part devoted to conducting research activities in 2019. We would like to thank the editors and reviewers for their constructive remarks and suggestions. We want also to thank Marta Pytel for English language proofreading. 21 Appendix A.

Table A.2: Deﬁnitions of the variables, coding and description (N=518)

Variable Code Description Group A (Variables of Phase 1) Demographics D1 - D10, D81 Gender D1 2 categories (nominal) Age D2 6 categories (ordinal) Relationship status D3 5 categories (nominal) Highest Educational Qualification D4 6 categories (ordinal) Occupation / Employment D5 6 categories (nominal) Monthly Household Income (in Euro per month) D6 13 categories (ordinal) Range of electricity bill (in TL per month) D7 6 categories (ordinal) Total members in the household D8 6 categories (ordinal) Number of children D81 5 categories (ordinal) Type of house D9 4 categories (nominal) Area of living D10 4 categories (ordinal) Belongings of smart devices & personal assets) B1 - B6 House B1 (1) yes / (0.5) no, but I plan to buy it within one year/ (0) no, and I do not plan to buy it. Flat or Apartment B2 Laptop or Desktop B3 Wifi / Internet connection at home B4 Smart TV B5 Other home appliances that can connect to the in- B6 ternet You regularly monitor energy usage in house- A1 (1) yes / (0.5) hard to say / (0) no hold Any renewable energy sources installed at the R1 (1) yes / (0.5) hard to say / (0) no household Social media platforms commonly used S01 - S08 Facebook S01 (1) yes/ (0) no Facebook Messenger S02 LinkedIn S02 Twitter S03 WhatsApp S04 Youtube S05 Instagram S02 Snap Chat S02 Source of information regarding electricity S1 - S14 (1) Yes / (0) No (Note: variables are listed (prices, new offers, etc.) in the table notes) Preferences regarding ESMplatforms P1-P6 Getting more details on electricity usage is desir- P1 able Getting real time information of electricity usage P2 (1) yes / (0.5) hard to say / (0) no would be useful You can reduce energy wastage if you had real P3 time information of electricity usage Continued on next page 22 Table A.2 – continued from previous page Variable Code Description Energy prices would be reduced if waster of en- P4 ergy decreases Prefer to be able to remotely turn on or off the P5 electricity supply Prefer to have fluctuating unit rates of electricity P6 usage Willingness to install SM De1-De4, De6-De8 Willing if ESMcould help save money De1 Willing if ESMcould help save money, but possi- De2 ble have adverse effect on health Willing if ESMcould have save money, but energy D 3 e (1) yes / (0.5) hard to say / (0) no companies would have access to electricity usage data Willing if company representative visits home and De4 explain all details Willing to install if upgrade to ESMis free De6 Willing to install if one of the friends / relatives/ De7 neighbours recommends it Willing to install if one of the friends / relatives / De8 neighbours installs ESM at their house Concerns about ESMusage F1,F4 Data privacy concerns F1 (1) yes / (0.5) hard to say / (0) no Fluctuations in unit rate of electricity would cause F4 additional stress Willingness to search or collect more informa- Q1 (1) yes / (0.5) hard to say / (0) no tion regarding SM Group B (Conditional Variable for Phase 2) Knows what is a ESM K1 (1) yes / (0.5) hard to say / (0) no Group B (Variables of Phase 2 - For K1 = 1) Knowledge about SM K2 - K4 Has ESMinstalled at home K2 ((1) yes / (0.5) hard to say / (0) no In process of installing ESMat home K3 Plans to install ESM at home K4 Source of information regarding SM I1, I2, I31 - I44 Search online about ESM I1 (1) yes / (0.5) hard to say / (0) no Hear about Government’s ESM rollout pro- I2 gramme Other sources (variables are listed in the table I31 - I44 (1) Yes / (0) No notes) Social influence W1 (1) yes / (0.5) hard to say / (0) no Willing to install ESM even if payment is re- De5 (1) yes / (0.5) hard to say / (0) no quired for upgrade Preferences regarding the role of the govern- G1-G3 ment in ESM enrollment Government should make it mandatory for all to G1 have ESM (1) yes / (0.5) hard to say / (0) no Continued on next page

23 Table A.2 – continued from previous page Variable Code Description Government should give an option to decline in- G2 stallation of ESM Would protest if government makes it mandatory G3 to install ESM Concerns about ESM usage F2,F3 Billing through ESM could be inaccurate F2 (1) yes / (0.5) hard to say / (0) no ESM could have adverse eﬀects on health F3 Willingness to have one’s home to be equipped X1 (1) yes / (0.5) hard to say / (0) no with ESM

Note: TV News (S1, I31); Radio (S2, I32); Newspaper (S32, I33); Friends, relatives, colleagues (S4, I34); Face- book (S5, I35); Facebook Messenger (S6, I36); Twitter (S7, I37); WhatsApp (S8, I38); LinkedIn (S9, I39); YouTube (S10 I40); Energy Companies (S11, I41); Oﬃcial government websites (S12, I42); Workshops / educational campaigns (S13, I43); Telephone / SMS (S14, I44)

24 Table A.3: Student’s t-test for correlation signiﬁcance between the information sources for ESM and chosen variables

I31 I32 I33 I34 I35 I37 I39 I41 I42 I43 I44 A1 -2.066* -2.229* -1.783* -1.859* R1 -2.402* P1 -8.306*** 2.065* -8.315*** P2 -6.81*** -6.825*** -6.815*** P3 -7.732*** -7.739*** P4 2.3* -1.704* P5 -7.961*** -7.985*** -7.969*** P6 -2.226* -8.769*** -1.926* F1 2.706** -12.02*** F2 6.252*** F4 -2.017* 2.704**

25 G1 -2.097* -2.034* G2 -7.854*** G3 -2.282* -2.824** -4.433*** -2.183* DE1 -4.994*** -2.733** -4.996*** DE2 -2.458** -1.746* 3.739*** DE3 -2.849** 2.236* DE4 -14.675*** 1.995* DE5 -2.078* DE6 -1.666* DE7 -14.828*** DE8 -8.076*** -8.101*** -8.084*** K3 -1.846* 2.486** K4 1.772* -18.104*** X1 -2.887** -10.818*** 2.086* -10.837***

S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 S13 S14 A1 -2.055* -2.93** -1.715* R1 -2.167* -1.883* 8.62*** P1 -2.808** -11.225*** -1.766* 2.218* -11.195*** P2 -9.516*** -9.511*** -9.517*** -9.504*** P3 -11.006*** -2.233* 2.605** -10.988*** P4 -15.053*** -2.029* -15.007*** P5 2.176* -12.465*** -2.994** 2.695** 1.71* -12.425*** P6 -13.8*** -13.78*** F1 -2.555** -2.193* F2 -2.262* 2.263* -1.94* -1.76* F4 -1.812* 14.255*** G1 -3.083**

26 G2 -1.74* -7.861*** -2.016* -2.376** G3 -2.436** -2.128* -3.413** DE1 2.801** 2.121* 1.967* -6.759*** -6.759*** -6.762*** -6.758*** 2.291* DE2 -1.773* -3.143*** -2.303* -2.281* -1.934* 1.789* 6.307*** DE3 1.684* 2.455** 1.732* DE4 -2.643* -1.728* -2.036* DE5 -1.984* -2.218* DE6 1.906* 1.737* DE7 3.885*** 2.117* 2.12* 2.059* DE8 -13.731*** K1 -2.662** -2.3* 16.716*** K2 2.235* -1.996*** K3 -2.055* -2.034* K4 1.729* -18.104*** X1 -10.818*** -2.018*

S01 S02 S03 S04 S05 S06 S07 S08 A1 2.253* 1.855* 1.97* 2.749** R1 -2.184* -1.793* -3.255*** -1.924* P1 -2.283* -1.666* -2.193* -4.121*** -3.828*** P2 -1.854* -2.912** -1.769* -2.281** -1.906* -3.075** -2.448** P3 -1.913* -2.492*** -2.831** -1.683* -2.232* -3.658*** P4 1.904* 1.783* P5 -1.744* -3.428*** -3.802*** -2.456** -2.926** -4.265*** -3.563*** -4.133*** P6 -1.939* F1 -1.748* F4 -2.568** -1.806* G1 -2.431** G2 -1.729* 27 G3 1.96* DE1 -3.864*** DE2 -1.929* -3.438*** -2.12* -3.09*** -2.777** DE3 -2.081* -2.172* -2.222** -0.674** DE4 -2.708** -2.479** -1.796* E5 -1.713* DE6 -1.748* -3.16*** -2.341** -2.289* -2.073* -3.448*** DE7 -2.627** -2.735** -3.009** DE8 -2.806** -3.239*** -3.793*** -2.728** -1.994* -3.403*** K1 2.141* K3 -2.498** -2.491** -1.881* -2.505** K4 -1.915* -1.991* -2.106* -1.986* -2.081* X1 -3.219*** -2.708** -2.8** -2.142*

Note: Student’s t-test for independent sample with different variances. The cells highlighted in red, indicate significant negative correlation, whereas the cells highlighted in green, show the significant positive correlation; Significance level: ***p¡0.001, **p¡0.01, *p¡0.05. References Abidin, S., Omar, N., Rahman, H., Rosli, M., 2010. Socio-informatics: identifying influential factors in digital elements. In Proceedings of the 14th WSEAS International Conference on Computers: Part of the 14th WSEAS CSCC multiconference organized by World Scientific and Engineering Academy and Society (WSEAS) 1, 397– 402. Aghaei, J., Alizadeh, M., 2013. Demand response in smart electricity grids equipped with renewable energy sources: A review. Renewable and Sustainable Energy Reviews 18, 64–72. Akroush, M., Zuriekat, M., Al Jabali, H., Asfour, N., 2019. Determinants of purchasing intentions of energy-efficient products: The roles of energy awareness and perceived benefits. International Journal of Energy Sector Manage- ment 13, 128–148. Alalwan, A., Rana, N., Dwivedi, Y., Algharabat, R., 2017. Social media in marketing: A review and analysis of the existing literature. Telematics and Informatics 34, 1117–1190. Avancini, D., Rodrigues, J., Martins, S., R.A.L., R., Al-Muhtadi, J., Solic, P., 2019. Energy meters evolution in smart grids: A review. Journal of Cleaner Production 217, 702–715. Barrios-O’Neill, D., Schuitema, G., 2016. Online engagement for sustainable energy projects: A systematic review and framework for integration. Renewable and Sustainable Energy Reviews 54, 1611–1621. Bellido, M., Rosa, L., Pereida, A., Falcoa, D., Ribeiro, S., 2018. Barriers, challenges and opportunities for microgrid implementation: The case of Federal University of Rio de Janeiro. Journal of Cleaner Production 180, 203–216. Bentoa, M., Martinez, L., Martinez, L., 2018. Brand engagement and search for brands on social media: Comparing generations x and y in Portugal. Journal of Retailing and Consumer Services 43, 234–241. Biresselioglu, M., Nilsen, M., Demir, M., Royrvik, J., Koksvik, G., 2018. Examining the barriers and motivators affecting european decision makers in the development of smart and green energy technologies. Journal of Cleaner Production 198, 417–429. Bogdal, M., 2013. Communication management in public sectors: the case of the Polish energy sector. Comunicacao Publica 8, 7–23. Burchell, K., Rettie, R., Roberts, T., 2016. Householder engagement with energy consumption feedback: the role of community action and communications. Energy Policy 88, 178–186. Chang, H., 2010. A new perspective on Twitter hashtag use: Diffusion of innovation theory. Proceedings of the American Society for Information Science and Technology 47, 1–4. Chawla, Y., Chodak, G., 2020. Social media marketing: Organic promotions of web-links on facebook. WORking papers in Management Science (WORMS), Department of Operations Research and Business Intelligence, Wrocław University of Science and Technology, Poland, WORMS/20/03 . Chawla, Y., Kowalska-Pyzalska, A., 2019. Public awareness and consumer acceptance of smart meters among Polish social media users. Energies 12, 2759. Chawla, Y., Kowalska-Pyzalska, A., Skowronska-Szmer, A., 2020. Perspectives of smart meters’ roll-out in india: an empirical analysis of consumers’ awareness and preferences. working paper, Hugo Steinhaus Center, Wroc/law University of Technology HSC Research Reports, HSC/19/03 . Chawla, Y., Kowalska-Pyzalska, A., Widayat, W., 2019. Consumer willingness and acceptance of smart meters in Indonesia. Resources 8, 177. Chodak, G., Chawla, Y., Dzidowski, A., Ludwikowska, K., 2019. The effectiveness of marketing communication in social media, in: ECSM 2019 6th European Conference on Social Media; Academic Conferences and Publishing Limited: Sonning Common, UK, p. 73. Crispim, J., Braz, J., Castro, R., Esteves, J., 2014. Smart grids in the eu with smart regulation: Experiences from the UK, Italy and Portugal. Utilities Policy 31, 85–93. Das, S., Lall, G., 2016. Traditional marketing vs digital marketing: An analysis. International Journal of Commerce and Management Research 2, 5–11. Doostizadeh, M., Ghasemi, H., 2012. A day-ahead electricity pricing model based on smart metering and demand-side management. Energy 46, 221–230. Droge, C., Stanko, M., Pollitte, W., 2010. Lead users and early adopters on the web: the role of new technology product blogs. Journal of Product Innovation Management 27, 66–82.

28 Duffett, R., 2015. Facebook advertising’s influence on intention-to-purchase and purchase amongst millennials. In- ternet Research 25, 498––526. Ellabban, O., Abu-Rub, H., 2016. Smart grid customers’ acceptance and engagement: An overview. Renewable and Sustainable Energy Reviews 65, 1285–1298. European Commission DG Energy, 2018. European smart metering benchmark. benchmarking smart metering deployment in the eu-28. revised final report. European Commission, DG Energy, 27 June 2019. Filo, K., Lock, D., Karg, A., 2015. Sport and social media research: a review. Sport Management Review 18, 166–181. Fiore, M., Silvestri, R., Conto,` F., Pellegrini, G., 2017. Understanding the relationship between green approach and marketing innovations tools in the wine sector. Journal of Cleaner Production 142, 4085–4091. Foulds, C., Robison, R., Macrorie, R., 2017. Energy monitoring as a practice: Investigating use of the imeasure online energy feedback tool. Energy Policy 104, 194—-202. Gans, W., Alberini, A., Longo, A., 2013. Smart meter devices and the effect of feedback on residential electricity consumption: Evidence from a natural experiment in northern ireland. Energy Economics 36, 729–743. Gerpott, T., Paukert, M., 2013. Determinnants of willingness to pay for smart meters: An empirical analysis of household customers in Germany. Energy Policy 61, 483–495. Ghazvini, A., Ramos, S., Soares, J., Castro, R., Vale, Z., 2019. Liberalization and customer behavior in the Portuguese residential retail electricity market. Utilities Policy 59, 100919. Gong, T., Li, D., Liu, Y., Wang, G., Zhu, H., 2019. Analysis of marketing strategy of electricity selling companies in the new situation. Journal of Physics: Conference Series 1187, 022043. Hanna, R., Rohm, A., Crittenden, V., 2011. We are all connected: The power of the social media ecosystem. Business Horizons 54, 265–273. Heikkinen, M., Nurminen, J., Smura, T., Hammainen, J., 2012. Energy efficiency of mobile handsets: Measuring user attitudes and behavior. Telematics and Informatics 29, 387–399. Hess, D., 2014. Smart meters and public acceptance: comparative analysis and governance implications. Health, Risk & Society 16, 243–258. Hille, S., Weber, S., Brosch, T., 2019. Consumers’ preferences for electricity-saving programs: Evidence from a choice-based conjoint study. Journal of Cleaner Production 220, 800–815. Hudson, S., Huang, L., Roth, M., Madden, T., 2016. The influence of social media interactions on consumer–brand relationships: A three-country study of brand perceptions and marketing behaviors. International Journal of Research in Marketing 33, 27–41. Kahma, N., Matschoss, K., 2017. The rejection of innovations? rethinking technology diffusion and the non-use of smart energy services in finland. Energy Resources & Social Science 34, 27–36. Kowalska-Pyzalska, A., 2018. What makes consumers adopt to innovative energy sources in the energy market? A review of incentives and barriers. Renewable and Sustainable Energy Review 82, 3570–3581. Kowalska-Pyzalska, A., 2019. Do consumers want to pay for green electricity? a case study from Poland. Sustain- ability 11. Kowalska-Pyzalska, A., Byrka, K., 2019. Determinants of the willingness to energy monitoring by residential consumers: A case study in the city of Wroclaw in Poland. Energies 12, 907. Kowalska-Pyzalska, A., Maciejowska, K., Suszczynski,´ K., Sznajd-Weron, K., Weron, R., 2014. Turning green: Agent-based modeling of the adoption of dynamic electricity tariffs. Energy Policy 71, 164–174. Kowalski, J., Matusiak, B.E., 2019. End users’ motivations as a key for the adoption of the home energy management system. International Journal of Management and Economics 55, 13–24. Krishnamutri, T., Schwartz, D., Davis, A., Fischoff, B., de Bruin, W.B., Lave, L., Wang, J., 2012. Preparing for smart grid technologies: A behavioral decision research approach to understanding consumer expectations about smart meters. Energy Policy 41, 790—-797. Kumar Verma, R., Kumar, S., Vigneswara Ilavarasan, P., 2017. Government portals, social media platforms and citizen engagement in india: Some insights. Procedia Computer Science 122, 842–849. Leong, L., Hew, T., Ooi, K., Lin, B., 2019. Do electronic word-of-mouth and elaboration likelihood model influence hotel booking? Journal of Computer Information Systems 59, 146–160. Lopes, M., Antunes, C., Janda, K., Peixoto, P., Martins, N., 2016. The potential of energy behaviors in a smart(er)

29 grid: Policy implications from a portuguese exploratory study. Energy Policy 90, 233–245. Ma, G., Lin, J., Li, N., 2018. Longitudinal assessment of the behavior-changing effect of app-based eco-feedback in residential buildings. Energy and Buildings 159, 486–494. Miguel, C., Mendes, A., Madeira, L., 2018. An overview of the portuguese energy sector and perspectives for power- to-gas implementation. Energies 11, 1–20. Nakarado, G., 1996. A marketing orientation is the key to a sustainable energy future. Energy Policy 24, 187–193. Nizetic, S., Djilali, N., Papadopoulos, A., Rodrigues, J., 2019. Smart technologies for promotion of energy efficiency, utilization of sustainable resources and waste management. Journal of Cleaner Production 231, 1–10. Ntanos, S., Kyriakopoulos, G., Chalikias, M., Arabatzis, G., Skordoulis, M., 2018. Public perceptions and willingness to pay for renewable energy: A case study from Greece. Sustainability 10, 687. Paetz, A.G., Duetschke, E., Fichtner, W., 2012. Smart homes as a means to sustainable energy consumption: A study of consumer perceptions. Journal of Consumer Policy 35, 23–41. Parag, Y., Sovacool, B., 2016. Electricity market design for the prosumer era. Nature energy 1, 16032–16053. Park, C., Kim, Y., Jeong, M., 2018. Influencing factors on risk perception of IoT-based home energy management services. Telematics and Informatics 35, 2355–2365. Priyanka, S., 2013. A study of online advertising on consumer behaviour. International Journal of Engineering and Management Science 3, 461––465. Saxena, A., Khanna, U., 2013. Advertising on social network sites: A structural equation modelling approach. Vision 17, 17–25. Schleich, J., Faure, C., Klobasa, M., 2017. Persistence of the effects of providing feedback alongside smart metering devices on household electricity demand. Energy Policy 107, 225–233. Sheau-Ting, L., Mohammed, A., Weng-Wai, C., 2013. What is the optimum social marketing mix to market energy conservation behaviour: an empirical study. Journal of Environmental Management 131, 196–205. Soroczynski,´ B., Szkutnik, J., 2015. New possibilities in electrical energy efficiency, in: 2015 16th International Scientific Conference on Electric Power Engineering (EPE), IEEE. pp. 13–16. Sovacool, B., Kivimaa, P., Hielscher, S., Jenkins, K., 2017. Vulnerability and resistance in the united kingdom’s smart meter transition. Energy Policy 109, 767–781. Streimikiene, D., Vveinhardt, J., 2015. Community based social marketing for implementation of energy saving targets at local level. Amfiteatru Economic Journal 17, 723–734. Todor, R., 2016. Blending traditional and digital marketing. Bulletin of the Transilvania University of Brasov, Series V - Economic Sciences 9, 51–56. Verbong, G., Beemsterboer, S., Sengers, F., 2013. Smart grids or smart users? involving users in developing a low carbon electricity economy. Energy Policy 52, 1175–125. Zhou, S., Brown, M., 2017. Smart meter deployment in Europe: A comparative case study on the impacts of national policy schemes. Journal of Cleaner Production 144, 22–32.

30 94 BIBLIOGRAPHY Paper 3

Y. Chawla, A. Kowalska-Pyzalska Public Awareness and Consumer Accep- tance of Smart Meters among Polish So- cial Media Users energies

Article Public Awareness and Consumer Acceptance of Smart Meters among Polish Social Media Users

Yash Chawla and Anna Kowalska-Pyzalska * Department of Operations Research, Faculty of Computer Science and Management, Wroclaw University of Science and Technology, Wyb. Wyspianskiego 27, 50-370 Wroclaw, Poland * Correspondence: [email protected]; Tel.: +48-71-3202524

Received: 17 May 2019; Accepted: 9 July 2019; Published: 18 July 2019

Abstract: Both people and things are becoming smarter by the day. Industrial evolution at the peak of the 4.0 phase and indications of 5.0 phase are fascinating. In these circumstances, fulﬁlling the demand for energy is a challenge faced by countries all over the world. Upgrading the current energy distribution systems with smart grids and smart meters are steps towards facing this challenge, especially for Poland, which is primarily relying on conventional sources of energy. For any innovation or new technology, creating public awareness and consumer acceptance enhances the chance for a fruitful deployment. To achieve this, various communication channels are adopted and social media is found to be one of the most effective tools for it. This study discusses the awareness level and consumer acceptance of social media users in Poland. The source through which they receive information regarding electricity in general and smart meters (SM) in particular and their preferences and willingness about the installation of SM under various conditions are discussed in detail. Findings show that there is low level of public awareness among the respondents which causes them to develop myths, fears and doubts about SM installation in their households. More effort is required from the government as well as from the energy companies in order to increase the public awareness which will result in an increase in consumer acceptance. Based on the results, the article also contains recommendations that can be used by governments as well as energy companies to create a positive feeling about SM to affect consumer behavior.

Keywords: smart meters; public awareness; social acceptance; knowledge; consumers; social media; Facebook

1. Introduction Energy markets are experiencing significant changes at various levels due to ecological, economic and technical challenges. Among them, the enrollment of smart meters, in many countries all over the world, opens new possibilities, not only to the energy suppliers, but also for the consumers. Energy smart meters (SM) are advanced electricity meters which can offer a range of intelligent functions and have intended benefits for energy consumers, suppliers and networks [1]. Consumers can benefit from smart meters because they provide more accurate bills, allow easier switching, enable clearer energy use through an in-home display, smart metering information systems, and have some potential for the reduction of utility costs based on reduced consumption. Smart meters would make it easier for the networks to balance the grid, as they could facilitate a smarter grid with real-time information of supply and usage. The suppliers also stand to be benefited by it, as it could reduce the customer service overhears through more accurate billing and avoiding site visits (for example to take meter readings) [1–3].

Energies 2019, 12, 2759; doi:10.3390/en12142759 www.mdpi.com/journal/energies Energies 2019, 12, 2759 2 of 27

The Electricity Directive 2009/72/EC requires the EU Member States to roll out intelligent electricity smart meters to 80% of consumers by 2020. One of the main goals of this enrollment is to enable active participation of consumers in the electricity supply market. The implementation of those metering systems may be subject to cost-benefit analysis. The example of British cost-benefit analysis from 2016 suggests that the majority of the total benefits connected with SM enrollment would come from supplier cost savings (49% of all total benefits) and energy savings (32% of total benefits). Benefits will also be achieved in terms of carbon savings and air quality benefits (8%), peak load shifting (6%), and network-related benefits (5%) [1]. Poland, which is one of the largest countries in Europe, has prolonged the national enrollment of SM until the year 2026. So far, no more than 10% are equipped with the meters. In Poland, just as in many other countries, the obligation to roll-out smart meters is supplier-led. Due to this, the planning and execution is carried out by the energy suppliers, as per their will, in a way that best suits their business and based on the needs of their customers. They are only required to achieve the overarching obligation to complete the roll-out by the end of 2026. The cost of installation (which is assumed to be around 80–160 USD per SM unit) is or will be inevitably transferred on the customers, who pay for it in their electricity bills. Although the concept of SM enrollment is fine, there are many doubts as to whether these new technologies and behavioral changes introduced by them will be widely adopted by the consumers in time to achieve the ambitious aims and targets. On the one hand, in Roger’s theory of innovation diffusion, the vital role of communication channels through social systems, such as word of mouth or recently also social media, in dissemination of information on innovations is emphasized [4]. On the other hand, according to the social capital theory, interpersonal communication is a significant way of attaining resources such as information on energy efficiency innovations for accomplishing certain objectives [5], which is also facilitated by social media. Communication channels have a huge role to play in shaping the message. Users on social media are exposed to content which is more personalized for them and more likely to be from people they have in the network. Studies have indicated that this has a great effect on the choices made by the users, their purchase intention, as well as their acceptance about new items [6]. There have already been several studies about the social acceptance and awareness of SM conducted by means of various tools (quantitative and qualitative methods; simulations; conceptual studies), paying attention to various aspects of this issue, such as: consumers’ expectations and perceptions about SM [7,8], consumers’ acceptance and engagement [9], effect of feedback by means of SM and smart metering information systems (SMP) on energy consumption [10–17], willingness to pay for SM [18], efficiency of education and training in SM and SMP adoption [19], or incentives and barriers of SM diffusion [19–21]. However, to the best of authors’ knowledge, the awareness and acceptance of SM among social media users have not yet been explored. Within our study, we seek to fill this gap. In our study, we assume that consumers, who possess common access to new, smart technologies, will have a higher acceptance level towards smart meters and will see more advantages to use them. The general aim of this paper is to investigate the acceptance level of smart meters among social media users. In particular, the socio-economic factors, possession and usage of various smart devices, knowledge and preferences regarding SM, are assumed to be potential predictors of consumers’ acceptance of SM and their willingness to install it. We have formulated the following objectives of the study: (1) to explore the source of information social media users have regarding electricity in general and smart meters in particular, (2) to analyze the consumers’ preferences and fears regarding installation of SM at their household, and finally, (3) to investigate the willingness to install SM under various conditions (e.g., financial, regulative ones). Energies 2019, 12, 2759 3 of 27

Although social acceptance towards SM has already been broadly investigated, we still see a gap in the literature. Namely, we believe that, presently, customers become “smart”, because of the access to various smart solutions in different markets. It is also true for the energy market. At the same time, smart customers are often present in social media, such as Facebook, Twitter or LinkedIn and are more aware regarding current and future technology trends. Hence, based on the literature review, we predict that social media users, who are usually in favor of new technologies, will be more aware and interested towards smart meters and especially SMP. They already have experience with modern communication technologies, such as mobile apps and internet web pages which are a communication channel between the smart meter and a user. We also believe that social media users can spread information about the innovation via their social network. However, at the same time, the literature shows the generally low interest levels among consumers towards issues connected with electricity and the energy market. By means of our survey, we seek to explore the awareness and acceptance level of SM among social media users and to verify our expectations. To the best of our knowledge, the awareness and acceptance of SM among social media users have not been checked yet. Therefore, within our study, we want to ﬁll this gap in. The structure of the article is as follows: First, in Section2 we discuss the Polish power system, followed by the public awareness of smart technologies in the energy market in Section3. In particular, we focus on the awareness and acceptance of market novelties among social media users. In Section4 we present the methods used and the theoretical background of the study. Next, Section5 describes the empirical setup of the survey, a statistical analysis of the results and discusses the main ﬁndings. Finally, in Section6, conclusions together with some market and policy implications are provided.

2. The Challenges of the Polish Power System Poland is currently facing a steep challenge related to infrastructure, economics and politics in the energy sector. One of the most severe problems is that over 80% dependence on fossil fuel-based energy generation (hard coal and lignite). The upward trend in energy consumption, weak transformation network, and limited generation capacity due to decommissioning of old generation units, fail to aid in this problem. To overcome this, the Polish energy sector needs to be revamped with increased efficiency of generation and transmission, while fulfilling its obligation to increase the share of RES in the total energy production (which was approx. 8% only in 2014) as well as limiting the CO2 consumption due to the Climate Policy Agreements [22–26]. Polish people are in favor of an energy policy which develops RES in Poland. According to a survey conducted in 2014, a staggering 95% of Poles are in agreement with this and 78% wish to produce their own energy if it were possible [27]. The results of another study, conducted to understand the preferences of Poles regarding investment in RES, its knowledge, legal regulations and willingness to adopt it, showed that 50% of Poles were not interested in RES primarily because of the inability to install the panel (e.g., respondents who were living in a flat or renting a house). 21% of the respondents showed readiness to invest in PV panels if the period to get return on investment would not be longer than five years [28]. This suggests that it would take a while for the Polish energy sector to move from conventional sources of energy to RES, even though there is high consent among the citizens towards implementation. There is a need to explore other remedies, which could be viable and effective in terms of time as well as the cost. Several proposals for remedies have been enlisted in the Polish Energy Policy 2050 (PEP 2050). Smart grids is one of the prominent elucidations listed among these remedies. There are several benefits of introducing smart grids and smart meter systems, while updating the current transmission and distribution systems or while constructing new ones [25]. Potential benefits of smart grids involves increased security and reduction in cost of exploitation through optimization of grid management and amplified potential of demand side management/demand response tools (DSM/DR) through the real time price signals and improved communication between the end-users and energy suppliers [25]. Concrete steps are already being taken towards upgrading the energy distribution system and Energies 2019, 12, 2759 4 of 27 establishing smart grids. In October 2017, the Polish energy distribution system operator Energa announced the signing of a deal with the Ministry of Energy to jointly fund the implementation of the smart grid program. This project to modernize the energy network was estimated to cost 65.2 million dollars, out of which the energy ministry had agreed to provide up to 45.1 million dollars in sources from the Operational Program Infrastructure and Environment (OPI&E) [29]. These are positive signs, but a question arises here on the readiness of the consumers to cope and accept these forthcoming changes. Previous studies, conducted for general users, show that there is a lack of awareness among the users regarding smart metering and smart grids. One of the reasons for this could be that the effectiveness of educational campaigns may not be percolating to enough users [19,30]. These studies however do not examine the communication channels through which the users receive their information, hence it would be interesting to examine the awareness level of people preferring communication channels, such as social media, which have been found to have higher diffusion effects. The lack of public awareness might also be due to the fact that the Polish government currently does not have any national roll-out plan for the smart-meters to all customers by 2020 [31].

3. Public Awareness of Smart Meters (SM)–Literature Review

3.1. Smart Grids, Smart Metering Systems, Smart Devices Presently, many products and services, as well as, concepts and ideas, must be “smart” to catch consumers’ attention. It is also true in the energy market. The word “smart” means intelligent, user friendly, and that the product or service can be used automatically and remotely without the customers’ presence. In the case of energy markets, a general concept of smart grids (SG) has been developed and implemented in various parts of the electricity network. This concept is based on the broad usage of modern communication technologies to exchange information between market participants, such as energy generators, energy suppliers and sellers, market operators, and end-users. It aims to increase the energy efﬁciency in production, transmission and consumption [2,9,10,30,32]. According to the experts, the transition of the power system into smart grids cannot be achieved without exchange of electricity meters at the customers’ level from traditional (analog) ones to smart meters [12,32]. One of the biggest advantages of SM is the access to the current data about electricity consumption for both sides: energy suppliers and customers. Consumers receive access mostly by means of smart metering information systems, which include internet platforms and mobile apps, and are often combined with other smart appliances, such as in-home displays, home area networks or smart plugs [7,10]. Various studies show that this access helps make informed decisions about electricity consumption, thus resulting in an average energy saving of between 2 to 4% [33,34]. Less energy consumption also results in fewer CO2 emissions, which is highly desirable. The access to data about electricity consumption also enables the customers to learn about different electricity tariffs offered, depending on the time and the spot electricity price [1,12,35], hence this could result in the reduction of electricity bills [36]. SM also have a potential to facilitate the growth in use of new products and services such as smart appliances and home batteries. These can be operated remotely or automatically in response to energy tariff price information [1]. Consumers could become active in the network by providing energy or demand side response services to balance the grid by using such appliances. SM would also make it easier for the consumers to change between suppliers, creating a more competitive market with lower tariffs [37]. Suppliers would also be beneﬁting from the installations of smart meters. More accurate data should reduce inquiries and customer service overheads, reduce debt management needs and also site visits for collecting meter readings. Suppliers would also be able to remotely monitor the customers for actions switching to pre-payment meters, disconnecting customers and so on, though the customers will still have the existing protections which are in place. In a competitive market, the suppliers may transfer some margins of cost saving to the customer as well [1]. Energies 2019, 12, 2759 5 of 27

To have an increase in efﬁcient energy consumption, companies and policy makers need to pitch in and do their part. This would not be possible without the acceptance, engagement and effort of the end users, i.e., the consumer. Their acceptance of SG products and services would be indispensable to successfully build SG, as mentioned by Ellabban and Abu-Rub [9]. Raising the cognizance of the consumers and actively involving them in the energy market is one of the major challenges that needs to be addressed, increasing the popularity of SG concept and broader possibility of implementing DSM/DR tools [9,38,39].

3.2. The Role of Social Media in Diffusion of Innovation The first part of the 21st century has seen a rapid shift in information dissemination mediums, from conventional one-way communication methods such as newspaper, television, radio, etc. to two-way or multi-way communication methods such as social media [40,41]. Driven by innovation, the impact of information technology has influenced all walks of life, directly or indirectly [42]. One of the key factors that has facilitated this growth has been the ease, pace and cost efficient methods of reaching out to consumers through digital means, especially through social media. Studies show that, around the globe, social media campaigns have been deployed for sales, marketing, branding, political gain, social causes, etc. [43,44]. Businesses, as well as researchers, take to social media to seek out an understanding of the reactions, opinions or attitude of people towards new products, technologies, policies or any changes that might be forthcoming [45,46]. User sentiments, activity patterns, choices, interests, network connections, preferences etc., all play a major part in this process [47]. Prior to the launch of new technologies/products in the market that would bring about changes, users are primed for adopting them through stimulated dissemination of informative content. This dissemination is carried out through official social media handles and/or, now more so, through influencers with a larger number of followers on social media. The latter has proven to be more effective [48]. Further amplification of this takes place through propagation of word of mouth, via user activities, reaching out to an avalanche of audiences [49]. This indicates the vital role played by social media towards spreading public awareness of new technologies. There have been several studies which illustrate this important role of social media in energy as well as other sectors. To increase public awareness of the prominence of monitoring photovoltaic (PV) systems in the Netherlands, a social media campaign was organized, which proved to be highly useful [50]. Social media was found to be effective for strengthening public awareness about wildlife conservation [51]. Social media has been used extensively by emergency managers, particularly to warn people and help in the co-ordination of response and recovery, as well as for disaster and risk reduction [52]. It has also been used for enhancing public access to relevant medical information [53], exploring and enhancing public awareness regarding CO2 emissions and climate change [54], marketing, spreading messages, engaging and raising public awareness towards culture [55], raising awareness among all stake holders in higher education regarding green infrastructure and green finance in the school/college curriculum [56], and stimulating public awareness about space exploration and human space flight [57]. For the energy sector, the potential role of online social networks in disseminating energy-related information is imperative, deserves greater attention, and will involve enhanced roles for both organizations and opinion leaders [58]. Literature supports the fact that, users who are active on digital platforms have far more information on current and upcoming trends as compared to users who prefer to remain offline [59]. Industry 4.0, the digital revolution, is inching towards the peak and dawn of industry 5.0 and is nearing reality, which calls for technologies to be more personalized for each user [60]. In such circumstances, social media is preordained to play a more influential role for public awareness and outreach. Energies 2019, 12, 2759 6 of 27

3.3. Public Awareness and Acceptance Smooth diffusion of any innovation in the market depends mostly on social acceptance [61]. Social acceptance is categorized in various levels, each with its own perspective. Wuestenhagen et al. have given a categorization of social acceptance, which distinguishes three, occasionally symbiotic, categories: socio-political acceptance—this is concerned with the acceptance of technologies and policies by the people, key stake holders and policy makers; community acceptance: this concerns the acceptance of procedures by law and trust building; and market acceptance—this concerns the acceptance by the consumers, investors, and intra-firms [62,63]. Socio-political acceptance creates a platform of (un-)favorable conditions for the other two types of acceptance [63]. Support for the smart-meters at the socio-political level would make spreading awareness easier or educating the community and market regarding it. In this article, the awareness and acceptance of the consumer component of market acceptance is being discussed. Studies indicate that there is limited awareness, knowledge and interest regarding SM among consumers and customers have shown concerns regarding the acceptance of the same [7,21,64]. Customer concerns regarding SM include privacy & security of data, network connections in remote areas with lower or no mobile coverage, installation visits & doorstep selling, effects on health, disconnection of meters on a prepayment basis, and the option to switch between suppliers and keep the ‘smart functionality’ [1,10,20]. Presently, the majority of customers remain passive in the energy market. They do not use all the possibilities given to them by the current market designs and regulations. The new smart grid (SG) approach provides consumers with more involvement opportunities [9,32]. For example, consumers may get access to their current data about energy consumption by means of SM and SMP. They may also engage in energy generation by installing solar photovoltaic, micro water or wind turbines in their households. Finally, they may participate in the energy market by means of DSM/DR tools and approval to adjust their energy usage to the current needs of the electricity balancing system and real-time electricity prices. Even though SM could be seen as a step towards consumers gaining more access and control to their energy consumption [19,30], many studies indicate a low level of knowledge, awareness and engagement towards new solutions available in the energy markets [9,32,38,65]. In particular, many studies have shown a relatively low level of consumers’ knowledge and interest about smart metering [64]. Most consumers are not even familiar with the terms “smart grid” and “smart metering”. However, at the same time, they declare the willingness to save energy while being informed about electricity prices and ways of consumption reduction [8,9]. It was also found that only those consumers who were already interested or involved in energy savings were willing to use SM combined with some feedback devices, such as SMP, and learn from them [66]. Even if consumers have the initial interest in SMP, it is uncertain whether such engagement will persist over a longer period of time [14,30,67]. Among various behavioral obstacles of SM acceptance, the following have been mostly mentioned in the literature: bounded rationality of consumers; confusion of choice (lack of professional advice); negative perceptions, mainly because of a lack of knowledge and understanding; negative word of mouth (spreading negative information among social networks); disbelief in climate change; discomfort of usage; privacy and security concerns, among others [9,20,21,68]. The studies and various pilot programs have shown that, to increase social acceptance and awareness levels among electricity consumers, and enable the smooth diffusion of SM in the energy market, energy companies must pay attention to: simplicity of the proposed solutions (i.e., its interface; possibility to set one’s preferences, etc.) and emphasize the benefits, both on a social and individual level from various perspectives (i.e., environment protection, financial benefits, social engagement, etc.). At the same time, field experiments have proved the significant role of social influence in forming the awareness and engagement level towards energy efficiency issues, renewable energy sources and other innovations in the energy market, see [68–70] for more details. Energies 2019, 12, 2759 7 of 27

Finally, consumer preferences and communication channels play a vital role in the diffusion of information as well as addressing their concerns [71]. In 2009, a case study for understanding the impact of social capital on information diffusion, for adoption of household energy efficiency measures, was carried out among three communities in the UK. Results showed the standard campaigns account for approximately two-thirds of information seeking behavior, while it may not be addressing the rest of the community who prefer to receive information from the people they knew. It was also concluded that the likelihood of acceptance towards innovations increases, up to four times, in cases where the information is shared through personal contact. This implied that it is very important to tailor the diffusion or awareness campaigns according to communities’ communication channel preferences, to have more efficient results. [5]. Social media is a very convenient medium for creating personalized social influence that has proven to be effective for the diffusion of information or creating public awareness [48]. To conclude, a state-of-the-art literature review indicates that consumer acceptance of SM, as a part of a broader market acceptance, raises with the increase of consumers’ knowledge about the energy market and SG solutions (including SM, SMP and DSM/DR), with positive social influence (the desire to “be like others”, to compete with neighbors, and the need for one’s actions and decisions to be supported by one’s peers), and with the belief in positive result of cost-benefit analysis of consumer engagement in the energy market (taking into account financial and non-financial aspects, e.g., discomfort of usage). Although social acceptance towards SM has already been broadly investigated, we still see a gap in the literature. Namely, we believe that, presently, customers become “smart”, because of the access to various smart solutions on different markets. It is also true for the energy market. At the same time, smart customers are often present in social media, such as Facebook, Twitter or LinkedIn and are more aware regarding current and future technological trends. Various studies have already proven that dissemination of innovation through social media channels can bring effective results, see for example: [41,42,46,50,51]. To the best of our knowledge, the awareness and acceptance of SM among social media users have not been checked yet. Within our study, we want to fill this gap.

4. Methods

4.1. Data Collection and the Sample Data for this research was collected through an online questionnaire which was disseminated through social media platforms, namely Facebook, LinkedIn, Twitter, Reddit, WhatsApp and Instagram. Targeted audiences were residents of Poland above the age of 18. To reach the targeted audience, the organic method–manual sharing, paid method–targeted social media campaigns and inﬂuencer method–sharing by highly followed users to their audience were used. An option to ﬁll out the survey in Polish and English was given to facilitate both Polish nationals and Non-Polish nationals residing in Poland to respond to the survey. In total, N = 505 complete questionnaires had been collected, during February–March 2019. Out of these, 65 responses were in English and the rest were in Polish. Number of users visiting the online questionnaire were monitored via Google Analytics and showed that 1374 users visited the landing page, out of which 349 exited from the landing page itself, 852 visited the questionnaire in Polish language and 173 visited the questionnaire in the English language. To explore the determinants of the willingness to accept and install SM, we performed the following statistical analyses. First, in order to describe the sample and show relations between the chosen variables, descriptive statistics were used. Then, to explore the predictors of willingness to accept SM, ordered logit and tobit regression models have been used. All calculations have been performed in SPSS and Gretl statistical programs. Energies 2019, 12, 2759 8 of 27

4.2. Theoretical Background As the study is focused on dissemination of innovative electricity smart meters among social media users, we rely on the theoretical concepts and models, which emphasize the role of the social network in the diffusion of innovation. The background of the study is based on Roger’s model of innovation diffusion (DoI) [4]. This model pays a great deal of attention to communication channels in spreading news about market novelties. Rogers argues that diffusion is the process by which innovation is communicated over time among the participants in a social network. According to the DoI model, four main elements influence the spread of a new product, service or idea: the innovation itself, communication channels, time, and a social network (system). Rogers emphasizes that the innovation must be widely adopted in society in order to self-sustain and spread. This is the reason this model pays great attention to human capital. There is a point, within a certain rate of adoption, at which innovation reaches the so-called “critical mass”, which allows the innovation to spread further in the market [72]. In our survey, we expect that social media users are the first, more aware and interested towards market novelties and, second, more experienced with modern communication technologies. As electricity SM are often connected with smart metering information systems (SMP), available for free on mobiles, tablets or just on internet web pages, social media users would be more inclined to have some experience with them. This assumption is caused by the general higher level of interest and engagement towards smart products and services among this group of customers. Social media users are usually in favor of new technologies, prefer to stay on-line, and are mostly familiar with various mobile applications and modern communication channels, such as Facebook Messenger or Whatsapp. We believe that staying in touch with one’s social networks may bring positive results in spreading the information about new products and ideas, as already proven on various markets, see for example [42,44,49,59]. At the same time, we are quite skeptical as to whether the energy market, electricity, and innovative energy services, such as e.g., SM, SMP or DSM/DR are advertised and offered to the potential customers via social media. Key words, such as “smart metering”, “electricity bills”, “energy saving” and others connected with the energy market and the power system, do not belong to the ones being often mentioned by the social media users. Various studies show that people generally are not interested in electricity and do not discuss it frequently [34,73], which may also turn out to be true among social media users. Our research has been conducted in order to find out whether our expectations regarding the awareness and acceptance level of SM among social media users are right or not.

4.3. Research Framework As presented in Figure1, we have started with exploring the results of the previous studies regarding SM awareness and acceptance such as e.g., [16–18,30,32]. The literature review let us make some theoretical assumptions (see Section 4.2), raise some research questions and decide about the questionnaire design. We have decided to conduct the study particularly among social media users, as based on the literature, we could have assumed that consumers, who are actively using social media, are much more familiar with and open-minded towards new technologies to which SM and SMP (i.e., smart metering platforms) belong. The dissemination of the survey has also been done through social media channels. Finally we have analyzed the collected results by conducting ﬁrst reliability and validity tests, and then by exploring the data by means of descriptive statistics and regression models. Energies 2019, 12, 2759 9 of 27

Figure 1. Research framework.

5. Results and Discussion

5.1. Descriptive Statistics The definitions of the variables and their coding is presented in Table1. Apart from the demographics, (D1–D10) following blocks of questions were included in the survey: belongings (possession) of smart devices (B1–B11), knowledge (K1–K5) and source of information about SM (S1–S14), social influence (W1–W3), preferences regarding the role of the government in SM enrollment (G1–G3) and SM information systems (P1–P6), concerns about SM (F1–F4), and decisions relating to options to install SM (De1–De6). Demographics (D1–D10) The statistical description of the socio-economic variables (D1–D10) is presented in Table2. The majority of the sample is represented by young people between 18–35 years old, with at least a secondary level of education, living currently in a flat or apartment in a major Polish city. Most of the respondents do not have children and (72%) and 31% are still students. According to Eurostats, in Poland, 63% of the population is active on social media, out of which over 90% of social media users are between the age of 18 to 35. This accounts for the majority responses from young people. Belongings of smart devices and personal assets (B1–B11) The structure of respondents’ belongings of smart devices and personal assets is presented in Figure2. The majority of respondents possess a smartphone (99.8%), a laptop (94.9%), and a Wi-Fi connection at home (98%). On the other hand, the possession of an owned car, motorcycle, house or apartment is lower. The respondents were also asked whether they have some renewable energy sources installed at their household. Only 6.7% confirmed to have such energy sourced installed, and 90.5% denied. Energies 2019, 12, 2759 10 of 27

Table 1. Deﬁnitions of the variables and coding (N = 505).

Variable Code Description Gender D1 nominal variable Age D2 ordinal variable Marital status D3 nominal variable Education D4 ordinal variable Occupation/Employment D5 nominal variable Household’s income (in PLN D6 interval variable per month) Electricity bill (in PLN per month) D7 interval variable Household size D8 ordinal variable Number of children D81 ordinal variable Type of a house D9 nominal variable Place of a living D10 ordinal variable Belongings (of smart devices & (1) yes/(2) no, but I plan to buy it in a year B1–B11 personal assets) time/(3) no, and I do not plan to buy it Regular monitoring of energy usage A1 (1) yes/(2) no/(3) hard to say Renewable energy sources installed R1 (1) yes/(2) no/(3) hard to say at the household Source of information regarding S1–S13 nominal variable electricity (prices, new offers, etc.) Knowledge about SM K1–K5 (1) yes/(2) no/(3) hard to say Source of information regarding SM I1–I8 (1) yes/(2) no/(3) hard to say Social inﬂuence W1–W3 (1) yes/(2) no/(3) hard to say Preferences regarding the role of the G1–G3 (1) yes/(2) no/(3) hard to say government in SM enrollment Preferences regarding SM platforms P1–P6 (1) yes/(2) no/(3) hard to say Concerns about SM usage F1–F4 (1) yes/(2) no/(3) hard to say

Decisions to install SM De1–De6 (1) yes/(2) no/(3) hard to say

Table 2. Frequencies of the demographic variables (D1–D10).

Variable Frequencies female 61.4% Gender (D1) male 38.6% 18–25 years old 35.2% 26–35 years old 41.2% Age (D2) 36–45 years old 18.8% 46–55 years old 3.4% over 56 years old 1.4% single 41% married 28.5% Marital status (D3) divorced/separated 4.2% in a relationship 25.5% widowed 0.8% Energies 2019, 12, 2759 11 of 27

Table 2. Cont.

Variable Frequencies high class pass 22.5% bachelor complete 26.6% Education (D4) masters complete 44.7% PhD complete 6.6% full time job in private sector 35.45% full time job in state sector 8.51% part time job in private sector 2.97% part time job in state sector 1.39% Occupation/Employment (D5) own business 12.08% unemployed 2.77% student in college/university 16.24% high school student (above 18 years old) 4.95% others (combining 2 or 3 of upper categories) 15.64% less than 1000 PLN 6.5% 1001 to 2500 PLN 9.7% 2501 to 4000 PLN 17.6% 4001 to 5000 PLN 12.1% Household’s income (in PLN per month) (D6) 5001 to 6000 PLN 10.9% 6001 to 7000 PLN 7.5% 7001 to 8000 PLN 6.3% 8001 to 10,000 PLN 10.9% more than 10,000 18.4% 0 to 20 PLN 3.8% 21 to 40 PLN 5.0% 41 to 60 PLN 11.4% 61 to 80 PLN 12.4% 81 to 100 PLN 19.8% Electricity bill (in PLN per month) (D7) 101 to 150 22.8% 151 to 200 12.2% 201 to 250 5.4% 251 to 300 3.0% more than 300 PLN 4.0% M = 2.65, SD = 1.29 (where the integer number indicates Household size (D8) the number of family members) M = 1.45, SD = 0.83 (where (1) indicates no kids, (2) 1 Number of children (D81) kid, (3) 2 kids and so on) apartment/flat (in a 4 stored building) 64.8% apartment/flat (in a more than 4 stored building) 27.1% Type of a house (D9) house (only ground floor) 3.2% house (ground and upper floor) 5.0% village 8.5% city up to 50,000 inh. 10.7% Place of a living (D10) city 50,000 to 1,001,000 inh. 5.9% city 100,000 to 500,000 inh. 11.9% city more than 500,000 inh. 63% Energies 2019, 12, 2759 12 of 27

Figure 2. Distribution of respondents’ belongings of smart devices and personal assets.

Knowledge about SM (K1–K5) Half of the respondents (51.9%) do not know what SM is. Even less (67.7) claim not to have a SM installed at their household. Only 5.3% of respondents declared as having SM installed. The majority of respondents (60.2%) do not know whether they plan to have a SM installed in the future, or do not plan to install it in the future (28.9%). At the same time 30.7%, would like to have SM in their household. This group is primarily composed of those respondents who are aware what SM is. Such results are not surprising, as other reports (such as, [1,9,73] indicate low levels of knowledge about SG and SM. It is also connected with a rather low level of awareness and interest towards energy itself and the energy market [67]. Source of information about electricity and SM (S1–S14, I1–I8) and social inﬂuence (W1–W3) The most common sources of information about electricity (tariffs, offers, energy sellers, etc.) include energy companies (50.3%), TV news (49.3%), friends and colleagues (39.2%), Facebook (34.7%), radio (23.1%), newspapers (17.1%) and YouTube (10.4%). The remaining sources include ofﬁcial government internet platforms, workshops and educational campaigns, Facebook messenger, Twitter, Whatsapp and LinkedIn, but they have not been mentioned often. Some of the respondents use various blogs, professional journals and internet platforms to collect some recent information about the energy market and electricity. The distribution of the variables I1–I8 indicates that the majority of the respondents (even above 90%) neither have seen any information about SM in conventional sources of media, such as newspapers, radio or TV, nor have heard about the national enrollment of SM (90%). On average, 11% of respondents have seen some posts abut SM in social media, have discussed this topic with their friends and colleagues, or have searched for some information about SM on the Internet. Half of the respondents do not know whether their peers have SM installed, and 38% think they have not (M = 2.45; SD = 0.64; where (1) means yes, (2) no, and (3) hard to say). They are also not sure whether the peers’ decision to install it would encourage them to do the same (M = 2.35; SD = 0.77). Around 30% of respondents indicated that the recommendations of their peers regarding SM installation would motivate them to install SM as well. Energies 2019, 12, 2759 13 of 27

Preferences regarding SM (P1–P6) Of respondents, 70.5% do not regularly monitor their energy usage. Only one fourth of the sample admit to monitoring energy consumption. At the same time, the majority (more than 70%) believe that getting more details and real time information about energy consumption would be desired (M = 1.36; SD = 0.69). 68.7% declare that access to such information would enable them to decrease the wastage of energy usage. 69.3% of respondents would like to have the possibility to remotely control their electricity supply by means of a mobile app (M = 1.45; SD = 0.73). Concerns and fear about SM (F1–F4) We have also explored the respondents’ fears and concerns. 38.4% indicated having some concerns regarding the safety of private data transferred through SM (M = 1.87; SD = 0.79). 36% believe that ﬂuctuations in the energy rates would result in additional stress (M = 1.97; SD = 0.83). At the same time, most of the respondents do not think or are not sure whether SM may have a negative effect on their health (M = 2.44; SD = 0.57). Respondents are also not convinced that SM might have a negative impact on the accuracy of electricity bills (M = 2.43; SD = 0.66). Preferences regarding government’s role in SM enrollment (G1–G3) The desired role of the government in SM roll-out has also been investigated. 52% are not sure if the government should oblige the citizens to install SM at their household (M = 2.34; SD = 0.76). More than 70% would like to have a choice whether to install SM or not (M = 1.49; SD = 0.82). 34% of the respondents would protest if they were forced to install SM and to pay for the installation (M = 2.06; SD = 0.86).

Willingness to accept/ install SM (De1–De6) Finally, we have explored the respondents’ preferences regarding the decision to install SM. We have taken into account the most vital factors which, according to the literature review, inﬂuence the acceptance of SM. In particular, we have included in our analysis: ﬁnancial aspects, saving energy (and money), impact on one’s health, free advice from the energy company and safety and privacy concerns. The following options (De1–DE6) were considered, see Table3 and Figure3.

Figure 3. Decisions to install SM. Energies 2019, 12, 2759 14 of 27

Table 3. Willingness to accept/install SM (Notation: (1) yes, (2) no (3) hard to say).

Option Description 68.3% yes; 5.1% no; If SM can help you to save energy/money, would you decide to D 1 26.5% hard to say; e install it? (M = 1.58; SD = 0.88) 6.1% yes; 71.1% no; If SM can help you save energy/money, but may have bad D 2 22.8% hard to say; e impact on your health, would you decide to install it? (M = 2.16; SD = 0.51) If SM can help you save energy/money, and does not have an 42.4% yes; 19% no; impact on your health, but energy companies will know all the D 3 38.6% hard to say; e details about your energy consumption, would you decide to (M = 1.96; SD = 0.90) install SM? 20.2% yes; 30.9% no; Would you decide to install SM, if the representative of the energy D 4 48.9% hard to say; e supplier would visit your house and present you the beneﬁts? (M = 2.28; SD = 0.78) 17.8% yes; 29.1% no; Would you decide to install SM, if you have to pay for D 5 53.1% hard to say; e the installation? (M = 2.35; SD = 0.76) 37.7% yes; 10.1% no; Would you decide to install SM, if you did not have to pay for D 6 51.8% hard to say; e the installation? (M = 2.13; SD = 0.95)

5.2. Validity and Reliability Test of the Collected Data The common method variance (CMV) is often used to control over common method bias (CMB). CMB happens when variations in responses are caused by the collection instrument and/or method of collection rather than the actual predispositions of the respondents that the research questionnaire attempts to uncover. As mentioned in the works of Podsakoff et al., it may happen that the survey instrument, such as a questionnaire, introduces a CMB, detected by the CMV as variances. When CMB is large then it can make a major contribution to the measured effects. In such a case the results are then contaminated by the ’noise’ stemming from the biased instruments (see [74,75] for more details). One of the simplest ways to verify the potential problems with CMB is to use Harman’s single factor score, in which all items (measuring latent variables) are loaded into one common factor. If the total variance for a single factor is less than 50%, it suggests that CMB does not affect the data, hence the results. Our research instrument does not introduce CMB, as the variance is equal only to 9.688%. We have also controlled the reliability of our variables by using Cronbach’s Alpha test. We have checked the following sets of variables: B1–B11; K1–K5; I1–I8; W1–W3; G1–G3; P1–P6; F1–F4 and S1–S13. Apart from variables, regarding the concern (F) with α = 0497, source of information about SM (S) with α = 0.399 and governmental impact and role (G) with α = 0.388, the rest of the variables have the following values of parameter α: 0.570 for (B), 0.587 for (K), 0.512 for (W), 0.688 for (P), and 0.792 for (I). It allows us to aggregate the variables in larger constructs, by taking their mean value in the further step of the analysis. Such an aggregation allowed to limit the number of signiﬁcant variables for the alternative decisions and let us draw some more general conclusions.

5.3. Modeling of Willingness to Install SM We assume that acceptance of SM can be expressed by the willingness to install SM at the consumer’s household. In the study, we have examined various options that can differentiate the acceptance level as well as the explanatory variables inﬂuencing the consumer’s choice. In particular, we have distinguished separate models for each decision alternative De1–De6. Altogether, 6 Tobit models (with the threshold D i 2, excluding uncertainty in (3) “hard to say” answers) and 6 ordered e ≤ logit models (for De1, De2, De4, De6) were included in the analysis. Energies 2019, 12, 2759 15 of 27

In the ordered logit model, it is assumed that there is a linear relationship between the unobserved value of the willingness to install SM under a certain condition (Dei∗) and exogenous variables

Dei∗ = α + Xi β + εi, (1) where α is an intercept, Xi is a vector of exogenous variables excluding the constant and εi as a residual. The probabilities of belonging to a certain class D i 1, 2, 3 , can be deﬁned in the following way, e ∈ { } Prob(D i = 1) = Prob(D i 1) = Λ(α X β) e e ∗ ≤ 1 − i Prob(D i = 2) = Prob(1 < De µ ) = Λ(α X Λ(α X β) (2) e ∗ ≤ 1 2 − i − 1 − i Prob(D i = 3) = Prob(µ < De µ ) = Λ(α X β) Λ(α X β) e 1 ∗ ≤ 2 3 − i − 2 − i where Λ() is a logit function and αk’s are thresholds, such as α1 < α2 < α3. The ordered logit model is often used to explore and analyze the willingness to pay or to accept a certain product. Since the actual values of Dei∗ are not observed, (apart from the respondents who have already SM installed, we cannot be sure whether those who declare interest to install SM under a certain condition would install it), the regression (1) cannot be estimated directly and the model (2) could serve as an approximation of factor effect on willingness to accept/ install SM. The models’ predictive capabilities are as follows: for model De1 69.8% with Log likelihood 331.49 and Chi-square 113.22(20) with p = 0.000; for model D 2 73.8% with Log likelihood 337.94 − e − and Chi-square 160.14(20) with p = 0.000; D 3 56.7% with Log likelihood 473.20 and Chi-square e − 111.84(20) with p = 0.000; D 4 52.2% with Log likelihood 463.16 and Chi-square 140.87(20) with e − p = 0.000; D 5 57.9% with Log likelihood 427.62 and Chi-square 181.66(20) with p = 0.000; and D 6 e − e 66.9% with Log likelihood 402.57 and Chi-square 206.88(20) with p = 0.000. − On the other hand, the tobit model assumes that the class number D i 1, 2, 3 is a linear e ∈ { } function of some exogenous variables, as in (1), however, the relationship cannot be observed for individuals who are not sure whether they would install SM under a certain condition (these are the respondents who choose the answer (3) “hard to say”). As a result, the model becomes

Dei∗ Dei 2 Dei = ≤ (3) ( 0 Dei > 2 where Dei∗ is a latent variable described similarly to (1). The results of the ordered logit model, presented in Table4, indicate that:

Decision D 1 to install SM if it allows saving energy/ money correlates positively with preferences to • e possess more information on how to consume energy in more efficient ways and how to decrease energy wastage (P), household size (D8), and place of living (D10). It means that saving money due to the installation of SM was a motivation to citizens of larger cities rather than smaller ones that were living in bigger families. At the same time, this alternative correlates negatively with concerns about privacy and negative impact on one’s health (F), and number of information sources regarding electricity (S1–S13). The negative relation with parameter F is not surprising, because those who are in favour of alternative De1, are at the same time against the statements regarding fears and concerns about negative impact of SM on their wellness and health and safety of data protection. Similarly, those who have revealed more sources of information about SM, are more inclined to confirm this decision alternative. Decision D 2 to install SM if it allows saving energy/ money but at the same time may have a negative • e impact of one’s health, correlates positively with (P)—preferences to possess more information on how to consume energy in more efficient ways and with an impact of the government on their obligation to install SM (G). Surprisingly, this alternative is also positively influenced by confirmed concerns and fears (F). With two demographic variables: age (D2) and marital state (D3), the Energies 2019, 12, 2759 16 of 27

probability of confirming this decision increases as well. It indicates that older consumers who are married or in a relationship are more interested in SM than younger single individuals. Decision D 3 to install SM if it allows saving energy/money and at the same time does not have a • e negative impact of one’s health but energy companies can know all details about one’s energy use depends positively on: parameter P, positive social influence (parameter W), meaning that one’s peers support installation of SM or already have an SM installed. This alternative is also influenced by household’s income (D6) and household’s size (D8), meaning that larger families with smaller income are more likely to accept this alternative. Decision D 4 to install SM if a company representative visits your home and explains the benefits to you • e correlates positively with social influence (W) and low education level (D4) and negatively with age (D2) and income (D6)- this option is rather chosen by older people, with higher income. Decision D 5 to install SM if one has to pay for the installation correlates strongly with parameters: • e W, G and F. Those who confirm this alternative, care about peers’ support of SM installation and would prefer the government not to force them to install SM if they do not want it. Again, this option depends on concerns and fears about negative impact of SM, which means that those who have some concerns are likely to choose the payable installation. Decision D 6 to install SM if one does not have to pay for it is likely to be confirmed if one cares about • e social support (W), lack of governmental obligation to install (G) and preferences to possess some information about energy saving because of SM (P). It correlates negatively with the number of information sources about SM (S).

The tobit models, in Table5, additionally show that knowledge about SM (K) and lower income (D6) and older age (D2) impact decisions De1 positively, whereas a lack of regular monitoring of energy use (A1) decreases the probability of the confirmation of this decision. In the case of decision De3, which also concerns about SM installation (F) and access to some sources of information regarding SM (I), have a positive impact. Concerns about SM installation have occurred to be significant also in terms of decision making De4, De5, and De6. Social influence (W) matters in case of all decisions, apart from De2. To conclude, the following parameters have been revealed to be essential to influence a positive willingness to accept SM installation: peer’s support (W), lack of governmental obligation to install SM (G), and preferences to get some more information about SM that will enable energy conservation (P). Consumers do not want to be pressed and feel obligated to install SM but would rather prefer to have a choice. At the same time, they see the benefits connected with SM installation, such as better control over one’s energy consumption and electricity prices. Consumers rely on the opinions of their friends and colleagues regarding SM. Peer support seems to be important for them as well. On the other hand, dependent on the decision alternative Dei, different socio-economic variables seem to be important. For example, readiness to install SM, in many cases, correlates with age (D2). Older consumers are more willing to accept SM installation. Confirmation of decisions De1 and De3 increases with several household members. The impact on the household’s average income may be negative (in case of De1 and De3) or positive (in case De4) on SM installation and acceptance. The size of the city seems to matter only for the first alternative De1 and martial state (De3) influences decision De2. Surprisingly, possession of smart devices and personal assets (B) and knowledge about SM (K) does not show to be significantly correlated with SM installation, apart from decision De1 with which parameter (K) is significantly correlated. Also, the source of information about electricity (S) and access to information about SM (I) are important only in the case of the two alternatives, De1 and De6. Energies 2019, 12, 2759 17 of 27

Table 4. Estimation results for ordered logit model (N = 496).

De1 De2 De3 De4 De5 De6 D1 0.204 (0.23) 0.141 (0.22) 0.276 (0.19) 0.048 (0.19) 0.137 (0.2) 0.171 (0.21) − D2 0.193 (0.14) 0.283 * (0.14) 0.163 (0.12) 0.220 * (0.12) 0.156 (0.12) 0.067 (0.14) − − − − − D3 0.078 (0.09) 0.146 * (0.09) 0.094 (0.08) 0.041 (0.08) 0.049 (0.08) 0.051 (0.08) − − − − − D4 0.060 (0.14) 0.058 (0.13) 0.113 (0.12) 0.225 (0.12) 0.194 (0.12) 0.038 (0.13) − D5 0.009 (0.02) 0.004 (0.02) 0.020 (0.02) 0.004 (0.02) 0.025 (0.02) 0.013 (0.02) − − − − D6 0.052 (0.04) 0.051 (0.04) 0.071 * (0.04) 0.105 ** (0.04) 0.006 (0.04) 0.013 (0.04) − − − D7 0.045 (0.06) 0.073 (0.06) 0.009 (0.05) 0.048 (0.05) 0.010 (0.05) 0.046 (0.06) − − − D8 0.198 ** (0.1) 0.082 (0.1) 0.143 * (0.08) 0.010 (0.08) 0.045 (0.08) 0.039 (0.09) − − − − D9 0.167 (0.15) 0.049 (0.14) 0.089 (0.12) 0.107 (0.12) 0.121 (0.13) 0.046 (0.13) − − D10 0.159 * (0.09) 0.003 (0.08) 0.010 (0.07) 0.051 (0.07) 0.069 (0.08) 0.085 (0.08) − B 0.057 (0.4) 0.202 (0.4) 0.029 (0.3) 0.167 (0.3) 0.096 (0.32) 0.036 (0.32) − − − − K 0.540 (0.41) 0.299 (0.38) 0.188 (0.33) 0.032 (0.32) 0.007 (0.33) 1.121 (0.23) − W 0.033 (0.26) 0.005 (0.24) 0.808 *** (0.22) 1.212 *** (0.22) 1.279 *** (0.23) 1.121 *** (0.23) G 0.383 * (0.21) 0.679 *** (0.21) 0.036 (0.18) 0.149 (0.18) 0.578 *** (0.19) 0.613 ** (0.20) P 1.586 *** (0.25) 0.701 *** (0.23) 0.790 *** (0.21) 0.245 (0.21) 0.125 (0.22) 0.680 ** (0.23) − F 0.676 ** (0.27) 0.444 * (0.26) 0.214 (0.22) 0.372 (0.22) 1.018 *** (0.23) 0.167 (0.66) − I 0.417 (0.72) 0.001 (0.67) 0.854 (0.63) 0.493 (0.6) 0.088 (0.6) 0.167 (0.67) − SO 1.704 * (0.91) 0.644 (0.9) 0.807 (0.77) 0.741 (0.78) 0.744 (0.79) 0.443 * (0.35) − − R1 0.179 (0.38) 0.056 (0.37) 0.007 (0.35) 0.014 (0.33) 0.090 (0.37) 0.443 (0.35) − − A1 0.293 (0.23) 0.238 (0.22) 0.035 (0.19) 0.162 (0.2) 0.157 (0.2) 7.421 (1.63) − − − cut1 3.655 ** (1.66) 0.278 (1.54) 5.8551 *** (1.52) 2.939 ** (1.40) 4.835 *** (1.5) 1.677 *** (1.69) − cut2 3.965 ** (1.66) 4.147 *** (1.55) 6.751 *** (1.52) 4.610 *** (1.41) 6.618 *** (1.51) 4.421 *** (1.63) LL 331.49 337.95 473.20 463.16 427.62 402.57 − − − − − − Note: *** p < 0.001, ** p < 0.01,* p < 0.05 (two-tailed test); Standard errors in brackets. LL stands for Log-Likelihood.

Table 5. Estimation results for tobit model (N = 496).

De1 De2 De3 De4 De5 De6 D1 0.163 (0.15) 0.038 (0.17) 0.147 (0.11) 0.046 (0.09) 0.015 (0.08) 0.056 (0.08) − D2 0.155 * (0.09) 0.241 ** (0.12) 0.084 (0.07) 0.103 * (0.06) 0.107 ** (0.06) 0.039 (0.05) − − − − − D3 0.031 (0.06) 0.128 * (0.07) 0.025 (0.04) 0.019 (0.04) 0.011 (0.03) 0.017 (0.03) − − − − − D4 0.017 (0.09) 0.003 (0.11) 0.064 (0.07) 0.097 * (0.06) 0.063 (0.05) 0.010 (0.05) − − − D5 0.005 (0.01) 0.008 (0.02) 0.006 (0.01) 0.004 (0.01) 0.010 (0.01) 0.002 (0.01) − − − − D6 0.055 * (0.03) 0.002 (0.03) 0.021 (0.02) 0.049 *** (0.02) 0.003 (0.02) 0.008 (0.02) − − D7 0.022 (0.04) 0.052 (0.04) 0.011 (0.03) 0.005 (0.02) 0.006 (0.02) 0.017 (0.02) − − − D8 0.040 (0.06) 0.051 (0.07) 0.031 (0.05) 0.041 (0.04) 0.032 (0.03) 0.029 (0.03) − − − D9 0.148 (0.1) 0.053 (0.1) 0.028 (0.1) 0.034 (0.06) 0.005 (0.05) 0.014 (0.05) − − D10 0.112 * (0.06) 0.009 (0.06) 0.014 (0.04) 0.005 (0.03) 0.028 (0.031) 0.037 (0.03) − − − B 0.128 (0.23) 0.218 (0.27) 0.072 (0.18) 0.076 (0.15) 0.024 (0.13) 0.002 (0.13) − − K 0.511 * (0.27) 0.073 (0.29) 0.232 (0.19) 0.045 (0.15) 0.056 (0.14) 0.203 (0.14) − W 0.235 (0.17) 0.011 (0.19) 0.448 *** (0.13) 0.472 *** (0.10) 0.361 *** (0.09) 0.501 *** (0.09) G 0.271 ** (0.13) 0.634 *** (0.16) 0.116 (0.10) 0.135 (0.08) 0.252 *** (0.08) 0.229 *** (0.07) P 0.891 *** (0.16) 0.436 ** (0.18) 0.107 (0.12) 0.061 (0.1) 0.203 ** (0.09) 0.083 (0.08) − − F 0.274 (0.17) 0.073 (0.2) 0.385 *** (0.13) 0.322 *** (0.11) 0.502 *** (0.1) 0.194 ** (0.09) − I 0.089 (0.46) 0.171 (0.49) 0.635 * (0.37) 0.181 (0.28) 0.288 (0.26) 0.121 (0.25) SO 0.907 (0.58) 0.234 (0.65) 0.679 (0.44) 0.002 (0.36) 0.052 (0.32) 0.739 ** (0.32) − − R1 0.179 (0.24) 0.247 (0.27) 0.190 (0.19) 0.081 (0.15) 0.035 (0.14) 0.109 (0.14) − − − − A1 0.255 * (0.15) 0.219 (0.17) 0.009 (0.11) 0.115 (0.09) 0.017 (0.08) 0.041 (0.08) − − − const 1.567 (1.08) 0.654 (1.15) 2.477 *** (0.88) 0.305 (0.66) 0.441 (0.61) 1.054 * (0.6) − − − − LL 344.08 335.59 433.85 476.95 464.31 446.32 − − − − − − Chi2(20) 70.92 p < 0.001 41.35 p = 0.003 71.65 p < 0.001 74.95 p < 0.001 108.98 p < 0.001 134.95 p < 0.001 Note: *** p < 0.001, ** p < 0.01,* p < 0.05 (two-tailed test); Standard errors in brackets. LL stands for Log-Likelihood; Chi2 stands for Chi-square test.

The detailed results, including all of the individual explanatory variables, of the ordered logit and tobit models (the latter with the threshold D i 2) are presented in the AppendixA in Tables A1–A3. e ≤ In each column, the upper rows show the results of the ordered logit regression and the bottom rows represent the results of the tobit model. For models De3 and De5, only tobit models are provided. Energies 2019, 12, 2759 18 of 27

From the demographic data set, age (D2), average income (D6) and household size (D8) are significantly correlated with willingness to accept SM. Older consumers, with higher incomes and larger families are more likely to accept SM installation. The correlation between willingness to accept/install SM and the possession of smart devices and personal assets is less clear. Consumers who do not have WI-FI/Internet connection in their household are less likely to accept alternatives De4, De5, De6. Positive influence of having a laptop has occurred to be significant for alternative De6, and partly for De2 and De4. In case of possessing knowledge about SM, mainly the desire to have SM installed in one’s house (K4 and K5) significantly increases the probability of willingness to accept alternatives: De1, De2, De5 and De6. Regular monitoring of energy usage (A1) declines the probability of confirmation of decision De1 and increases in the case of decision De6. Having multiple sources of information about SM, both in social media, on the Internet and in conventional sources, such as TV or radio, seems to have an impact, especially on first four alternatives. In particular, those who have not seen a post about SM in social media and have not talked about this with their peers are less likely to confirm alternatives De1, De2, De3 and, partly, De4. At the same time, those who have ever looked for some information about SM on the Internet are highly likely to accept the options De1 and De3. Receiving positive advice and recommendations from one’s peers regarding the installation of SM has been revealed as a significant factor, explaining the likelihood of all of the alternatives. Consumers, whose friends and colleagues have recommended SM installation, are much more likely to accept and install it in their households under various conditions proposed in alternatives De1–De6. In the case of governmental impact on SM installation, most of the respondents want to have a choice to accept or decline SM installation. This opinion increases the probability of confirming most of the alternatives, apart from De3 and De4. The following preferences (P) correlate with the willingness to accept/install SM, especially in alternatives De1, DE2 and De3: receiving more details about energy consumption (P1), the ability to remotely turn on/off the electricity supply in one’s household (P5), and the desire to have a dynamic electricity tariff with different electricity prices in load peaks and to consume more energy when it is cheaper (P6). Only in alternative De5 do none of the preferences show as significant.

5.4. Final Discussion The results show that social media users between the age of 18 to 35 were the most active in responding to the questionnaire. This also compliments the literature which stats that this age group of users are the largest in number on social media. As presumed before carrying out the survey, the respondents are open to new technologies and are users of smart phones, laptops and internet. It was surprising though, to find out that social media users, described in the literature to be really active in checking their notification and news updates, are not very keen on monitoring their energy consumption regularly. At the same time they rather agree that real time information of energy consumption would lead to energy saving at their households. More than half of the respondents do not know about SM and even more do not know whether they would be installing SM at home in the future. This indicates that if the consumers’ awareness regarding SM increased, they would be open towards using it for saving energy. Lack of information in conventional communication channels as well as in social media, is one of the responsible factors for low levels of SM awareness among consumers. Respondents also indicated that discussion among peers regarding SM is not that common, which is another reason for lower diffusion of SM, especially that even one third of the respondents agreed that a recommendation from their peer would motivate them to install SM. Consumers also indicated that they fear about their data privacy and increase in stress levels (because of additional energy consumption information) through the use of SM. These fears would be addressed if the users have proper knowledge regarding SM, its functions and benefits. There are studies in the literature which indicate the fear of health problems and impact on accuracy of billing, but the results of this Energies 2019, 12, 2759 19 of 27 study do not support those findings as majority of the respondents do not think that it would be a factor of concern to them. Energy companies and the government need to proceed with caution and raise the awareness level of consumers regarding the benefits of SM, so that they accept it voluntarily rather than forcing it on them. This is because over two thirds of the respondents indicated that they should have the choice of whether or not to install SM at their houses. Over one third of the respondents also expressed that they would protest if they were not given a choice for it. Apart from all these factors which warrant for large scale awareness campaign regarding SM among consumers, this study also drew some conclusions from the question where users were given different conditions of acceptance to install SM. Majority of the users were positive to install SM if they would be able to save energy and decrease their expenses. The acceptance level decreases if consumers’ information about their energy consumption can be known by the energy companies. The numbers fall even further where consumers were asked if they would accept the use of SM if they had to pay extra for its installation and around two thirds of the users responded negatively to the prospect of accepting SM if there were any ill effects to their health because of it.

6. Conclusions and Recommendations

6.1. Conclusions Within this study, we wanted to explore the awareness and acceptance level of SM among social media users. Although various aspects of consumers’ acceptance and engagement towards innovative, smart energy services, such as micro technologies based on renewable energy sources, dynamic electricity tariffs, green electricity tariffs or even SM and SM information systems (SMP) have already been investigated in the literature, to the best of our knowledge, the awareness and acceptance of SM among social media users has not been explored yet. The literature also supports that, presently, social media platforms, such as Facebook or Twitter, have become very important communication channels in social systems. At the same time, communication channels play a vital role in every diffusion process of innovation (see DOI model in Section4), which is also true for the energy market and innovative energy services being launched and offered to electricity consumers. Even though our survey has been limited to one country, we have been able to draw some general conclusions. In particular, we have observed that consumers’ willingness to accept/install SM depends on consumers’ age, income and household size. Generally, older consumers with higher incomes and larger families are more likely to accept SM installation. Willingness to accept/install SM is also influenced by peer’s support (people care about peer’s recommendations, experiences and advice), lack of governmental obligation to install SM (people do not want to be forced to install SM and pay for it) and preferences to receive some additional information about one’s energy consumption profile and real-time electricity prices through SM and SMP. This last parameter should be emphasized because it shows that social media users, who are generally keen on modern communication technologies, could be interested in using SMP mobile apps, connected with their SM, to remain informed about their current energy consumption. Polish consumers hold similar concerns regarding SM as consumers from other countries. Namely, they are afraid about the safety of their private data being transferred through SM to energy companies. They are also worried about the impact of dynamic electricity prices on their comfort of living. Surprisingly, possession of smart devices and personal assets, sources of information about electricity, and access to information about SM have occurred to matter only in case of two installation alternatives, both of which are connected with financial aspects, namely: energy and money savings, and lack of consumers’ payment for SM installation. Such a result may be caused by the fact that electricity, in general, and SM or SMP in particular, are not common subjects spread throughout social media. At the same time, consumers interested in some aspects of the energy market, e.g., in comparing the offers of energy companies, admit to looking for such data on the Internet. They also ask their Energies 2019, 12, 2759 20 of 27 friends and colleagues for opinions and advice. Additionally, Facebook itself has been revealed as one of the most common sources of information, especially for the younger group of consumers, who generally rely on the information found on various social media platforms. This finding shows some optimistic perspective on how social media could be used in the dissemination process of energy efficiency, energy conservation and other energy related subjects. Our study has shown that, presently, SM and SMP are mostly not advertised through conventional or social media. Consumers do not have a chance to view informative posts or messages, explaining the concept of SM enrollment, its advantages and potential benefits for electricity consumers. It is one of the reasons for a rather low level of consumers’ awareness. At the same time, we have observed a general positive interest and a desire to have SM installed in ones’ household. We believe that there is still plenty to be done in order to increase consumers’ awareness and knowledge about SM and SMP. Ideas for how it could be achieved are presented below.

6.2. Recommendations for Increasing Consumer Acceptance through Social Media The government as well as energy companies must put in a more vigorous effort to increase public awareness regarding SM. Based on the results of this study, ﬁndings on awareness and acceptance of smart meters in the literature, and studies based on diffusion of innovation through social media, the following are recommendations that could be fruitful if implied by the government and energy companies.

Facebook and YouTube were mentioned by 34.7% and 10.4% of the respondents as a source of • receiving information about various aspects of electricity. The remainder of the social platforms were not that popular. The authors briefly browsed Facebook, LinkedIn, Twitter, YouTube and Instagram accounts of energy companies in Poland. It was found that Facebook and YouTube had some content related to smart meters, but the rest did not. Additionally, platforms such as Instagram, which are gaining popularity, did not show presence of energy companies. To increase this number, social media campaigns must be more diverse, in terms of content type, theme and means. It has been proven by several studies that the most effective content types on social networks • are photos, graphics, illustrations and motion graphics. The use of these means would help in improving the outreach as well as understanding of social media users about SM. Themes of the campaign are also very important and, while planning the campaign, the platform • of dissemination and targeted age groups must be kept in mind. Through the results of the study, some of the recommended themes are as follows: addressing the knowledge about SM (What are smart meters, its function, benefits, myths, long term implications, financial impacts and so on.); demonstrating the controls users get through installation of SM (monitoring energy use, reducing bills, as well as wastage of energy, remotely controlling energy usage with real time information and so on.); addressing the concerns about SM (security of personal data, fluctuations in the energy rates, health effects, accuracy of billing, etc.); and social discussions through experts and current users of SM (positive feedback/experiences, expert advice/assurances and so on). Through the literature as well as the results, it is evident that individuals are more open towards • accepting information received though the people they know. Hence, we recommend that, instead of running paid campaigns on social media, organic campaigns and influencer campaigns would be more effective. Through organic campaigns, the users will receive information through their peers and connections, although the effect would be limited in terms of reach and would depend on the network of the page or profile where the content is posted. Through influencer campaigns, the users would be receiving information from a person well-known to them and, hence, the impact of information, as well as its lasting effect, would be greater as compared to an unknown source. Energies 2019, 12, 2759 21 of 27

6.3. Limitations of the Study and Future Work The conducted survey has some limitations. Firstly, the study was geographically limiting as it was only conducted in Poland. Furthermore, the study was also limited by the means of dissemination of filling out the web-based survey, it only focused on social media users. Hence, the sample does not account for consumers who are not active on social media. Further studies could be aimed at more diverse demographics and at a wider audience. The authors tried to include as many factors as possible, which have an influence on consumers towards their willingness or decision to install smart meters. Future studies might add more factors, such as ownership or tenancy of the house, whether the respondent is the decision maker, their field of education and profession. A similar study could be carried out for understanding public awareness and consumer acceptance of other innovations and novelties. It would also be interesting to measure the effects of recommendations suggested by the authors in this article.

Author Contributions: Y.C. conceived and designed the survey; A.K.-P. reviewed the design of the survey; Y.C. performed the survey in social media; A.K.-P. analyzed the data; A.K.-P. and Y.C. reviewed the literature; A.K.-P. and Y.C. drafted, reviewed and edited the paper. Funding: This work was supported by the National Science Center (NCN, Poland) by grant no. 2016/23/B/HS4/00650. Acknowledgments: The authors are grateful to the reviewers for their constructive and insightful remarks and suggestions. The authors would also like to thank Marta Pytel for checking and proofreading the language of the manuscript. The authors also acknowledge the support by Monika Czaplicka from WoBuzz in Poland for the dissemination of the survey questionnaire to collect responses through social media. Conﬂicts of Interest: The authors declare no conﬂict of interest.

Abbreviations The following abbreviations are used in this manuscript:

SG smart grids SM electricity smart meters SMP smart metering platform (SM information systems) DSM/DR Demand Side Management& Demand Response tools DoI diffusion of innovation model

Appendix A. Estimation Results for Ordered Logit and Tobit Models

Table A1. Estimation results of the ordered logit and tobit models (part 1).

De1 De2 De3 De4 De5 De6 0.365 (0.294) 0.106 (0.259) 0.001 (0.230) 0.261 (0.252) D1 0.050 (0.100) − 0.082 (0.183) 0.056 (0.059) 0.305 (0.493) 0.189 (0.168) 0.063 (0.108) 0.328 * (0.178) 0.361 ** (0.164) 0.255 * (0.143) 0.063 (0.163) D2 − − 0.220 *** (0.065) − 0.131 (0.118) 0.080 ** (0.037) 0.543 * (0.301) − 0.129 (0.168) 0.026 (0.072) − − − 0.037 (0.106) 0.107 (0.096) 0.030 (0.084) 0.068 (0.093) D3 − − 0.068 * (0.037) − 0.026 (0.066) − 0.007 (0.022) 0.037 (0.184) − 0.003 (0.065) 0.0114 (0.041) − 0.0003 (0.172) 0.076 (0.154) 0.135 (0.134) 0.081 (0.152) D4 − 0.007 (0.061) 0.292 ** (0.117) 0.004 (0.036) 0.279 (0.359) 0.038 (0.107) 0.019 (0.066) − − − − 0.017 (0.023) 0.008 (0.019) 0.012 (0.017) 0.001 (0.018) D5 − 0.011 (0.007) 0.005 (0.012) 0.002 (0.004) 0.031 (0.032) − 0.0123 (0.012) − 0.0033 (0.008) − 0.074 (0.055) 0.104 ** (0.049) 0.106 ** (0.043) 0.056 (0.048) D6 0.052 *** (0.019) − 0.072 ** (0.036) − 0.014 (0.011) 0.491 *** (0.156) 0.021 (0.032) − 0.0127 (0.207) − − 0.089 (0.071) 0.118 * (0.066) 0.003 (0.062) 0.019 (0.063) D7 − 0.021 (0.026) − 0.005 (0.048) − 0.013 (0.015) 0.036 (0.103) − 0.018 (0.044) − 0.004 (0.027) − 0.171 (0.114) 0.039 (0.100) 0.021 (0.091) 0.053 (0.096) D8 − 0.078 ** (0.039) − 0.066 (0.077) − 0.0555 ** (0.023) 0.253 (0.188) − 0.176 *** (0.064) 0.083 * (0.043) − − − − Energies 2019, 12, 2759 22 of 27

Table A1. Cont.

De1 De2 De3 De4 De5 De6 0.312 * (0.188) 0.134 (0.158) 0.172 (0.136) 0.030 (0.152) D9 − 0.047 (0.063) 0.147 (0.114) − 0.067 * (0.036) 0.359 (0.337) 0.225 * (0.118) − 0.049 (0.068) − − 0.163 (0.114) 0.044 (0.095) 0.027 (0.084) 0.039 (0.094) D10 0.010 (0.037) 0.082 (0.070) 0.023 (0.022) 0.340 ** (0.172) 0.038 (0.059) − 0.005 (0.039) 00.303 (0.231) 0.315 (0.193) 0.169 (0.161) 0.011 (0.184) B1 0.011 (0.072) 0.211 (0.136) − 0.023 (0.043) 0.082 (0.288) − 0.044 (0.117) − 0.022 (0.077) − − 0.073 (0.169) 0.016 (0.03) 0.071 (0.132) 0.059 (0.146) B2 − 0.104 * (0.057) 0.168 (0.109) 0.004 (0.034) 0.592 * (0.338) − 0.004 (0.097) − 0.023 (0.062) − 0.146 (0.162) 0.014 (0.147) 0.318 ** (0.131) 0.182 (0.145) B3 − − 0.046 (0.058) − 0.034 (0.109) − 0.036 (0.034) 0.423 (0.275) − 0.268 *** (0.102) − 0.071 (0.063) − − − − 0.218 (0.251) 0.344 (0.226) 0.223 (0.188) 0.287 (0.211) B4 − 0.111 (0.086) − 0.015 (0.149) − 0.075 (0.052) 1.296 ** (0.547) 0.471 ** (0.184) 0.227 ** (0.098) − − − 0.048 (0.149) 0.255 * (0.134) 0.051 (0.114) 0.107 (0.127) B5 − 0.015 (0.052) − 0.084 (0.09) 0.0248 (0.030) 0.116 (0.274) − 0.073 (0.089) − 0.042 (0.055) − 0.126 (0.320) 0.915 *** (0.314) 0.571 * (0.292) 0.786 ** (0.329) B6 − 0.107 (0.135) 0.419 (0.113) 0.014 (0.069) 0.379 (0.503) 0.353 (0.245) 0.336 ** (0.155) − 0.111 (0.547) 0.119 (0.261) 0.115 (0.589) 0.981 (0.981) B7 0.837 (0.377) 1.639 (0.655) 0.155 (0.260) 0.193 (0.384) 0.252 (0.560) − 1.408 (0.366) 0.049 (0.145) 0.107 (0.128) 0.037 (0.112) 0.106 (0.124) B8 0.008 (0.049) − 0.102 (0.092) − 0.007 (0.0294) 0.968 *** (0.348) − 0.014 (0.081) 0.029 (0.054) − − − 0.159 (0.525) 0.115 (0.465) 1.627 *** (0.435) 1.140 *** (0.431) B9 − 0.0236 (0.172) − 0.053 ** (0.264) − 0.009 (0.108) 1.140 (0.797) − 0.832 *** (0.254) − 0.387 ** (0.188) − − − 0.183 (0.155) 0.265 * (0.136) 0.052 (0.117) 0.019 (0.130) B10 0.045 (0.052) 0.053 (0.096) 0.0388 (0.031) 0.214 (0.274) 0.263 *** (0.088) − 0.0204 (0.056) − 0.255 * (0.151) 0.272 ** (0.137) 0.075 (0.117) 0.081 (0.128) B11 − − 0.041 (0.052) 0.047 (0.095) 0.044 (0.031) 0.237 (0.249) − 0.068 (0.088) − 0.074 (0.056) − − − 0.186 (0.226) 0.016 (0.203) 0.089 (0.173) 0.022 (0.191) K1 − 0.108 (0.081) 0.058 (0.146) 0.033 (0.047) 0.285 (0.405) 0.159 (0.139) − 0.005 (0.087) − 0.262 (0.316) 0.211 (0.283) 0.515 ** (0.240) 0.098 (0.277) K2 − − 0.109 (0.108) − 0.196 (0.212) − 0.006 (0.064) 0.611 (0.515) − 0.006 (0.184) − 0.074 (0.118) − − − − 0.279 (0.796) 0.736 (0.602) 0.788 (0.506) 0.076 (0.559) K3 − 0.069 (0.223) 0.048 (0.328) − 0.051 (0.136) 0.061 (0.781) 0.663 * (0.355) 0.087 (0.240) 0.382 (0.264) 0.805 *** (0.232) 0.296 (0.189) 0.241 (0.214) K4 − 0.073 (0.084) 0.138 (0.142) − 0.051 (0.049) 1.495 *** (0.430) − 0.027 (0.137) − 0.099 (0.088) − − − 0.853 *** (0.231) 0.083 (0.175) 0.011 (0.149) 0.665 *** (0.163) K5 − 0.044 (0.065) 0.247 ** (0.117) 0.146 *** (0.039) 0.098 (0.257) 0.126 (0.106) 0.299 *** (0.070) − Note: *** p < 0.001, ** p < 0.01,* p < 0.05 (two-tailed test); Standard errors in brackets. LL stands for Log-Likelihood. The upper rows represent the results of the ordered logit model, whereas the bottom rows show the results of the tobit model.

Table A2. Estimation results of the ordered logit and tobit models (part 2).

De1 De2 De3 De4 De5 De6 0.619 ** (0.291) 0.305 (0.248) 0.142 (0.227) 0.502 ** (0.253) A1 − − 0.142 (0.096) − 0.217 (0.182) 0.116 ** (0.057) 0.981 (0.617) − 0.210 (0.170) 0.226 ** (0.107) − 0.918 ** (0.402) 0.724 ** (0.339) 0.007 * (0.281) 0.292 (0.306) I1 − − 0.242 * (0.126) − 0.307 (0.213) 0.171 ** (0.076) 2.490 ** (0.953) − 0.093 (0.198) 0.015 (0.134) − − − 1.115 (0.794) 0.080 (0.571) 0.072 (0.505) 0.691 (0.543) I2 0.333 (0.126) − 0.430 (0.382) 0.1888 (0.128) 0.217 (0.897) 0.054 (0.333) − 0.126 (0.222) − 1.389 *** (0.529) 0.716 (0.447) 0.060 (0.376) 0.546 (0.381) I3 − − 0.551 *** (0.173) 0.119 (0.288) − 0.290 *** (0.101) 0.296 (0.717) − 0.093 (0.272) − 0.227 (0.174) − − − 1.122 * (0.197) 0.495 (0.498) 0.283 (0.188) 0.157 (0.475) 1.015 * (0.544) I4 − − 0.142 (0.353) 0.193 * (0.115) 0.282 (0.782) 0.336 (0.292) − 0.272 (0.198) 0.745 (0.572) 0.887 * (0.489) 0.194 (0.182) 0.263 (0.438) 0.385 (0.497) I5 0.474 * (0.287) − 0.058 (0.113) 1.036 (0.735) 0.055 (0.273) − 0.231 (0.194) − 1.051 (0.812) 1.173 * (0.660) 0.865 (0.609) 0.222 (0.631) I6 − 0.051 (0.248) 0.202 (0.403) 0.086 (0.149) 1.207 (0.385) − 0.138 (0.359) 0.192 (0.250) − Energies 2019, 12, 2759 23 of 27

Table A2. Cont.

De1 De2 De3 De4 De5 De6 1.560 *** (0.591) 0.499 (0.455) 0.211 (0.358) 0.272 (0.393) I7 0.356 ** (0.175) − 0.114 (0.277) − 0.284 *** (0.102) 0.615 (0.775) 0.443 (0.274) 0.178 (0.177) − − − 0.065 (0.489) 0.127 (0.446) 0.336 (0.415) 0.284 (0.452) I8 − 0.195 (0.179) 0.519 * (0.301) − 0.098 (0.103) 0.274 (0.763) 0.443 (0.281) − 0.0109 (0.187) − − 0.149 (0.267) 0.150 (0.240) 0.211 (0.205) 0.250 (0.222) W1 − 0.036 (0.090) − 0.312 * (0.174) 0.027 (0.054) 0.346 (0.388) 0.087 (0.149) 0.106 (0.094) − − − 0.193 (0.253) 0.152 (0.204) 0.246 (0.169) 0.139 (0.196) W2 − − 0.106 (0.075) 0.201 (0.131) − 0.021 (0.045) 0.146 (0.367) 0.124 (0.119) 0.122 (0.078) − − − 0.315 * (0.188) 0.178 (0.151) 0.781 *** (0.132) 0.879 *** (0.147) W3 0.207 *** (0.055) 0.617 *** (0.104) 0.041 ** (0.034) 0.766 ** (0.334) 0.509 *** (0.094) 0.382 *** (0.059) 0.523 (0.469) 0.136 (0.417) 0.146 (0.377) 0.923 ** (0.404) R1 − − 0.024 (0.166) 0.386 (0.268) 0.06 (0.097) 1.649 ** (0.704) − 0.435 (0.280) 0.3111 * (0.179) − − 0.029 (0.222) 0.156 (0.183) 0.048 (0.158) 0.007 (0.173) G1 − 0.081 (0.068) 0.089 (0.125) 0.039 (0.041) 1.309 *** (0.434) − 0.084 (0.111) − 0.032 (0.072) − − − − 0.527 *** (0.164) 0.309 ** (0.151) 0.061 (0.139) 0.248 (0.159) G2 0.042 (0.062) 0.401 *** (0.139) 0.1333 *** (0.036) 1.521 *** (0.532) 0.03 (0.106) 0.175 ** (0.073) − − 0.315 ** (0.156) 0.076 (0.142) 0.084 (0.124) 0.270 * (0.139) G3 − 0.123 ** (0.056) 0.148 (0.108) 0.069 ** (0.032) 0.220 (0.275) − 0.041 (0.097) − 0.066 (0.060) − − 0.556 *** (0.185) 0.638 *** (0.183) 0.097 (0.168) 0.240 (0.198) P1 0.282 *** (0.083) 0.101 (0.150) 0.163 ** (0.045) 0.610 (0.435) 0.123 (0.146) − 0.173 * (0.094) 0.101 (0.219) 0.042 (0.202) 0.349 ** (0.173) 0.071 (0.198) P2 0.113 (0.085) − 0.016 (0.167) − 0.039 (0.048) 0.735 (0.507) 0.168 (0.149) 0.030 (0.092) − 0.189 (0.183) 0.372 ** (0.179) 0.103 (0.153) 0.082 (0.166) P3 − 0.027 (0.068) 0.043 (0.135) − 0.042 (0.040) 0.811 ** (0.401) − 0.244 * (0.148) 0.031 (0.075) − 0.043 (0.184) 0.125 (0.156) 0.288 ** (0.139) 0.075 (0.156) P4 0.049 (0.062) 0.151 (0.119) − 0.007 (0.036) 0.337 (0.286) 0.292 *** (0.109) 0.046 (0.066) − − 0.880 *** (0.191) 0.402 ** (0.175) 0.169 (0.154) 0.139 (0.176) P5 0.178 ** (0.074) − 0.071 (0.142) 0.213 *** (0.042) 1.293 ** (0.635) 0.197 (0.144) 0.174 ** (0.083) 0.034 (0.170) 0.211 (0.150) 0.015 (0.131) 0.368 ** (0.144) P6 0.097 * (0.057) 0.082 (0.111) 0.006 (0.034) 0.585 ** (0.275) 0.053 (0.098) 0.157 ** (0.062) − − Note: *** p < 0.001, ** p < 0.01,* p < 0.05 (two-tailed test); Standard errors in brackets. LL stands for Log-Likelihood. The upper rows represent the results of the ordered logit model, whereas the bottom rows show the results of the tobit model.

Table A3. Estimation results of the ordered logit and tobit models (part 3).

De1 De2 De3 De4 De5 De6 0.569 *** (0.181) 0.006 (0.161) 0.206 (0.142) 0.177 (0.158) F1 − 0.332 *** (0.069) − 0.024 (0.124) − 0.116 *** (0.037) 0.903 ** (0.383) − 0.292 ** (0.117) − 0.122 * (0.069) − − − − 0.130 (0.234) 0.222 (0.209) 0.619 *** (0.178) 0.210 (0.200) F2 − − 0.228 *** (0.080) 0.450 *** (0.146) 0.067 (0.047) 0.181 (0.384) 0.447 *** (0.133) 0.0187 (0.097) − − 0.269 (0.277) 0.791 *** (0.252) 0.036 (0.209) 0.063 (0.235) F3 0.164 * (0.094) 0.251 (0.165) 0.046 (0.055) 2.417 *** (0.762) 0.145 (0.156) − 0.0485 (0.102) − 0.072 (0.169) 0.246 * (0.149) 0.044 (0.133) 0.152 (0.149) F4 − 0.034 (0.058) − 0.129 (0.110) 0.008 (0.034) 0.763 * (0.394) − 0.015 (0.098) 0.075 (0.063) 0.419 (0.294) 0.085 (0.251) 0.179 (0.217) 0.577 ** (0.249) S1 − 0.006 (0.098) − 0.218 (0.184) − 0.096 * (0.057) 0.456 (0.464) 0.349 ** (0.167) − 0.247 ** (0.107) − − − − 0.334 (0.348) 0.306 (0.304) 0.039 (0.257) 0.713 ** (0.287) S2 − 0.006 (0.116) 0.213 (0.226) − 0.015 (0.068) 0.589 (0.567) 0.074 (0.184) 0.319 *** (0.119) − − 0.051 (0.372) 0.340 (0.335) 0.086 (0.291) 0.138 (0.313) S3 0.119 (0.132) 0.101 (0.254) − 0.036 (0.068) 0.525 (0.720) 0.186 (0.207) − 0.004 (0.138) − − − 0.540 * (0.281) 0.082 (0.247) 0.273 (0.218) 0.209 (0.237) S4 − − 0.137 (0.095) 0.423 ** (0.184) − 0.0‘01 * (0.056) 0.725 (0.454) − 0.134 (0.154) 0.110 (0.104) − − − 0.248 (0.345) 0.168 (0.288) 0.287 (0.247) 0.201 (0.287) S5 − 0.363 *** (0.116) − 0.476 ** (0.214) 0.075 (0.066) 0.606 (0.537) 0.028 (0.190) 0.076 (0.123) Energies 2019, 12, 2759 24 of 27

Table A3. Cont.

De1 De2 De3 De4 De5 De6 0.187 (0.536) 0.709 (0.474) 0.519 (0.416) 0.087 (0.491) S6 − 0.217 (0.183) − 0.146 (0.351) 0.042 (0109) 0.449 (0.982) − 0.242 (0.279) 0.067 (0.202) − − 0.665 (0.724) 1.106 * (0.631) 0.092 (0.607) 0.366 (0.743) S7 0.257 (0.260) 0.469 (0.471) − 0.096 (0.153) 0.088 (0.257) − 0.051 (0.436) − 0.262 (0.283) − − 0.612 (0.914) 0.029 (0.723) 0.597 (0.637) 0.181 (0.669) S8 − 0.253 (0.275) − 0.468 (0.392) 0.120 (0.162) 0.701 (0.862) 0.0519 (0.422) − 0.224 (0.306) − − 1.812 ** (0.714) 0.747 (0.575) 1.071 ** (0.535) 0.619 (0.574) S9 − − 0.243 (0.223) 0.137 (0.393) 0.347 *** (0.129) 0.254 (0.732) − 0.426 (0.400) 0.303 (0.263) − − 0.493 (0.477) 0.212 (0.437) 0.343 (0.380) 0.369 (0.401) S10 0.032 (0.164) − 0.319 (0.288) − 0.109 (0.097) 0.258 (0.725) − 0.404 (0.256) − 0.117 (0.174) − − − 0.045 (0.124) 0.188 (0.249) 0.343 (0.380) 0.369 (0.401) S11 0.019 (0.105) − 0.054 (0.079) − 0.099 (0.137) 0.129 (0.16) 0.024 (0.086) 0.068 (0.077) − − 0.321 (0.362) 0.245 (0.428) 0.343 (0.380) 0.369 (0.401) S12 0.152 (0.164) − 0.125 (0.145) − 0.147 (0.245) 0.05 (0.299) − 0.203 (0.161) − 0.021 (0.137) − − − 0.512 (0.231) 0.239 (0.484) 0.152 (0.361) 0.261 (0.281) S13 − 0.128 (0.226) − 0.272 (0.17) − 0.133 (0.279) 0.281 (0.343) − 0.138 (0.171) − 0.146 (0.162) − − − 277.913 297.134 395.041 420.643 360.383 LL − − − − 205.796 − 396.297 265.206 395.041 230.876 − 367.956 − − − − − Note: *** p < 0.001, ** p < 0.01,* p < 0.05 (two-tailed test); Standard errors in brackets. LL stands for Log-Likelihood. The upper rows represent the results of the ordered logit model, whereas the bottom rows show the results of the tobit model.

References

1. Hinson, S.; Bolton, P.; Barber, S. Energy smart meters. Commons Library Briefing, 3 April 2019; No. 8119. 2. Strbac, G. Demand-side management: Benefits and challenges. Energy Policy 2008, 36, 4419–4426. [CrossRef] 3. Darby, S.; McKenna, E. Social implications of residential demand response in cool temperature climates. Energy Policy 2012, 49, 759–769. [CrossRef] 4. Rogers, E.M. Diffusion of Innovations; The Free Press: New York, NY, USA, 2003. 5. McMichael, M.; Shipworth, D. The value of social networks in the diffusion of energy-efficiency innovations in UK households. Energy Policy 2013, 53, 159–168. [CrossRef] 6. Berger, J.; Iyengar, R. Communication channels and word of mouth: How the medium shapes the message. J. Consum. Res. 2013, 40, 567–579. [CrossRef] 7. Krishnamutri, T.; Schwartz, D.; Davis, A.; Fischof, B.; de Bruin, W.B.; Lave, L.; Wang, J. Preparing for smart grid technologies: A behavioral decision research approach to understanding consumer expectations about smart meters. Energy Policy 2012, 41, 790–797. [CrossRef] 8. Paetz, A.-G.; Duetschke, E.; Fichtner, W. Smart homes as a means to sustainable energy consumption: A study of consumer perceptions. J. Consum. Policy 2012, 35, 23–41. [CrossRef] 9. Ellabban, O.; Abu-Rub, H. Smart grid customers’ acceptance and engagement: An overview. Renew. Sustain. Energy Rev. 2016, 65, 1285–1298. [CrossRef] 10. Nachreiner, M.; Mack, B.; Matthies, E.; Tampe-Mai, K. An analysis of smart metering information systems: A psychological model of self-regulated behavioral change. Energy Res. Soc. Sci. 2015, 9, 85–97. [CrossRef] 11. Gans, W.; Alberini, A.; Longo, A. Smart meter devices and the effect of feedback on residential electricity consumption: Evidence from a natural experiment in Northern Ireland. Energy Econ. 2013, 36, 729–743. [CrossRef] 12. Faruqui, A.; Sergici, S. Household response to dynamic pricing of electricity—A survey of the experimental evidence. J. Regul. Econ. 2010, 38, 193–225. [CrossRef] 13. Burchell, K.; Rettie, R.; Roberts, T. Householder engagement with energy consumption feedback: The role of community action and communications. Energy Policy 2016, 88, 178–186. [CrossRef] 14. Ma, G.; Lin, J.; Li, N. Longitudinal assessment of the behavior-changing effect of app-based eco-feedback in residential buildings. Energy Build. 2018, 159, 486–494. [CrossRef] 15. Podgornik, A.; Sucic, B.; Blazic, B. Effects of customized consumption feedback on energy efficient behavior in low-income households. J. Clean. Prod. 2016, 130, 25–34. [CrossRef] Energies 2019, 12, 2759 25 of 27

16. Schleich, J.; Faure, C.; Klobasa, M. Persistence of the effects of providing feedback alongside smart metering devices on household electricity demand. Energy Policy 2017, 107, 225–233. [CrossRef] 17. Foulds, C.; Robison, R.; Macrorie, R. Energy monitoring as a practice: Investigating use of the i-measure online energy feedback tool. Energy Policy 2017, 104, 194–202. [CrossRef] 18. Gerpott, T.; Paukert, M. Determinants of willingness to pay for smart meters: An empirical analysis of household customers in Germany. Energy Policy 2013, 61, 483–495. [CrossRef] 19. Weron, T.; Kowalska-Pyzalska, A.; Weron, R. The role of educational trainings in the diffusion of smart metering platforms: An agent-based modeling approach. Phys. A Stat. Mech. Appl. 2018, 505, 591. [CrossRef] 20. Good, N.; Ellis, K.; Mancarella, P. Review and classification of barriers and enablers of demand response in the smart grid. Renew. Sustain. Energy Rev. 2017, 16, 57–72. [CrossRef] 21. Kahma, N.; Matschoss, K. The rejection of innovations? Rethinking technology diffusion and the non-use of smart energy services in Finland. Energy Res. Soc. Sci. 2017, 34, 27–36. [CrossRef] 22. Gawlik, L. The Polish power industry in energy transformation process. Miner. Econ. 2017, 31, 229–237. [CrossRef] 23. Rosicki, R. Poland’s energy policy: Main problems and forecasts. Srodkowoeuropejskie Stud. Polit. 2017, 2, 59–87. [CrossRef] 24. Antosiewicz, M.; Nikas, A.; Szpor, A.; Witajewski-Baltvilks, J.; Doukas, H. Pathways for the transition of the Polish power sector and associated risks. Environ. Innov. Soc. Transit. 2019, in press. [CrossRef] 25. Wierzbowski, M.; Filipak, I.; Lyzwa, W. Polish energy policy 2050. An instrument to develop a diversified and sustainable electricity generation mix in coal-based energy system. Renew. Sustain. Energy Rev. 2017, 74, 51–70. [CrossRef] 26. Manowska, A.; Tobor-Osadnik, K.; Wyganowska, M. Economic and social aspects of restructuring Polish coal mining: Focusing on Poland and the EU. Resour. Policy 2017, 52, 192–200. [CrossRef] 27. OptimalEnergy.pl. 2014. Reports on Polish Energy Market. Available online: https://optimalenergy.pl/ raport-o-rynku-energii-w-polsce/ (accessed on 14 May 2019). 28. Institute of Renewable Energy. 2015. Poles Want to Produce Energy from RES-Results of the Latest Research Included in the Report: National Development Plan for Micro Installation of Renewable Energy Sources. Available online: https://ieo.pl/pl/aktualnosci/380-polacy-chc-produkowa-energi-z-oze-wyniki- najnowszych-bada-zawartych-w-raporcie-krajowy-plan-rozwoju-mikroinstalacji-odnawialnych-rode- energii (accessed on 14 May 2019). 29. Nhede, N. Poland Invests in Smart Grid Rollout, Smart Energy International. 2017. Available online: https://www.smart-energy.com/regional-news/europe-uk/energa-energy-distribution-system- operator-poland/ (accessed on 17 May 2019). 30. Kowalska-Pyzalska, A.; Byrka, K. Determinants of the Willingness to Energy Monitoring by Residential Consumers: A Case Study in the City of Wroclaw in Poland. Energies 2019, 12, 907. [CrossRef] 31. RAP. Report on the Polish Power System 2018; Version 2.0 Study commissioned by Agora Energie-Wende; Agora Energie-Wende: Berlin, Germany, 2018. 32. Verbong, G.P.J.; Beemsterboer, S.; Sengers, F. Smart grids or smart users? Involving users in developing a low carbon electricity economy. Energy Policy 2013, 52, 117–125. [CrossRef] 33. Meijer, F.; Straub, A.; Mlecnik, E. Impact of Home Energy Monitoring and Management Systems (HEMS): Triple-A: Stimulating the Adoption of Low-Carbon Technologies by Homeowners through Increased Awareness and Easy Access D2.1.1; Report on Impact of HEMS; Interreg: Lille, France, 2018. 34. Ofgem. Energy Demand Research Project Final Analysis; OFGEM: London, UK, 2011. 35. Star, A.; Isaacson, M.; Haeg, D.; Kotewa, L. The dynamic pricing mousetrap: Why isn’t the world beating down our door? In Proceedings of the ACEEE Summer Study on Energy Efficiency in Buildings 2010, Pacific Grove, CA, USA, 15–20 August 2010; Volume 2, pp. 257–268. 36. UK Department of Energy and Climate Change. Written Evidence Submitted by the Department of Energy and Climate Change (SME0031); UK Department of Energy and Climate Change: London, UK, 2016. 37. British Infrastructure Group. Not So Smart—A Comprehensive Investigation into the Roll-Out of Energy Smart Meters; British Infrastructure Group: London, UK, 2018. 38. Claudy, M.C.; Michelsen, C.; ODriscoll, A.; Mullen, M.R. Consumer awareness in the adoption of microgeneration technologies. An empirical investigation in the Republic of Ireland. Renew. Sustain. Energy Rev. 2010, 14, 2154–2160. [CrossRef] Energies 2019, 12, 2759 26 of 27

39. Frederiks, E.; Stenner, K.; Hobman, E. Household energy use. Applying behavioral economics to understand consumer decision making and behavior. Renew. Sustain. Energy Rev. 2014, 41, 1385–1394. [CrossRef] 40. Abidin, S.Z.; Omar, N.; Rahman, H.F.; Rosli, M.M. Socio-informatics: Identifying influential factors in digital elements. In Proceedings of the 14th WSEAS international conference on Computers: Part of the 14th WSEAS CSCC multiconference-Volume I, Corfu Island, Greece, 23–25 July 2010; World Scientific and Engineering Academy and Society (WSEAS): Cambridge, UK, 2010; pp. 397–402. 41. Hanna, R.; Rohm, A.; Crittenden, V. We’re all connected: The power of the social media ecosystem. Bus. Horizons 2011, 54, 265–273. [CrossRef] 42. Ma, L.; Lee, C.S.; Goh, D.H. Understanding news sharing in social media: An explanation from the diffusion of innovations theory. Online Inf. Rev. 2014, 38, 598–615. [CrossRef] 43. Chawla, Y.; Chodak, G. Recommendations for Social Media Activities to Positively Influence the Economic Factors. In Double-Blind Peer-Review Proceedings Part I. of the International Scientific Conference Hradec Economic Days 2018, 30–31 January 2018; Jedliˇcka,P., Marešová, P., Soukal, I., Eds.; University of Hradec Králové: Hradec Králové, Czech Republic, 2018; Volume 8, pp. 328–338, ISSN 2464-6059. 44. Reid, C.K. The state of digital advertising. EContent 2014, 37, 16–17. 45. Yang, B.; Liu, Y.; Liang, Y.; Tang, M. Exploiting user experience from online customer reviews for product design. Int. J. Inf. Manag. 2019, 46, 173–186. [CrossRef] 46. Heinonen, S.; Hiltunen, E. Creative Foresight Space and the Futures Window: Using visual weak signals to enhance anticipation and innovation. Futures 2012, 44, 248–256. [CrossRef] 47. Daelemans, W. Keynote: Profiling the Personality of Social Media Users; In of the Final Workshop 7 December 2016; ELRA: Naples, Italy, 2016; p. 18. 48. Booth, N.; Matic, J. Mapping and leveraging influencers in social media to shape corporate brand perceptions. Corp. Commun. Int. J. 2011, 16, 184–191. [CrossRef] 49. Amblee, N.; Bui, T. Harnessing the Influence of Social Proof in Online Shopping: The Effect of Electronic Word of Mouth on Sales of Digital Microproducts. Int. J. Electron. Commer. 2011, 16, 91–114. [CrossRef] 50. Moraitis, O.T.P.; Kausika, B.B.; Van Der Velde, H.; Hart’T, S.; De Vries, A.; De Rijk, P.; De Jong, M.M.; Van Leeuwen, H.; Van Sark, W. Three years experience in a Dutch public awareness campaign on photovoltaic system performance. IET Renew. Power Gener. 2017, 11, 1229–1233. 51. Wu, Y.; Xie, L.; Huang, S.; Li, P.; Yuan, Z.; Liu, W. Using social media to strengthen public awareness of wildlife conservation. Ocean. Coast. Manag. 2018, 153, 76–83. [CrossRef] 52. Dufty, N. The use of social media in countrywide disaster risk reduction public awareness strategies. Aust. J. Emerg. Manag. 2015, 30, 12–16. 53. Rubin, G.D.; Krishnaraj, A.; Mahesh, M.; Rajendran, R.R.; Fishman, E.K. Enhancing Public Access to Relevant and Valued Medical Information: Fresh Directions for RadiologyInfo.org. J. Am. Coll. Radiol. 2017, 14, 697–702. [CrossRef] 54. Ghazali, Z.; Zahid, M.; Kee, T.S.; Yussoff Ibrahim, M. A step towards sustainable society: The awareness of carbon dioxide emissions, climate change and carbon capture in Malaysia. Int. J. Econ. Financ. Issues 2016, 6, 179–187. 55. Poulopoulos, V.; Vassilakis, C.; Antoniou, A.; Lepouras, G.; Wallace, M. Personality Analysis of Social Media Influencers as a Tool for Cultural Institutions; Springer: Cham, Switzerland, 2018. 56. Meena, L.; Alamelu, R.; Amudha, R.; Nalini, R.; Motha, L.C.S. Public awareness towards India Innovation Lab for Green Finance. Int. J. Econ. Res. 2017, 14, 391–400. 57. Bertr, ; P.J.; Niles, S.L.; Newman, D.J. Human Spaceflight in Social Media: Promoting Space Exploration Through Twitter. New Space 2015, 3, 117–133. [CrossRef] 58. Mohammadi, N.; Wang, Q.; Taylor, J.E. Diffusion dynamics of energy saving practices in large heterogeneous online networks. PLoS ONE 2016, 11, e0164476. [CrossRef] 59. Gaál, Z.; Szabó, L.; Obermayer-Kovács, N.; Csepregi, A. Exploring the role of social media in knowledge sharing. Electron. J. Knowl. Manag. 2015, 13, 185–197. 60. Rada, M. Industry 5.0 Definition, Medium.com, 21 July 2018. Available online: https://medium.com/ @michael.rada/industry-5-0-definition-6a2f9922dc48 (accessed on 30 April 2019) 61. Gouws, T.; Van Rheede van Oudtshoorn, G.P. Correlation between brand longevity and the diffusion of innovations theory. J. Public Aff. 2011, 11, 236–242. [CrossRef] Energies 2019, 12, 2759 27 of 27

62. Wuestenhagen, R.; Wolsink, M.; Buerer, M.J. Social acceptance of renewable energy innovation: An introduction to the concept. Energy Policy 2007, 35, 2683–2691. [CrossRef] 63. Wolsink, M. Distributed generation of sustainable energy as a common pool resource: Social acceptance in rural setting of smart (micro-) grid configurations. In New Rural Spaces: Towards Renewable Energies, Multifunctional Farming, and Sustainable Tourism; Frantál, B., Martinát, S., Eds.; University of Amsterdam: Amsterdam, The Netherlands, 2014; pp. 36–47. 64. Buchanan, K.; Banks, N.; Preston, I.; Russo, R. The British public’s perception of the UK smart metering initiative: Threats and opportunities. Energy Policy 2016, 91, 87–97. [CrossRef] 65. van der Werff, E.; Steg, L. The psychology of participation and interest in smart energy systems: Comparing the value-belief-norm theory and the value-identity-personal norm mode. Energy Res. Soc. Sci. 2016, 22, 107–114. [CrossRef] 66. Wallenborn, G.; Orsini, M.; Vanhaverbeke, J. Household appropriation of electricity monitors. Int. J. Consum. Stud. 2011, 35, 146–152. [CrossRef] 67. Hargreeaves, T.; Nye, M.; Burgess, J. Keeping energy visible? Exploring how households interact with feedback from smart energy monitors in the longer term. Energy Policy 2013, 52, 126–134. [CrossRef] 68. Kowalska-Pyzalska, A. What makes consumers adopt to innovative energy services in the energy market? A review of incentives and barriers. Renew. Sustain. Energy Rev. 2018, 82, 3570–3581. [CrossRef] 69. Allcott, H. Social norms and energy conservation. J. Public Econ. 2011, 95, 1082–1095. [CrossRef] 70. Nolan, J.; Schultz, P.; Cialdini, R.; Goldstein, N.; Griskevicius, V. Normative social influence is underdetected. Personal. Soc. Psychol. Bull. 2008, 34, 913–923. [CrossRef] 71. Khan, M.Z.; Miankhel, A.K.; Nawaz, A. Information & Communication Technology and ‘Individual’: Prospects & Concerns. Glob. J. Comput. Sci. Technol. 2013, 13. Available online: https://computerresearch. org/index.php/computer/article/view/387 (accessed on 18 July 2019). 72. Kowalska-Pyzalska, A.; Maciejowska, K.; Suszczyński,K.; Sznajd-Weron, K.; Weron, R. Turning green: Agent-based modeling of the adoption of dynamic electricity tariffs. Energy Policy 2014, 71, 164–174. [CrossRef] 73. RWE Report. What Poles Think about the Energy Market? The Report for RWE Polska S.A. 2014. Available online: https://www.innogy.pl/en/about/about-group (accessed 12 May 2019). (In Polish) 74. Podsakoff, P.M.; MacKenzie, S.B.; Lee, J.-Y.; Podsakoff, N.P. Common method biases in behavioral research: A critical review of the literature and recommended remedies. J. Appl. Psychol. 2003, 88, 879–903. [CrossRef] 75. Podsakoff, P.M.; MacKenzie, S.B.; Podsakoff, N.P. Sources of method bias in social science research and recommendations on how to control it. Annu. Rev. Psychol. 2012, 63, 539–569. [CrossRef]

c 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access

article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). Paper 4

Y. Chawla, A. Kowalska-Pyzalska, W. Widayat Consumer Willingness and Acceptance of Smart Meters in Indonesia resources

Article Consumer Willingness and Acceptance of Smart Meters in Indonesia

Yash Chawla 1,*,† , Anna Kowalska-Pyzalska 1,† and Widayat Widayat 2 1 Department of Operations Research, Faculty of Computer Science and Management, Wroclaw University of Science and Technology, 50-370 Wrocław, Poland; [email protected] 2 Department of Management, Faculty of Economics and Business, University of Muhammadiyah Malang, Malang 65145, Indonesia; [email protected] * Correspondence: [email protected]; Tel.: +48-693-290-935 † Current address: Wyb. Wyspianskiego 27, 50-370 Wroclaw, Poland.

Received: 26 October 2019; Accepted: 22 November 2019; Published: 24 November 2019

Abstract: Indonesia is the fourth most populous country in the world and is currently facing some challenges, such as pollution and a growing energy demand. One of the solutions to these problems is upgrading the electricity transmission and distribution system to avoid losses of energy, and encourage consumer engagement in energy saving as well as energy generation. The government of Indonesia has initiated projects for smart grids and advanced metering infrastructure (AMI), but consumer awareness and willingness to accept these new technologies is still uncertain. This study focused on analyzing consumers’ knowledge and willingness to accept one of the key components in grid modernization, being smart meters (SM). An online questionnaire was used to record responses from 518 social media users from different parts of Indonesia. The analysis shows that, among social media users who are seen as early adopters of technology, there is certainly a lack of awareness about SM, but they are largely open towards the acceptance of SM. Based on the ﬁndings, we have also drawn recommendations for energy companies, which would help in raising consumer awareness, as well as acceptance of SM in Indonesia.

Keywords: smart meters; sustainable development; Indonesian energy market; consumers’ preferences; on-line questionnaire; social media users

1. Introduction and Literature Review

1.1. Sustainable Development of the Energy Markets and Power Systems The 2030 Agenda for Sustainable Development was adopted by all United Nations Member States in 2015. This document provides 17 Sustainable Development Goals (SDGs), which should be implemented by all developed and developing countries, creating a global partnership (for more details, see https://sustainabledevelopment.un.org/sdgs (accessed on 29 September 2019)). Among the proposed SDGs, SDG 7 is ensure access to affordable, reliable, sustainable and modern energy for all, refers directly to the energy sector; SDG 9 is build resilient infrastructure, promote inclusive and sustainable industrialization and foster innovation; and SDG 12 is ensure sustainable consumption and production patterns, which are directly and indirectly related to the energy industry. Energy is also very important in other strategic documents, such as the 2030 Agenda for Sustainable Development and the Paris Agreement on Climate Change. Both documents emphasize that “ensuring access to affordable, reliable, sustainable and modern energy for all will open a new world of opportunities for billions of people, through new economic opportunities and jobs, empowered women, children and youth, better education and health, more sustainable, equitable and inclusive communities, and greater protections from, and resilience to, climate

Resources 2019, 8, 177; doi:10.3390/resources8040177 www.mdpi.com/journal/resources Resources 2019, 8, 177 2 of 23 change” (https://sustainabledevelopment.un.org/topics/energy (accessed on 19 September 2019)). Furthermore, to achieve these ambitious goals, many of the partnering countries all over the world have proposed their own regulations and strategies, leading to increases in energy efficiency, a decrease of energy wastage and digitization of the power system [1–5]. The aim of those regulations is to provide a major transition of the existing power systems, traditionally based on the large power plants (nuclear or coal based), with a passive role of the consumers, towards sustainable and smart power grids, where communication among the market participants, i.e. energy generators, suppliers, sellers, and consumers, plays a major role. Within the smart grid (SG) concept, an implementation of advanced metering system (AMI) is required. This advanced IT and communication technology offers great opportunities to all market participants, leading to cost optimization and savings. AMI consists of an integrated system of smart meters (SM), communication networks, and data management systems that enables two-ways communication between utilities and customers. The level and scope of transition of power system into smart grid, and the progress in introduction of AMI and SM is different in various countries or regions of the world. It is mainly due to the legislative and economic reasons. For example, in European Europe, based on the Directive 2012/27/EC on energy efficiency, the decision whether to exchange the traditional meters into smart ones is based on the long-term cost–benefit analysis made by the individual member states. In result, in countries such as Denmark, Sweden, or Spain, the introduction of AMI and the exchange of traditional meters into smart meters have been already finalized [3,5,6], whereas, in Ireland, Greece or the Czech Republic, it has not even started [7]. In many other countries all over the world, such as India, Indonesia, Turkey, or Brazil, this process has just begun or is going to begin soon [8–10]. There are many initiatives, such as the European Electricity Grid Initiative (EEGI) and European Strategic Energy Technology Plan (SETplan), Turkey Smart Grid 2023 Vision and Strategy Roadmap (TSG’2023), National Smart Grid Mission in India, or Making Indonesia 4.0, that encourage the sustainable transition of the power systems in terms of its economic, ecological, technical, and social aspects in the coming decades [8,11,12]. Each of those documents provides some milestones, hints how to achieve them, and indicators to verify the progress being made.

1.2. Indonesian Energy Sector Within this paper, our attention is focused on Indonesia—the fourth most populated country in the world with 250 million of citizens—and its significant role as a major producer and consumer of energy in regional and international markets. The country is also the largest economy in the Association of South-East Asian Nations (ASEAN) and an active member of the G20 Summit. Indonesia is one of the world’s largest producers of natural gas and coal [13]; however, the renewable energy potential in Indonesia is also very high. Apart from solar, wind, and hydro energy, geothermal energy can also be harvested (for more details, see: https://www.nortonrosefulbright.com/en/knowledge/publications/ 0552a1f0/renewable-energy-snapshot-indonesia (accessed on 14 October 2019)). The country experiences a rapid increase in electricity demand (up to 10–15% per year), which leads to the danger of blackouts, due to the insufficient energy supply and technical problems of low quality infrastructure in the Indonesian power system [14–16]. Indonesia belongs to those developing countries which are experiencing rapid economic growth combined with rising urbanization, and this leads to enormous pressures on the environment [13]. Nowadays, it is one of the most polluted countries in the world. Extremely poor air quality in Jakarta, the capital city of Indonesia, places Jakarta at the top of the list of most polluted places to live. According to the Energy Policy Institute at the University of Chicago, an increase in coal-fired power stations, burning of land for agriculture plantations, and a rising level of vehicle exhaustion fumes are responsible for the worsening pollution in Indonesia. If the air quality does not improve significantly, the life expectancy of Indonesians will drop drastically in the coming years. Resources 2019, 8, 177 3 of 23

The Ministry of Energy and the Ministry of Economic Affairs of the Indonesian government are aware of the current problems. One of the potential cures is the increase of energy efficiency in production and consumption of energy. Since 2014, the state-owned electricity company Perusahaan Listrik Negara (PLN), has been in the process of switching from the still dominating manual electricity meters to digital ones, using AMI technology. It is expected that the new system can be used by one million customers already in 2019. The installation of digital meters takes place, firstly, in big cities and tourist destinations, such as Jakarta, Bali, and Labuan Bajo in East Nusa Tenggara. The installation is free of charge, as the customers do not have to pay directly for the switch from manual to digital. The costs are borne by the PLN company. Due to the exchange of smart meters, energy wastage in generation, transmission, and consumption is expected to decrease. Indonesia has received external support from organizations such as the Asian Development Bank (ABD) to overcome the challenges it is facing in the energy market. Since 2015, ABD has financed projects for the modernization and strengthening of electricity grids in Indonesia, such as the Sumatra Program (result-based $600 million project) (for more details, see: https://www.adb.org/ projects/49080-001/main#project-overview and https://www.adb.org/sites/default/files/linked- documents/50016-001-pid.pdf). In 2017, ABD also approved a loan of $1.1 billion for strengthening and diversifying Indonesia’s energy sector, which is considered a key element to promote inclusive growth and sustainable development in the country. One of the key disbursement-linked indicators for these projects with ADB is the growth in the number of smart meters through the replacement of the current manual electricity meters. It is expected that 75% of consumers in the regions covered by the project would have SM installed by 2021. The base line taken at the time of disbursement of the load was 48% in 2016. According to the current status of the two smart grid projects in Selayar and Sangihe, 57.78% of consumers in those regions have SM already installed in their homes (for more details, see: https://www.adb.org/projects/50016-001/main#project-pds). Engaging the consumers, making them more aware about the features and uses of SM, and managing their electricity expenditures were also among the key program activities of these projects.

1.3. Indonesia 4.0 and Smart Meters Not only in Indonesia, but all over the world, the implementation of smart grids and increase in the usage of renewable energy sources seems to be a promising cure to the current challenges of the power systems [4,17]. The role of advanced metering infrastructure (AMI) development, in general, and smart metering, in particular, may have a tremendous effect on the realization of some of the Sustainable Development Goals (SDGs) in the area of sustainable production and consumption in the energy sector through by the implementation of advanced technology that allows better control over one’s energy consumption (for more details, see: https://sustainabledevelopment.un.org/topics/energy (accessed on 20 October 2019)). The realization of SDGs in the area of digitization of the industry, as emphasized by Hidayatno et al. [16], may also lead to a sustainable energy transition. The authors indicated that the fourth industrial revolution (Industry 4.0) is now believed to have significant relations with sustainable energy. The United Nations Industrial Development Organization (UNIDO), in its 2017 report, elaborated on the relevance between SDGs about sustainable energy and inclusive industry development. In that sense, Industry 4.0 and the sustainable energy transition share important concerns that can be interconnected to pursue a sustainable energy transition in the coming future [16]. Sustainable energy itself is defined to have two central components: renewable energy and energy efficiency. The latter is combined with the implementation of smart grids, together with the AMI system and including SM as its core component. The Indonesian government and the Ministry of Industry have released the road-map called Making Indonesia 4.0, in which they set some goals and priorities to be achieved by 2030 [18]. According to this document, Indonesia wants to become one of the Top 10 global economies by building strong manufacturing infrastructure by 2030 (see [16,18] for more details). Among the priorities mentioned in this document, National Priority Number 3 declares Resources 2019, 8, 177 4 of 23 that Indonesia will accommodate the standards in sustainability, including that of energy consumption in industrial sectors [16]. Although the industrial sector is, without any doubt, very important, great attention must be also be paid to the residential consumers. Due to the significant changes and decentralization of energy production in most countries, this group has become an important player in the energy market [6,19–22]. Consumers’ activities in energy generation (i.e., by means of small-scale generators) and conscious control and reduction of energy consumption may bring positive effects to the whole power system. The question is, however, how to engage consumers and make them aware of the opportunities they have. Energy companies see this as one of the major challenges. In a 2017 report published by PwC, over 70% of energy companies emphasized that the management of expansion programs, including raising customer engagement and awareness, was a major impediment at that time and would still be significant even five years from then [23]. Major efforts, in this regard, are required to be undertaken, as the up-to-date literature reveals that consumers do not pay much attention to the energy market, unless they are motivated by financial, ecological, normative, or other incentives [24–27].

1.4. Current Findings from the Literature Review The up-to-date literature reveals that, even if the idea of an energy transition into smart grids looks optimistic and brings many advantages, consumers’ engagement, interest, and acceptance are needed to achieve the ambitious goals regarding the increase of energy efficiency and reduction of environmental pollution and climate change [20,24,26,28,29]. Smart meters themselves are not very user friendly and offer much more benefits to the energy providers than the consumers. However, if SM is combined with some advanced enabling technologies, such as smart metering platforms (i.e., mobile apps or Internet widgets), in-home displays, smart plugs, or smart appliances, they automatically offer a spectrum of opportunities to the end-users of electricity. First, by offering feedback regarding energy consumption in real-time, they allow consumers to better understand their energy consumption and help them to increase their energy efficiency by lowering energy wastage [9,30–32]. The literature emphasizes that feedback received from SM about one’s energy consumption may create some behavioral change and lead to monitoring the amount of energy consumed and to the reduction in this consumption [28,30,31,33]. Based on the literature review, the following problems with SM deployment among consumers should be emphasized: (1) low-level of consumer awareness and knowledge [31]; (2) low level of consumer acceptance and, hence, engagement [20,34]; (3) concern regarding data privacy and security [27]; (4) lack of experience or technical feasibility regarding eco-feedback received through enabling technologies, such as smart metering platforms or in-home displays, and a lack of usefulness of SM from the consumers’ point of view [28,30,35–37]; and (5) improper or insufficient communication channels between the energy providers and their customers [17,32]. The current findings about the Indonesian SM deployment indicates that consumers’ expectations regarding the usefulness, ease of use, and risk of smart meters influence their attitudes and behaviors regarding smart meter adoption [14]. Moreover, as the authors of [9,14] revealed, perceived usefulness and ease of use from one side, and perceived risk from the other side, are the most important determinants of consumer acceptance of SM. Within our study, we wanted to explore the points of view regarding SM of Indonesians even further, focusing on social media users, who are supposed to act as early adopters in further SM deployment in the energy market.

1.5. Research Questions and Aim of the Study Taking all of that into consideration, within our study, we wanted to focus on Indonesian social media users and verify, first, their knowledge about SM and, second, their preferences and concerns, as well as their willingness to install SM under various conditions. We also aimed to explore the consumers’ communication channels in the energy market. We believe that the analysis of consumers’ Resources 2019, 8, 177 5 of 23 preferences and communication channels may lead to constructive conclusions regarding further SM deployment. Although SM acceptance and deployment have already been studied in various cultural and socioeconomic contexts, very few have referred or focused on the communication channels between end-users of electricity and energy companies (see, e.g., [17,38]), especially among social media users. As our study was based on the Roger’s model of innovation diffusion (DoI) [39], it paid great attention to the channels through which innovation, smart metering in this case, is spread. In the DoI model, not only do communication channels matter, but also time, the social network, and innovation attributes. We decided to explore the consumers’ acceptance of SM only among social media users, because, based on the literature, this group of consumers is predominated in usage of modern, Internet based technologies [40–42]. Moreover, we chose to conduct this study in Indonesia, where plenty of papers regarding the Indonesian renewable energy sector are available [15,43–45], but there are very few dealing with the deployment of smart meters in this region of the world [9,14]. Our survey contributes to the literature by further enriching the literature with the most recent consumer awareness levels and preferences regarding SM. In particular, our paper contributes to emphasizing that communication channels, both traditional and modern ones, cannot be neglected in the bilateral contacts between energy providers and consumers. Nowadays, consumers need to be informed, not only about the new opportunities they have in the energy market, but also must feel empowered to use them in order to make their energy consumption more efficient and sustainable. The structure of the paper is as follows. After discussing in Section1 the aspects of sustainable development in the context of energy transition into smart grids and presenting some current findings from the literature regarding SM adoption among consumers, the introduction of the Indonesian energy sector is provided. Then, in Section2, the research and survey frameworks are presented, and the definitions of the variables are elaborated. In Section3, the data collection and the sample are described. Section4 presents and discusses the results of the analysis. Finally, Section5 summarizes the findings.

2. Methods and Framework

2.1. Research Framework The literature analysis and the specifics of Indonesian challenges in their energy market and SM deployment allowed us to define the research objective. We aimed to investigate the consumers’ knowledge about SM and willingness to accept SM under various conditions. We also wanted to explore the consumers’ communication channels in the energy market. The findings from this study allowed us to provide some recommendations that can be used by energy companies to enhance the awareness and acceptance of SM in Indonesia. Our research framework is shown in Figure1. Once the objectives were decided, we carried out another extensive literature review. Several studies, reports, and new articles were studied about challenges in the Indonesian energy markets, pollution problems, consumer awareness, acceptance of technology, and AMI roll-out plans. Based on the findings from the literature and taking into account the previous studies, the framework of the survey questionnaire was formulated. The questionnaire was then disseminated through social media channels, as the opted target audience were the social media users. Once the data were collected, we conducted a descriptive analysis to gain a better understanding of our respondents. We especially paid attention to the communication channels they use (in particular, social media channels), channels through which they received information about electricity, information about SM (only for those who knew what SM was), and where they would look for information regarding SM. This was done through analyzing the independent responses to all variables, hence presenting a preliminary understanding of the consumers point-of-view. Thereafter, by using the Tobit regression models, we analyzed the willingness of the respondents to adopt SM under Resources 2019, 8, 177 6 of 23 various conditions, followed by elaborating on the conclusions and recommendations to enhance further acceptance of SM.

Figure 1. Research framework and ﬂow of the study.

From the methodological point of view, we based our study on the famous model of innovation diffusion (DoI) by Rogers [39]. DoI model pays great attention to the communication channels in spreading news about market novelties [32]. Apart from communication channels, the model takes into account: attributes of innovation, time, and a social network. Within the model, great emphasis is placed on the acceptance of innovation in society. Without such an acceptance, reaching of high adoption rates (i.e., market penetration) is not possible. That is why, in our research framework, consumers’ awareness and acceptance are perceived as a key element of the diffusion of SM in the market.

2.2. Survey Framework The proposed questionnaire was designed in three stages and consisted of several variables, as described in Tables1 and2 respectively. The variables included in the survey were motivated by the Resources 2019, 8, 177 7 of 23 literature and similar studies regarding consumers’ acceptance and preferences towards SM and other enabling technologies (see, e.g., [25,29,31,32,46,47]). The literature states that social and economic attributes are important for SM diffusion, but, additionally, the consumers’ knowledge, awareness, and preferences, as well as social inﬂuence, may play a role in SM diffusion. The questionnaire was divided into three stages, as shown in Table1.

Table 1. Survey Framework with different variables at each stage

The Survey Framework D1–D12 A1–A2, A31–A39 B1–B7 S01–S09 Stage 1 S1–S15 R1 P1–P4 F1,F4

De1–De4, De6–De8 K1: Do you know what is an “Electricity Smart Meter”? Yes No/Not Sure I1, I2, I31–I35 Stage 2 K2–K4 G1–G3

De5 X1 Q1: Would you search for more information about Electricity Smart Meters ? Stage 3 Yes No/Not Sure Q21–Q25

In the first stage, all respondents (N = 518) were asked about their demographic attributes, such as gender, age, educational qualification, household income, area of residence, and so on (D1–D12). Thereafter, they were asked about their preferred social media platforms (S01–S09), their possessions or belongings (B1–B7), the sources of information regarding electricity (S1–S15), potential use of renewable sources in their household (R1), and if they regularly monitored their electricity usage (R1). To check the respondents’ attitude and habits towards the conservation of the environment and energy, they were asked questions related to Variables A1, A2 and A31–A39. Further, the respondents were asked about their preferences (P1–P4), concerns (F1 and F4) and willingness to install SM under various conditions (D 1 D 4 and D 6 D 8). e − e e − e In the second stage, the respondents were asked if they knew what an SM was (K1). All respondents who said “No” or “Not Sure” to K1, were not asked any questions from the second stage and were directed to the third stage. Respondents who knew about SM (N1 = 129) were asked about the sources they received information about SM from (I1, I2, and I31–I35), preferences regarding the government’s role in SM roll out (G1–G3), additional concerns about SM (F2 and F3), whether they have, want to, plan to, or are in the process of installing SM in their home (K2–K4 and X1) and a verification question regarding their willingness to pay for SM (De5). In addition to the questions asked in this first stage and related to social influence, an additional question related to it (W1) was asked in this stage. These respondents (N1 = 129) were then directed to the third stage. Resources 2019, 8, 177 8 of 23

In the third stage, all respondents in the study (N = 518) were asked if they would want to know more details or collect more information on SM (Q1). The respondents who opted “No” or “Not sure” directly moved to submit their responses to the database. On the other hand, the respondents who showed a willingness (Q1 = 1) to know more about SM (N2 = 319) were asked about the communication channels where they would look for information regarding SM (Q21–Q25).

Table 2. Deﬁnitions of the variables and coding (N = 518).

Variable Code Description Gender D1 nominal variable Age D2 ordinal variable Relationship status D3 nominal variable Highest Educational Qualiﬁcation D4 ordinal variable Occupation/Employment D5 nominal variable Collective Household Income (in Indonesian D6 interval variable Rupiah per month) Range of electricity bill (in Indonesian Rupiah per D7 interval variable month) Total members in the household D8 ordinal variable Number of children D81 ordinal variable Type of house D9 nominal variable Place of living D10 ordinal variable (1) yes/(2) no, but I plan to buy it within one Belongings (of smart devices and personal assets) B1–B7 year/(3) no, and I do not plan to buy it (1) I buy thrice or more per year/(2) I buy twice per year/(3) I buy once in a year/(4) I Behavior towards buying new technology A1–A2 buy once every two years/(5) I buy once in more than three years Behavior towards environment and energy saving A31–A39 (1) yes/ (2) no Renewable energy sources installed at the R1 (1) yes/(2) no/(3) uncertain household Social media platforms commonly used S01–S09 nominal variable Source of information regarding electricity (prices, S1–S15 nominal variable new offers, etc.) Knowledge about SM K1–K4 (1) yes/(2) no/(3) uncertain I1, I2, Source of information regarding SM (1) yes/(2) no/(3) uncertain I31–I45 Social inﬂuence W1 (1) yes/(2) no/(3) uncertain Preferences regarding the role of the government G1–G3 (1) yes/(2) no/(3) hard to say in SM enrollment Preferences regarding SM platforms P1–P4 (1) yes/(2) no/(3) uncertain Concerns about SM usage F1–F4 (1) yes/(2) no/(3) uncertain Willingness to have one’s home to be equipped X1 (1) yes/(2) no/(3) uncertain with SM

Decisions to install SM De1–De8 (1) yes/(2) no/(3) hard to say Willingness to search or collect more information Q1 (1) yes/(2) no/(3) uncertain regarding SM Source of Information preferred to search or collect Q21–Q25 (1) yes/(2) no more information regarding SM

2.3. Data Collection and Sampling A self-administered online anonymous questionnaire, hosted on a web page, was used to gather the primary data for carrying out this empirical quantitative study. The target pool of respondents Resources 2019, 8, 177 9 of 23 included individuals residing in Indonesia (irrespective of nationality), who were responsible or co-responsible for making decisions in regards to managing the household. The survey was published at the beginning of August 2019 and data collection was completed with N = 518 responses a month later. We used two sampling techniques, one after the other, to recruit respondents for this study. At ﬁrst, a non-probability sampling method, convenience sampling, was used. In this phase, direct social connections via social media posts and personal messages or emails were used. In the second phase, snowball sampling was implemented, by reaching a wider range of respondents through those who participated in the ﬁrst phase of the survey. The questionnaire was made available in two languages, English and Indonesian (Bahasa). The questions, as well as the answer options, were exactly the same and had the same sequence as well. The user sessions, for the landing page from where users selected their language of choice and the webpages with the questionnaire, was tracked using Google Analytics (GA), to track the response rate. This did not effect the anonymity of the respondents, as the IP addresses of the respondents were not available to us through GA. In total, 1354 user sessions were recorded on the landing page, out of which 559 user sessions dropped out and did not continue to the questionnaire. Out of the 795 user sessions that went through to the questionnaires, 28 and 767 sessions were recorded on the web pages with English language and Indonesian language questionnaires, respectively. N = 518 valid responses were collected, 15 in English and 503 in Indonesian. While planning the survey framework, we deliberated over the number of responses that could be satisfactory for our study. From the literature, we observed that previous studies conducted through online questionnaires among social media users, had collected 300–500 responses. Hence, we aimed to get at-least 500 responses. Once the data were collected, they were translated to the English language for analysis.

3. Description of the Data Through the questionnaire, we collected information regarding a number of variables, all of which are deﬁned and coded in Table2. The elaboration in regards to the variables, based on the collected responses are described in Sections 3.1–3.8.

3.1. Demographics of the Respondents (D1–D10, D81) Table3 shows the statistical description of the variables, which explained the socioeconomic factors of the respondents. The majority of the sample was young, between the ages of 18 and 35 years, with a close balance between genders. Almost half of the respondents were married and the other half were single, with a small number (3.7%) of respondents being widowed or widowers. Respondents were well educated, with most possessing qualifications higher than a high school graduation or equivalent. A large part of the respondents were employed in the public/private sector or had their own business, while about a quarter of them were still students. The majority of the respondents had three or more members in their household, but 46.7% had no children. The most popular type of residence option was a house with no floors, followed by apartments or flats in a high rise building with more than four floors. According to the annual report by “Hootsuite” and “We are Social” in 2019 (for more details, see: https://datareportal.com/reports/digital-2019-indonesia (accessed on 21 October 2019)), social media users comprise around 56% of Indonesia’s population, with an annual recorded growth rate of 15% in 2018. The demographic statistics of the collected sample were compared with the data for the whole country (for more details: https://www.cia.gov/library/publications/the-world-factbook/geos/ id.html (accessed on 21 October 2019)), which yielded that the weight-age of various demographic variables in this study and their counterpart in the actual (for Indonesia) are largely similar. Due to this resemblance, this sample could be considered as representative for the country. Resources 2019, 8, 177 10 of 23

Table 3. Frequencies of the demographic variables (D1–D10 and D81).

Variable Frequencies Male (46.3%) (D1) Gender Female (53.7%) 18–25 years old (41.9%) 26–35 years old (22.4%) 36–45 years old (10.2%) (D2) Age 46–55 years old (19.3%) 56–65 years old (5.8%) 66+ years old (0.4%) Single (46.5%) (D3) Relationship status Married (49.8%) Widowed or Widower (3.7%) Elementary School (2.7%) Junior High School or Equivalent (4.8%) High school graduation or (D4) Highest Educational Qualiﬁcation equivalent (18.9%) Completed Diploma (3.7%) Graduate (44.6%) Master’s Degree (19.5%) PhD Complete (5.8%) Job in Private Sector (37.6%) Job in Public Sector (10.6%) Business (14.1%) (D5) Occupation/Employment Student (25.3%) Unemployed (10.4%) Retired (1.9%) Under Rp. 1 million (14.5%) Rp. 1 million to 1.5 million (14.3%) Rp. 1.5 to Rp. 2 million (10.8%) (D6) Collective Household Income Rp. 2 million to Rp. 3 million (12.5%) (in Indonesian Rupiah per month) Rp. 3 million to Rp. 5 million (16.2%) Rp. 5 million to Rp. 7.5 million (14.9%) More than Rp. 7.5 million (16.8%) Not more than Rp. 50,000 (6.4%) Rp. 50,000 to Rp. 100,000 (16.8%) Rp. 100,000 to Rp. 150,000 (19.7%) (D7) Range of electricity bill Rp. 200,001 to Rp. 250,000 (20.1%) Rp. 250,001 to Rp. 300,000 (8.5%) Rp. 300,001 to Rp. 350,000 (5.2%) (in Indonesian Rupiah per Month) Rp. 350,001 to Rp. 400,000 (4.4%) Rp. 400,001 to Rp. 450,000 (2.7%) Rp. 450,001 to Rp. 500,000 (5.2%) More than Rp. 500,000 (11.0%) One (8.3%) Two (10.4%) Three (21.4%) (D8) Total members in the household Four (33.6%) Five (18.1%) Six or more (8.1%) None (46.7%) One (17.6%) (D81) Number of children Two (22.2%) Three (10.6%) Four or More (2.9%) Resources 2019, 8, 177 11 of 23

Table 3. Cont.

Variable Frequencies Apartment/Flat (in a building up to 4 floors) (0%) Apartment/Flat (in a building with 4+ floors) (31.9%) (D9) Type of house House (only ground floor) (67.2%) House (multiple floors) (1.0%) City with population less than 50,000) (12.7%) City with population between 50,000 and 100,000 (28.0%) (D10) Place of living City with population between 100,000 and 500,000 (16.6%) City with population more than 500,000 (16.6%) Village (26.1%)

3.2. Belongings and Possessions of the Respondents (B1–B7) Figure2 shows the distribution of respondents’ various belongings and possessions. Among the most popular possessions were a house, laptop, WiFi/internet connection at home, and appliances that could connect to the Internet. The popularity of these possessions would still be higher in a year’s time, as a substantial number of respondents expressed a plan to buy these assets within a year. Owning a ﬂat or apartment and electric vehicle were among the least popular, even though their numbers were also estimated to increase in a year. About 25% of respondents owned smart technologies that enabled the monitoring and control of energy consumption in the household, with around 15% of respondents having a plan to buy smart technologies within a year.

Figure 2. Distribution of respondent’s belongings of smart devices and personal assets, where: B1, a house; B2, ﬂat or apartment; B3, a laptop; B4, a WiFi /internet connection at home; B5, home appliances that can connect to the internet; B6, electric vehicle; B7, smart technologies that enable monitoring or control of energy consumption at home

The distribution of consumers’ belongings of smart devices have been introduced in order to verify if people are already experienced with smart devices connected with Internet and enabling a real-time access via a mobile apps are also more open-minded towards SM installation [32]. Secondly, people having their own houses or apartments are usually supposed to be more eager to monitor their energy consumption, which could be done by means of the enabling technologies, such as smart metering platforms combined with SM [31]. The relation between belongings and respondents’ willingness to accept/install SM is veriﬁed in Section 4.1. Resources 2019, 8, 177 12 of 23

3.3. Communication Channels and Sources of Information (S01–S08, S1–S15, I1, I2, I31–I35, Q21–Q25) Respondents were asked about various communication channels, in general, and source of information regarding electricity and SM, in particular. The recorded responses are shown in Table4. Among the sources of information for electricity,TV news was the most popular, followed by conversations with friends, relatives, and colleagues; Whatsapp; and newspapers. For the sources regarding information about SM, the same channels were the most common. In addition, the respondents also indicated receiving some information regarding SM through Youtube and Facebook. Out of N = 518 respondents, N2 = 319 respondents expressed a desire to look for more information regarding SM after participating in the survey (Q21–Q35). The following sources of information regarding SM were indicated by the respondents: TV news; friends, relatives, and colleagues; Facebook; WhatsApp; YouTube; and ofﬁcial government websites. A small number of respondents also indicated to look for information through search engines, radio, or Twitter and expressed an interest in attending workshops or educational sessions regarding SM. The use of social media was quite popular among respondents, especially Facebook, Youtube, and Instagram. Although respondents revealed to receive some information regarding SM through Facebook and Youtube, none of the respondents indicated to get any information through Instagram, which was the most common social media platform. Additionally, the mean of the responses showed that there was further scope to extensively improve the diffusion of information regarding SM through social media. Based on the ﬁndings, we have made some recommendations regarding various perspectives to increase the knowledge and acceptance of SM in Indonesia in Section5.

Table 4. Communication channels.

Var Mean (N = 518) Var Mean (N = 518) Var Mean (N1 = 130) Var Mean (N2 = 319) S01 0.47 S1 0.67 I31 0.67 Q21 0.58 S02 0.15 S2 0.04 I32 0.09 Q22 0.07 S03 0.06 S3 0.17 I33 0.27 Q23 0.2 S04 0.17 S4 0.23 I34 0.29 Q24 0.24 S05 0.9 S5 0.12 I35 0.24 Q25 0.16 S06 0.42 S6 0.01 I36 0.02 Q26 0.04 S07 0.64 S7 0.03 I37 0.1 Q27 0.07 S08 0.04 S8 0.18 I38 0.33 Q28 0.26 S09 0.03 S9 0.01 I39 0.02 Q29 0.02 S10 0.08 I40 0.49 Q30 0.32 S11 0.03 I41 0.15 Q31 0.08 S12 0.12 I42 0.12 Q32 0.14 S13 0 I43 0.09 Q33 0.06 S14 0.05 I44 0.02 Q34 0.02 S15 0.06 I45 0 Q35 0.05

3.4. Attitudes towards Buying New Technology, the Environment and Energy Saving (A1, A2, A31–A39, R1) The respondents were predominantly environmentally friendly, not in favor of buying new technology too frequently, and keeping track of their energy usage. Overall, 70.7% of respondents bought new technology once in over three years, while 58.9% bought a new mobile phone to keep up with the latest technology, once in over three years. This shows the trend for lasting and sustainable use of personal and home electronic gadgets. The majority of respondents regularly monitored their energy consumption (57.3%), having searched on the internet to know more about eco-friendly ways of living (67.4%), had returned home to check whether they turned off all home appliances or lights to save energy (67.8%), and have picked up trash left by someone else on the road side (79.5%). Some respondents also indicated to have paid more for energy efﬁcient appliances (39%), to have re-used grocery bags (42.7%), and were following organizations or proﬁles on social media which promote energy saving (17.8%). The installation of equipment to harness renewable energy was Resources 2019, 8, 177 13 of 23 found to be quite low, with 79.5% of respondents indicating to not have any such installation in their household.

3.5. Knowledge about SM, Stage of Installation and Social Inﬂuence (K1–K4, X1, W1) Almost three quarters (74.9%) of respondents did not have any knowledge about SM and a small group, only 6.6%, had a SM installed in their home. The number of respondents who were in the process of installing a SM or had a plan to install a SM in their homes were also very low, 5.8% and 13.7%, respectively. At the same time, 17.4% of respondents expressed that they wanted to have SM in their homes; however, these were the respondents who knew what a SM was. Social inﬂuence has a role to play in the dissemination of knowledge, as well as acceptance of smart meters [32] and, in this study, almost 30% of respondents who knew what a SM was indicated that their friends, relatives, or neighbors had a SM installed in their homes.

3.6. Preferences Regarding SM (P1–P4, G1–G3) The preferences of respondents were found to be inline with the various features and beneﬁts SM has to offer. A large majority of the respondents expressed a desire to receive more details about how they use electricity (76.3%), indicated that it would be useful for them to have real-time information about energy consumption (80.3%), and would like to have ﬂuctuating unit rates of electricity during the day so that they can use more electricity when it is cheaper (55.8%). Almost two-thirds of the respondents (74.7%) preferred to be able to remotely turn on or off the electricity supply through their mobile phones. Respondents who knew what a SM was were asked about their preferences on government policies regarding SM. About 50% of these respondents were in favor of the government making SM installations mandatory in homes. On the other hand, a higher number of these respondents (67.7%) would prefer that the government would offer SM as an option, instead of making its installation mandatory. About 46% of these respondents also expressed that they would protest if they were not given an option to say no to SM.

3.7. Concerns and Fears about SM (F1–F4) One of the concerns indicated by 69.7% respondents was the additional stress caused by the ﬂuctuations in energy rates. Another concern was connected with data privacy, which was expressed by 40.7% of respondents. At the same time, 40.3% did not indicate any concern regarding it, whereas the rest were not sure if it would be a cause of concern for them. Among the respondents who had knowledge about SM, 63.7% said that SM would make the billing process more accurate and 57.8% felt that SM would not have any adverse effect on their health.

3.8. Willingness to Accept/Adopt SM (De1–De8) Consumers’ acceptance of a certain technology is one of the main aspects of overall social acceptance of that technology. In Table5, the willingness to install SM under various conditions, such as ﬁnance, health, face-to-face advice, and social inﬂuence, are shown. These results were further used to create models for each decision and to understand other factors affecting Decisions De1–De8. Resources 2019, 8, 177 14 of 23

Table 5. Willingness to accept/adopt SM under various conditions (N = 518).

Code Question Mean Std Dev Yes % No % Not Sure % Accept if SM would help D 1 1.57 0.838 66.2 11 22.8 e save money Accept if SM would help De2 save money but might have 1.51 0.808 69.5 10.4 20.1 adverse effect on health Accept if SM would help save money, have no have De3 adverse effect on health but 1.47 0.796 71.6 9.3 19.1 companies have access to data of energy usage Accept if company representative visit home D 4 1.95 0.938 46.5 12 41.5 e and explains all details about SM Accept even if upgrade to De5 SM was not free but paid 1.92 0.886 43.1 21.5 35.4 (N1 = 130) Accept if upgrade of SM D 6 1.7 0.885 58.5 12.9 28.6 e was free Accept if De7 friends/relatives/neighbors 1.79 0.915 54.8 11.8 33.4 recommends SM Accept if De8 friends/relatives/neighbors 1.65 0.874 61.4 11.8 26.8 installs SM

4. Model, Results and Discussion

4.1. Modeling for Willingness and Acceptance of SM Respondents were asked about their willingness to use or install smart meters under certain conditions (De1–De8). We assumed that the willingness expressed showed their acceptance towards the installation of SM in their household. To understand the various factors inﬂuencing the willingness expressed by the respondents in this study, we examined the regression of Dei variables, with respect to the other variables, such as demographics, preferences, fears, and so on. For this purpose, we constructed eight separate Tobit regression models, for each Dei variable. These Tobit models, shown in Tables6–13, have a threshold of D i 1. Such a threshold emphasizes e ≤ that the relation between positive (“yes”) answers to the given decision alternative and the rest of the explanatory variables is investigated. The Tobit model assumes that the class number D i 1, 2, 3 e ∈ { } is a linear function of some exogenous variables, as in Equation (2). As we wanted to focus on the positive answers on decision alternatives, the model became

Dei∗ Dei 1 Dei = ≤ (1) ( 0 Dei > 1 where Dei∗ is a latent variable described by

Dei∗ = α + Xi β + εi, (2) where α is an intercept, Xi is a vector of exogenous variables excluding the constant, and εi as a residual. Resources 2019, 8, 177 15 of 23

The predictive capabilities of the model are as follows: for Model D 1: Log likelihood 443.79 e − and Chi-square 180.63 (61) with p = 0.000; for Model D 2: Log likelihood 408.505 and Chi-square e − 168.376 (61) with p = 0.000; for Model D 3: Log likelihood 384.653 and Chi-square 168.598 (61) e − with p = 0.000; for Model D 4: Log likelihood 621.393 and Chi-square 171.348 (61) with p = 0.000; e − for Model D 5: Log likelihood 124.892 and Chi-square 100.148 (61) with p = 0.001; for Model D 6: e − e Log likelihood 522.172 and Chi-square 178.645 (61) with p = 0.000; for Model D 7: Log likelihood − e 546.382 and Chi-square 192.802 (61) with p = 0.000; and for Model D 8: Log likelihood 489.383 and − e − Chi-square 192.286 (61) with p = 0.000. The value in the brackets accompanying the value of chi-square is the degree For each of Models De1–De8, only the variables which were statistically signiﬁcant (p < 0.05) are show in Tables6–13, respectively, while the rest of the variables were discarded. The interpretations of the models are described below in Sections 4.1.1–4.1.8.

4.1.1. Decision (De1) to Install SM If It Would Help Save Money When making a decision based on this condition, none of the demographic variables were found to be statistically significant. The model shows that users inclined towards spending more for energy efficient appliances and were more likely to favor the acceptance of SM if it would help save money. This indicated that they were ready to invest in energy saving appliances, expecting returns through savings, over a period of time. Preferences P3 and P4 were also found to be statistically significant for this decision, which suggested that the ability for users to turn the electricity supply on and off and fluctuating unit rates of electricity were positively co-related with this decision condition. Users expected to accept SM under this condition; however, they expressed a data privacy concern if the energy companies had access to the consumption data. Users interest to receive more information regarding SM was also positively co-related with this decision condition. Surprisingly, none of the communication channels were statistically significant for this model. The detailed statistics of this model, excluding the statistically insignificant variables, is shown in Table6.

Table 6. Estimation results of Tobit model for Variable De1 (N = 518).

Coefﬁcient Std. Error z p-Value A38 0.459298 0.201398 2.281 0.0226 P3 0.301042 0.147649 2.039 0.0415 P4 0.368473 0.121833 3.024 0.0025 F1 0.378323 0.127808 2.960 0.0031 Q1 0.416144 0.104268 3.991 0.0001 const 3.98193 1.36675 2.913 0.0036 − −

4.1.2. Decision (De2) to Install SM If It Would Help Save Money but Might Have Adverse Effect on Health According to the model, women seemed to share more concerns regarding their health, whereas men would be ready to accept SM, even if it might have adverse effects on their health. As for the communication sources, information received through energy companies would be highly positively evaluated, whereas searching on the internet was perceived negatively. One of the reasons for this might have been the lack of information suitable for the user to have a better understanding for SM, as there was a positive relation between attitudes towards searching on the internet for eco-friendly ways of living and the tendency to accept SM under this condition. As in the model for Decision De1, positive attitudes towards paying more to buy higher energy efficient appliances had a positive relation with Decision De2 but it had a higher impact and significance, as compared to Decision De1. The desire to look for more information regarding SM, receiving more details about electricity usage, and perceived usefulness of real-time information about energy Resources 2019, 8, 177 16 of 23 consumption would lead to a positive decision under this condition. The detailed statistics of this model, excluding the statistically insignificant variables, is shown in Table7.

Table 7. Estimation results of Tobit model for Variable De2 (N = 518).

Coefﬁcient Std. Error z p-Value D1 0.426938 0.207638 2.056 0.0398 − − S11 1.07450 0.527156 2.038 0.0415 S15 1.26069 0.527275 2.391 0.0168 − − A32 0.558343 0.217264 2.570 0.0102 A38 0.604122 0.220646 2.738 0.0062 P1 0.486778 0.199553 2.439 0.0147 P2 0.360111 0.178394 2.019 0.0435 Q1 0.439514 0.112342 3.912 0.0001 const 2.00271 1.37239 1.459 0.1445 − −

4.1.3. Decision (De3) to Install SM, If It Would Help Save Money, Had No Adverse Effect on Health but Companies Had Access to the Data of Energy Usage The concern about data privacy (F1) was not found to be statistically significant, confirming the decision under the condition of this model. Moreover, this model revealed that, for such users, income and type of residence was negatively related with Decision De3. This suggested that consumers with lower incomes and smaller households would be more interested in accepting SM under this condition. Features of SM, such as real time information of energy consumption and the ability to remotely turn electricity on or off, contributed positively to this decision alternative. These consumers were also interested in receiving more information regarding SM. The detailed statistics of this model, excluding the statistically insignificant variables, is shown in Table8.

Table 8. Estimation results of Tobit model for Variable De3 (N = 518).

Coefﬁcient Std. Error z p-Value D6 0.131180 0.0668288 1.963 0.0497 − − D9 0.566800 0.212691 2.665 0.0077 − − P2 0.364382 0.182679 1.995 0.0461 P3 0.351821 0.161554 2.178 0.0294 Q1 0.571528 0.117001 4.885 0.0000 const 2.79852 1.46845 1.906 0.0567 − −

4.1.4. Decision (De4) to Install SM If Company Representatives Would Visit Homes and Explain All the Details Table9 shows the detailed statistics of this model, excluding the statistically insignificant variables. The model revealed the segment of consumers who were highly concerned about data privacy, which is denoted through the statically significant positive relations with the fear, F1, as well as the negative influence of using social media platforms, such as LinkedIn and SnapChat. Apart from the visitation of a company representative, these consumers also trusted newspapers as a source of information, as it had a positive relation with this alternative decision. Consumers who owned a house were less likely to accept the installation of SM under this condition, whereas those who had a renewable energy source were more likely to accept the installation of SM, under this condition. The availability of remotely turning the electricity supply on or off also positively influenced this decision. These consumers were also interested in seeking more information regarding SM and this would increase the probability of acceptance of SM under this condition. Resources 2019, 8, 177 17 of 23

Table 9. Estimation results of Tobit model for Variable De4 (N = 518).

Coefﬁcient Std. Error z p-Value S03 0.651626 0.317034 2.055 0.0398 − − S08 0.878655 0.408831 2.149 0.0316 − − B1 0.277428 0.125653 2.208 0.0273 − − S3 0.397866 0.190581 2.088 0.0368 R1 0.396932 0.177439 2.237 0.0253 P3 0.252233 0.121232 2.081 0.0375 F1 0.350940 0.0997990 3.516 0.0004 Q1 0.346318 0.0821941 4.213 0.0000 const 2.09742 1.01987 2.057 0.0397 − −

4.1.5. Decision (De5) to Install SM Even If Upgrade to SM Was Not Free but Paid (N1 = 130) This model took into account the respondents who indicated to have knowledge of SM and had the highest number of influencing factors that were statistically significant. Table 10 shows that men who were mildly educated were more likely to accept paying for SM. They used social media platforms and had a higher likelihood to accept SM from social peers. They would be less likely to attend any workshops or seminars of awareness campaigns but were interested to receive information regarding SM. A highly negative relation of this decision, with the habit of monitoring energy, indicated that, those consumers who did not have a habit of monitoring energy consumption would be willing to pay for a device such as SM, which would make it easier for them to monitor energy consumption. This was further confirmed through the positive relation with the variable P2, indicating that they perceived receiving real time usage of energy consumption to be useful. Having a renewable source of energy installed was an additional motivation to accept SM under this condition. The Indonesian consumer does not have to pay for SM in the current plan. However, the results of this model would be useful in the case that the consumer would have to pay for SM.

Table 10. Estimation results of Tobit model for Variable De5 (N = 130).

Coefﬁcient Std. Error z p-Value D1 0.526502 0.266806 1.973 0.0485 − − D4 0.299052 0.133785 2.235 0.0254 − − S08 1.99339 0.906438 2.199 0.0279 S09 1.97660 0.588132 3.361 0.0008 B5 0.398364 0.169221 2.354 0.0186 − − S4 0.602266 0.303052 1.987 0.0469 S8 0.609118 0.287638 2.118 0.0342 − − S14 1.68988 0.747223 2.262 0.0237 − − A35 0.951932 0.322951 2.948 0.0032 − − R1 1.15989 0.349880 3.315 0.0009 P2 0.738628 0.316637 2.333 0.0197 Q1 0.549663 0.153058 3.591 0.0003 const 11.9877 882.376 0.01359 0.9892

4.1.6. Decision (De6) to Install SM If It Was a Free Upgrade Consumers with a higher number of children and those owning a flat or an apartment were less likely to accept SM, even if it was free, but the likelihood increased if they had invested in energy saving appliances before. Providing more details about usage of electricity and emphasizing on the availability of real time information of energy consumption would have a positive influence for the acceptance of SM under this condition. Addressing the concern for data privacy with more details and giving reassurance would also increase the probability of SM acceptance. Providing more details through conventional marketing mediums, such as newspapers and micro-blogging platforms, such as Twitter, would also aid in enhancing the acceptance of SM. As Indonesian consumers currently do not have to Resources 2019, 8, 177 18 of 23 pay to SM, the variable in this scenario would be more significant for them to address. The detailed statistics of this model, excluding the statistically insignificant variables, is shown in Table 11.

Table 11. Estimation results of Tobit model for Variable De6 (N = 518).

Coefﬁcient Std. Error z p-Value D81 0.215990 0.109612 1.970 0.0488 − − S04 0.570765 0.248447 2.297 0.0216 S08 1.14777 0.489555 2.345 0.0191 − − B2 0.399133 0.192045 2.078 0.0377 − − S3 0.468690 0.220941 2.121 0.0339 S8 0.484939 0.244529 1.983 0.0474 − − A34 0.593223 0.245430 2.417 0.0156 P1 0.365108 0.174351 2.094 0.0363 P2 0.363549 0.158399 2.295 0.0217 F1 0.492495 0.117634 4.187 0.0000 Q1 0.441846 0.0964924 4.579 0.0000 const 3.15554 1.20167 2.626 0.0086 − −

4.1.7. Decision (De7) to Install SM If It Was Recommended by Friends/Relatives/Neighbors Social influence was found to have a significant impact on the acceptance of technology. As the numbers in Table 12 show, consumers with higher household incomes were less likely to be influenced by social recommendations; however, if they possessed an attitude to pay more for energy saving appliances and were open to look for more information regarding SM, it would aid in increasing the acceptance of SM. Surprisingly, social recommendations through WhatsApp, search engines, LinkedIn, and SnapChat would have a negative influence on the acceptance of SM.

Table 12. Estimation results of Tobit model for Variable De7 (N = 518).

Coefﬁcient Std. Error z p-Value D6 0.111553 0.0495975 2.249 0.0245 − − S03 0.907934 0.391621 2.318 0.0204 − − S08 1.61036 0.509745 3.159 0.0016 − − S8 0.461269 0.222007 2.078 0.0377 − − S15 1.27823 0.405237 3.154 0.0016 − − A38 0.421198 0.170078 2.476 0.0133 F1 0.338865 0.109706 3.089 0.0020 Q1 0.524352 0.0889067 5.898 0.0000 const 2.47158 1.13916 2.170 0.0300 − −

4.1.8. Decision (De8) to Install SM If Friends/Relatives/Neighbors Installed It Social influence, through the action of peers, would have an impact on consumers who had lower monthly expenses on electricity, were willing to pay a higher price for more energy efficient appliances, and had a tendency to upgrade their appliance after long periods. The assurance regarding data privacy would lead to a higher acceptance of SM, especially if it was communicated directly by energy companies. The perceived usefulness of real time information on energy consumption and the ability to control the flow of electricity remotely, by turning it on or off, had a positive and statistically significant impact on the acceptance of SM. Higher use of social media platforms would have a positive influence on the willingness to accept SM; however, Messenger users would be more skeptical about it. The detailed statistics of this model, excluding the statistically insignificant variables, is shown in Table 13. Resources 2019, 8, 177 19 of 23

Table 13. Estimation results of Tobit model for Variable De8 (N = 518).

Coefﬁcient Std. Error z p-Value D7 0.0675710 0.0342000 1.976 0.0482 − − S02 0.623260 0.281593 2.213 0.0269 − − S04 0.576861 0.253421 2.276 0.0228 S11 1.29209 0.441498 2.927 0.0034 A1 0.246666 0.0782935 3.151 0.0016 A38 0.483869 0.184021 2.629 0.0086 P2 0.333437 0.160555 2.077 0.0378 P3 0.645629 0.138027 4.678 0.0000 F1 0.429969 0.117665 3.654 0.0003 Q1 0.205751 0.0968163 2.125 0.0336 const 3.10468 1.21801 2.549 0.0108 − −

5. Conclusions and Recommendations Indonesia has abundant energy resources but still faces problems and challenges to fulfill the increasing demand for electricity. It also suffers from high pollution and enormous pressures on environmental impacts, due to the increase in the use of coal to produce electricity, as well as vehicular emissions, which is a threat to the life expectancy of Indonesians. Positive steps have been taken by the Indonesian government to counter these issues by aiming to increase energy efficiency and develop AMI. To get the desired output, it is very important for consumers to have knowledge about smart meters—a major element of the AMI system, and to engage with interest and accept AMI applications. As per the recent studies in the literature, the outreach and communication was insufficient to increase the acceptance and awareness about SM for all segments of the population [17]. These studies also stated that the outreach and communication focusing on the benefits of SM for climate change would be particularly productive. Our study concentrated on the socioeconomic perspective for the willingness and acceptance of adopting SM by Indonesian residents. Initial descriptive statistics, regarding the willingness to accept or adopt, showed that consumers of different groups were very much in favor of accepting SM under various conditions, as shown in Table5, with only a small number showing an unwillingness to accept SM. There were relatively high numbers of respondents who were “Not Sure” whether they were willing to accept SM. Increasing the dissemination of detailed information regarding SM with the emphasis of its advantages and opportunities could shift the unsure consumer towards acceptance. This led us to our first recommendation, regarding creation of a briefing package for the consumers, in layman’s language, which would explain the details, features, and benefits of SM, before replacement of standard meters. The package could be through mobile applications, or a digital package, through email or even through a community briefing by a company representative. If the package addresses the statistically significant variables stated in our Models De1–De8 elaborated in Section 4.1, it would help in increasing the consumers perceived usefulness of SM, as well as their acceptance and engagement. The package may also contain ready content, such as infographics or images with features/benefits of SM, that the consumers could share with their connections through the most popular messengers in Indonesia, such as, FB messenger, Line messenger, or WhatsApp (Table4). We were surprised to notice that respondents in this study were ready to accept SM, even if they were given a hypothetical situation where their electricity usage data could be accessed by energy companies, which was in contradiction with the finding by Chou et al. [9], stating that data privacy was one of the major concerns among consumers. The reason for this difference of opinion was that the study by Chou et al. [9] was conducted among the general population, whereas the current study was conducted among social media users. Regular social media users were seen as early adopters of new technology and the findings from this study also showed that they were more aware about privacy concerns as well. Resources 2019, 8, 177 20 of 23

This led us to the second recommendation, with regards to the target audience for raising the awareness and knowledge about SM. It would be fruitful to aim for dissemination to social media users first, due to the ease with which they can be reached at the relatively lower cost. This target group (i.e., social media users) includes 56% of Indonesia’s populations and is growing quickly. We found that social influence among lower and middle class communities, which form the majority of Indonesia’s population, would have a high impact on the acceptance of SM. The snowball effect, through personal and social connections, online or offline, of social media users, would enhance the required outcome manifold. Currently, PLN—the national electricity company undertaking the roll-out of SM—has profiles/pages on social media platforms and a certain number of followers (Facebook, more than 10,000; LinkedIn, 46,279; and YouTube, more than 5680) but there are no activities or engagement with the consumers. On YouTube, certain videos have been shared, whereas, on Facebook, the last activity was in 2010 and no posts on LinkedIn were observed. Utilizing the existing network, through activities and engagement, while growing the network on other social media platforms would be quite effective for increasing the acceptance of SM. The results from the modeling, for willingness to accept SM, revealed various factors that would play an important role in the acceptance of SM under certain conditions. As the consumer in Indonesia does not have to pay for upgrading to SM, according to the current policy, emphasis on factors, such as energy saving resulting in savings on the cost of electricity and how the consumers could achieve it, would push the consumer towards accepting SM. The models also revealed that, for certain consumers, data privacy was still a concern but, as the responses to the Variables De1–De8 indicated, even with this concern, consumers would be willing to accept SM. Addressing the data privacy concern through the visitation of a company representative or social influence through friends, family, and peers, would further enhance the acceptance of SM. Social influence was found to be a vital factor that would increase the acceptance of SM. Attitudes of the consumers towards energy efficient purchases or willingness to pay more for energy saving appliances had a major direct effect on the consumer’s acceptance of SM. Even though the study was conducted among social media users, the importance of dissemination of information through conventional channels, such as newspapers, TV news, radio, and direct communication through energy companies, has shown itself to be highly effective. Findings from the literature (see [17]), as well as our findings, showed that there was a very low level of consumer awareness regarding AMI, in general, or SM, in particular, and communication channels had a huge role to play in this part. Social media was especially significant, as over 56% of the Indonesian population was active on social media and the dissemination of information through social media channels is relatively easier, less time consuming, and cost effective [48]. Based on these findings, our third recommendation is regarding communication channels for raising awareness and knowledge among consumers regarding SM. The use of digital media, not just social media, and conventional marketing channels would prove to be effective in reaching out to a greater number of customers. The reason for this recommendation is the choice of channels selected by the respondents in this study, which suggested that they would like to receive information by searching online, through TV news, newspapers and so on, apart from social media platforms. We also recommend having articles on the most visited news websites (Table4), such as tribunnews. com, detik.com, liputan6.com, kompas.com, sindonews.com, okezone.com, idntimes.com and so on. These articles could compliment the briefing package we suggested in the first recommendation.

6. Limitations and Future Scope of Research There are certain limitations to this study, although vigorous efforts were made to expand the scope of this research. Respondents were located from almost all regions of Indonesia in this study, but we did not consider the regional effect on the willingness or the knowledge of SM. Further in-depth research could be conducted independently in different regions of Indonesia and a comparative analysis of these regions would help to get more reﬁned results of this study, speciﬁcally for individual regions. Resources 2019, 8, 177 21 of 23

The future scope of this research also encompasses the testing of communication channels and their signiﬁcance through awareness campaigns, which would show the responsiveness of the consumer to various channels.

Author Contributions: Y.C. conceived and designed the survey; A.K.-P. reviewed the design of the survey; W.W. translated the questionnaire and the collected data into Indonesian; Y.C. created and managed the online questionnaire; W.W. collected the data through the online questionnaire; Y.C. analyzed the data; A.K.-P. reviewed the literature; A.K.-P. and Y.C. drafted and edited the paper; and A.K.-P. and Y.C. reviewed the paper. Funding: This work was supported by the Faculty of Computer Science and Management, Wrocław University of Science and Technology from funds of the Ministry of Science and Higher Education subsidy in the part devoted to conducting research activities in 2019. Conﬂicts of Interest: The authors declare no conﬂict of interest.

Abbreviations The following abbreviations are used in this manuscript:

AMI Advanced Metering Infrastructure PLN Perusahaan Listrik Negara SM Smart Meters SG Smart Grids SDGs Sustainable Development Goals

References

1. Fadaeenejad, M.; Saberian, A.; Fadaee, M.; Radzi, M.; Hizam, H.; AbKadir, M. The present and future of smart power grid in developing countries. Renew. Sustain. Energy Rev. 2014, 29, 828–834. [CrossRef] 2. Barrios-O’Neill, D.; Schuitema, G. Online engagement for sustainable energy projects: A systematic review and framework for integration. Renew. Sustain. Energy Rev. 2016, 54, 1611–1621. [CrossRef] 3. Crispim, J.; Braz, J.; Castro, R.; Esteves, J. Smart Grids in the EU with smart regulation: Experiences from the UK, Italy and Portugal. Util. Policy 2014, 31, 85–93. [CrossRef] 4. Nizetic, S.; Djilali, N.; Papadopoulos, A.; Rodrigues, J. Smart technologies for promotion of energy efﬁciency, utilization of sustainable resources and waste management. J. Clean. Prod. 2019, 231, 1–10. [CrossRef] 5. Zhou, S.; Brown, M. Smart meter deployment in Europe: A comparative case study on the impacts of national policy schemes. J. Clean. Prod. 2017, 144, 22–32. [CrossRef] 6. Avancini, D.; Rodrigues, J.; Martins, S.; Rabêlo, R.A.L.; Al-Muhtadi, J.; Solic, P. Energy meters evolution in smart grids: A review. J. Clean. Prod. 2019, 217, 702–715. [CrossRef] 7. Chawla, Y.; Kowalska-Pyzalska, A.; Silveira, P. Marketing and Communications Channels for Diffusion of Smart Meters in Portugal; Working Paper; Hugo Steinhaus Center, Wroclaw University of Technology HSC Research Reports HSC/19/; Wrocław University of Science and Technology: Wroclaw, Poland, 2020. 8. Bhatt, J.; Jani, O. Smart Development of Ahmedabad-Gandhinagar Twin City Metropolitan Region, Gujarat, India; Springer: Singapore, 2019; pp. 313–356. 9. Chou, J.; Kim, C.; Ung, T.; Novi Yutami, G.; Lin, G.; Son, H. Cross-country review of smart grid adoption in residential buildings. Renew. Sustain. Energy Rev. 2015, 48, 192–213. [CrossRef] 10. Acharjee, P. Strategy and implementation of Smart Grids in India. Energy Strateg. Rev. 2013, 1, 193–204. [CrossRef] 11. Biresselioglu, M.; Nilsen, M.; Demir, M.; Royrvik, J.; Koksvik, G. Examining the barriers and motivators affecting European decision makers in the development of smart and green energy technologies. J. Clean. Prod. 2018, 198, 417–429. [CrossRef] 12. Colak, I.; Bayindir, R.; Fulli, G.; Tekin, I.; Demirtas, K.; Covrig, C.F. Smart grid opportunities and applications in Turkey. Renew. Sustain. Energy Rev. 2014, 33, 344–352. [CrossRef] 13. Dutu, R. Challenges and policies in Indonesia’s energy sector. Energy Policy 2016, 98, 513–519. [CrossRef] 14. Chou, J.; Novi Yutami, G. Smart meter adoption and deployment strategy for residential buildings in Indonesia. Appl. Energy 2014, 128, 336–349. [CrossRef] Resources 2019, 8, 177 22 of 23

15. Maulidia, M.; Dargusch, P.; Ashworth, P.; Ardiansyah, F. Rethinking renewable energy targets and electricity sector reform in Indonesia: A private sector perspective. Renew. Sustain. Energy Rev. 2019, 101, 231–247. [CrossRef] 16. Hidayatno, A.; Rahmawan Destyanto, A.; Arauna Hulu, C. Industry 4.0 Technology Implementation Impact to Industrial Sustainable Energy in Indonesia: A Model Conceptualization. Energia Procedia 2019, 156, 227–233. [CrossRef] 17. Bugden, D.; Stedman, R. A synthetic view of acceptance and engagement with smart meters in the United States. Energy Res. Soc. Sci. 2019, 47, 137–145. [CrossRef] 18. Perindustrian, K. Making Indonesia 4.0; Ministry of Industry, Republic of Indonesia: Jakarta, Indonesia, 2018. 19. Frederiks, E.; Stenner, K.; Hobman, E. Household energy use. Applying behavioral economics to understand consumer decision making and behavior. Renew. Sustain. Energy Rev. 2014, 41, 1385–1394. [CrossRef] 20. Ellabban, O.; Abu-Rub, H. Smart grid customers’ acceptance and engagement: An overview. Renew. Sustain. Energy Rev. 2016, 65, 1285–1298. [CrossRef] 21. Geelen, D.; Reinders, A.; Keyson, D. Empowering the end-user in smart grids: Recommendations for the design of products and services. Energy Policy 2013, 61, 151–161. [CrossRef] 22. Hyysalo, S.; Johnson, M.; Juntunen, J. The diffusion of consumer innovation in sustainable energy technologies. J. Clean. Prod. 2017, 162, S70–S82. [CrossRef] 23. Kamarudin, Y.; Boothman, T. Powering the Nation: Indonesian Power Industry Survey. 2017. Available online: https://www.pwc.com/id/en/energy-utilities-mining/assets/power/power-survey-2017.pdf (accessed on 14 November 2019). 24. Kowalska-Pyzalska, A. What makes consumers adopt to innovative energy sources in the energy market? A review of incentives and barriers. Renew. Sustain. Energy Rev. 2018, 82, 3570–3581. [CrossRef] 25. Kahma, N.; Matschoss, K. The rejection of innovations? Rethinking technology diffusion and the non-use of smart energy services in Finland. Energy Resour. Soc. Sci. 2017, 34, 27–36. [CrossRef] 26. Verbong, G.P.J.; Beemsterboer, S.; Sengers, F. Smart grids or smart users? Involving users in developing a low carbon electricity economy. Energy Policy 2013, 52, 1175–125. [CrossRef] 27. Good, N.; Ellis, K.; Mancarella, P. Review and classiﬁcation of barriers and enablers of demand response in the smart grid. Renew. Sustain. Energy Rev. 2017, 16, 57–72. [CrossRef] 28. Schleich, J.; Faure, C.; Klobasa, M. Persistence of the effects of providing feedback alongside smart metering devices on household electricity demand. Energy Policy 2017, 107, 225–233. [CrossRef] 29. Krishnamutri, T.; Schwartz, D.; Davis, A.; Fischoff, B.; de Bruin, W.B.; Lave, L.; Wang, J. Preparing for smart grid technologies: A behavioral decision research approach to understanding consumer expectations about smart meters. Energy Policy 2012, 41, 790–797. [CrossRef] 30. Foulds, C.; Robison, R.; Macrorie, R. Energy monitoring as a practice: Investigating use of the iMeasure online energy feedback tool. Energy Policy 2017, 104, 194–202. [CrossRef] 31. Kowalska-Pyzalska, A.; Byrka, K. Determinants of the willingness to energy monitoring by residential consumers: A case study in the city of Wroclaw in Poland. Energies 2019, 12, 907. [CrossRef] 32. Chawla, Y.; Kowalska-Pyzalska, A. Public awareness and consumer acceptance of smart meters among Polish social media users. Energies 2019, 12, 2759. [CrossRef] 33. Buchanan, K.; Russo, R.; Anderson, B. Feeding back about eco-feedback: How do consumers use and respond to energy monitors? Energy Policy 2014, 73, 138–146. [CrossRef] 34. Wolsink, M. Distributed generation for sustainable energy as a common pool resource: Social acceptance in rural setting of smart (micro-) grid conﬁgurations. In New Rural Spaces: Towards Renewable Energies, Multifunctional Farming, and Sustainable Tourism; Frantal, B., Martiant, S., Eds.; University of Amsterdam: Amsterdam, The Netherlands, 2014; pp. 36–37. 35. Burchell, K.; Rettie, R.; Roberts, T. Householder engagement with energy consumption feedback: The role of community action and communications. Energy Policy 2016, 88, 178–186. [CrossRef] 36. Ma, G.; Lin, J.; Li, N. Longitudinal assessment of the behavior-changing effect of app-based eco-feedback in residential buildings. Energy Build. 2018, 159, 486–494. [CrossRef] 37. Vassileva, I.; Campillo, J. Consumers’ perspective on full-scale adoption of smart meters: A case study in Västerås, Sweden. Resources 2016, 5, 3. [CrossRef] 38. Nachreiner, M.; Mack, B.; Matthies, E.; Tampe-Mai, K. An analysis of smart metering information systems: A psychological model of self-regulated behavioral change. Energy Res. Soc. Sci. 2015, 9, 85–97. [CrossRef] Resources 2019, 8, 177 23 of 23

39. Rogers, E. Diffusion of Innovations, 5th ed.; Free Press: New York, NY, USA, 2003. 40. Sheau-Ting, L.; Mohammed, A.; Weng-Wai, C. What is the optimum social marketing mix to market energy conservation behaviour: An empirical study. J. Environ. Manag. 2013, 131, 196–205. [CrossRef][PubMed] 41. Hudson, S.; Huang, L.; Roth, M.; Madden, T.J. The influence of social media interactions on consumer–brand relationships: A three-country study of brand perceptions and marketing behaviors. Int. J. Res. Mark. 2016, 33, 27–41. [CrossRef] 42. Alalwan, A.; Rana, N.; Dwivedi, Y.; Algharabat, R. Social media in marketing: A review and analysis of the existing literature. Telemat. Inf. 2017, 34, 1117–1190. [CrossRef] 43. Burke, P.; Widnyana, J.; Anjum, Z.; Aisbett, E.; Resosudarmo, B.; Baldwin, K. Overcoming barriers to solar and wind energy adoption in two Asian giants: India and Indonesia. Energy Policy 2019, 132, 1216–1228. [CrossRef] 44. Kennedy, S. Indonesia’s energy transition and its contradictions: Emerging geographies of energy and finance. Energy Res. Soc. Sci. 2018, 41, 230–237. [CrossRef] 45. Al Irsyad, M.; Halog, A.; Nepal, R. Estimating the impacts of financing support policies towards photovoltaic market in Indonesia: A social-energy-economy-environment model simulation. J. Environ. Manag. 2019, 230, 464–473. [CrossRef] 46. Gerpott, T.; Paukert, M. Determinnants of willingness to pay for smart meters: An empirical analysis of household customers in Germany. Energy Policy 2013, 61, 483–495. [CrossRef] 47. Paetz, A.G.; Duetschke, E.; Fichtner, W. Smart homes as a means to sustainable energy consumption: A study of consumer perceptions. J. Consum. Policy 2012, 35, 23–41. [CrossRef] 48. Chodak, G.; Chawla, Y.; Dzidowski, A.; Ludwikowska, K. The Effectiveness of Marketing Communication in Social Media. In ECSM 2019 6th European Conference on Social Media; Academic Conferences and Publishing Limited: Sonning Common, UK, 2019; p. 73.

c 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access

article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). Paper 5

G. Chodak, Y. Chawla, A. Dzidowski, K. Ludwikowska The Effectiveness of Marketing Commu- nication in Social Media The Effectiveness of Marketing Communication in Social Media

Grzegorz Chodak1, Yash Chawla1, Adam Dzidowski2, and Kamila Ludwikowska3 1Department of Operational Research, Finance and Information Technology, Faculty of Computer Science and Management, Wroclaw University of Science and Technology, Poland 2Department of Management Systems, Faculty of Computer Science and Management, Wroclaw University of Science and Technology, Poland 3Department of Humanities and Social Sciences, Wroclaw University of Science and Technology, Poland [email protected] [email protected] [email protected] [email protected]

Abstract: The paper proposes the metrics for marketing communication in social media. The metrics are based on typical social media engagement measurements, but are calculated in relation to the concepts of content quality, valence and volume. The authors present a social media marketing communication experiment implemented in a real business environment of the e-shop Facebook fan page. The experiment results show that different effects were obtained depending on the content of posts (picture, video, album) related to the same products. In the case of video, the highest level of reach, while in the case of pictures, the highest level of engagement was achieved. The analysis of social media engagement metrics in relation to the published content type, dissemination process and time were proposed. The obtained results clearly show how complicated the analysis of social media communication could be and that the metrics have to be carefully designed to capture the relevant outcomes and aims.

Keywords: social media marketing communication, types of content on social media, social media experiment, Facebook users 1. Introduction There is a significant increase in social media users that is noticeable across the globe. The study of Pew Research Centre in America, concerning social media usage, reveals a constant growth in the number of users in the upcoming years (Perrin, 2015). The power of the social media ecosystem is embodied in the sheer numbers of users - 3.196 billion (Kemp 2018) but its high importance is related to the fact that it connects online and offline elements of market reality (Hanna et al., 2011). Consequently, activities in social media often become the basis of the marketing strategy of enterprises. The studies of social media research over a period of 20 years clearly show its growing importance for businesses (Kapoor et al., 2018). In a world where one person can communicate about products, companies or brands with hundreds or even thousands of peers, the impact of consumer engagement has been greatly magnified (Liu et al. 2018). In current culture, when planning the reach of the marketing content, managers are dependent on the users and the algorithms of social media platforms (Kanuri et. al., 2018). This is in opposition to traditional integrated marketing communications paradigm (Mangold et al., 2009). Especially Facebook (FB), with more than 2 billion active users, has proven to be a very popular platform to market products, promote brands, manage relationships and lead discourse with customers (Chodak & Suchacka, 2017), (Myers, 2018).

However, the studies concerned with social media communication often do not take into consideration the effectiveness of communication for a variety of content, levels of interaction and time of posting. There is also a lack of agreement about which social media marketing indicators define communicational performance (Lamberton & Stephen, 2016). The purpose of this article is to analyze the effectiveness of different types of content (video, image and album) in generating user engagement, through organic and paid promotions. The authors also took into consideration time of posting and the number of fans online, to analyze the reach and engagement patterns for different types of content. Thereafter, the metrics for indicating engagement patterns among different types of content are proposed. Subsequently, the set-up experiment, results, conclusion and future scope of research are presented.

Grzegorz Chodak et al.

2. Literature background The evolution of the social networking platforms gave users the medium to share content in various forms, such as text, visuals, audio-visuals, weblinks etc. General impact of different kinds of content on the brand’s page have been studied and shown to exhibit variations in engagement levels among different types of content (de Vries et al. 2012), (Myers, 2018), (Kanuri et. al., 2018). In this regard, the literature presents further determinations of the type of social media content and its placement for most-desired business outcomes, for best brand awareness and sales (Salo, 2017). Researches show that the type of content and the time of posting affects the reach as well as engagement, but there is a need to establish metrics for linking the type of content with respective engagement patterns (Dolan et al. 2017). User engagement, such as likes, comments, shares and clicks, is the basis on which the reach of the content is determined. The higher the content engagement, the higher the reach is obtained (Lipsman et al., 2012). Social Media platforms keep updating and modernizing their algorithms to make the content more relevant for the users. Through an analysis of EdgeRank, the algorithm structuring the flow of information and communication on Facebook’s ‘News Feed’, researchers have argued that distribution of content is biased based on its type (Bucher, 2012). There has been research comparing different types of content used by global brands in social media marketing. During one of such studies, it was observed that photos are more effective in drawing user interaction as compared to video or text (Kim et al., 2015).

To gain deeper insights on social media marketing effectiveness, the authors studied various metrics. All social media platforms provide some measurements that help to understand users’ engagement and they are provided in the simplest possible manner to ensure that no expertise is necessary to analyze them. Although researchers have raised the question of needed expertise and adequacy of widely available metrics (Baym, 2013), the engagement is one of the most popular metrics for social media marketing. However, further research on that measurement is clearly required (Barger & Labrecque, 2013). Based on those findings, authors decided to propose and calculate metrics focusing on effectiveness and interaction patterns among Facebook users.

The effectiveness of communication is defined by the ISO 9000: 2005 standard, as the degree of achieving the planned goals. The term effectiveness reaches praxeology approach - the theory of efficient action (Kotarbinski, 2013). Accordingly, to general meaning, the efficacy describes each component of „good work” as constituting: effectiveness, favorableness and economy. The fundamental form of efficient action is effectiveness, described as compatibility of the action with intended aim (Pszczolowski, 1976). The action can be effective when all performed activities enable reaching set goals. This can be measured as the level of reaching goals or the degree of approach to reach them (defined as purposefulness). Effectiveness can be characterized by different intensity, for instance, when the goal is reached partially, the action is also partially effective and, when the goal is not achieved, the action is not efficacious. Users presence on social media pages is not a sufficient indicator for marketing communications. Its effectiveness should rather relate to what the users pay attention to. In their extensive review of social media metrics, Peters et al. (2013) distinguish four dimensions of social media analysis: motives, content, network structure, and social roles and interactions. In that perspective, content has three distinct aspects: quality (e.g. interactivity, vividness, education, entertainment, information), valence (e.g. emotions, tonality, rating variance), and volume (counts and volumes). Since the purpose of the presented experiment was to measure the difference between the interaction patterns of FB users, among different types of content, the proposed metrics should focus on that area. 3. Proposed metrics The metrics used in the experiment focus mainly on volume, with less focus on the type (quality) of the content, omitting the valence dimension since Facebook generally is based on positive endorsement rather than rating (negative reactions are possible, but they were negligible during the devised experiment for this study). Despite the fact that volume related measurements are easiest to obtain and calculate, they could be treated as quality indicators, especially when calculations are based on effectiveness of observed actions and mutual relations between the reach, reaction and engagement rate. In order to capture these considerations in a way that would help to choose an effective type of the content, authors propose the metrics illustrated in the following paragraphs.

For each advertised product title t (t = 1, 2, … n), the following variables were defined:

Grzegorz Chodak et al.

t t t R video, R album, R image – the number of FB users, to whom the video, album and image posts of the board game t, reached respectively, during the whole experiment. t t t L video, L album, L image – the number reactions (Likes, Comments and Shares) by FB users that were observed in the video, album and image posts of the board game t, respectively, during the whole experiment.

t t t C video, C album, C image – the number of clicks by FB users that were recorded on the video, album and image posts of the board game t, respectively, during the whole experiment.

t t R organic, R paid – the number of FB users to whom the posts related to the board game reached through the organic and paid dissemination, respectively.

t t L organic, L paid – the number of reactions (Likes, Comments and Shares) by FB users that were observed on the posts of the board game t, when it reached them through the organic and paid dissemination, respectively.

t t C organic, C paid – the number of clicks by FB users, that were recorded on the posts related to the board game t, when it reached them through the organic and paid dissemination, respectively.

Rf, Rnf – the number of fans and non-fans reached during the whole experiment.

Tpost – the specific time of the day, when the posts were published. Based on the typical measurements and the dimension of social media content analysis, the following metrics were proposed: Metrics related to goal attainment that focus on conversion rate (St - the number of advertised units of given type sold during the experimental campaign): Effectiveness of the campaign – Et = St / Ct (related to the planned conversion rate, compared between campaigns or compared between advertised units)

t t t t t t t Effectiveness of the content type – E video, E album, E image = S / C video, C album, C image (mutually compared)

t t t t t Effectiveness of the campaign dissemination – E organic, E paid = S / C organic, C paid (mutually compared) Metrics related to content valence that describe the attractiveness of the content and the willingness to interact (like, comment or share), especially among new followers:

t t t Attractiveness of the campaign – A = [(L nf / Rnf) - (L f / Rf)] (compared between campaigns)

t t t t t t t t t Attractiveness of the content type – A video, A album, A image = L video, L album, L image / R video, R album, R image (mutually compared)

t t t t t t Attractiveness of the campaign dissemination – A organic, A paid = L organic, L paid / R organic, R paid (mutually compared Metrics related to content quality that illustrate the power of the content message in regard to the desired action (click): Persuasiveness of the campaign – Pt = Ct / Rt (compared between campaigns)

t t t t t t t t t Persuasiveness of the content type – P video, P album, P image = C video, C album, C image / R video, R album, R image (mutually compared)

t t t t t t Persuasiveness of the campaign dissemination – P organic, P paid = C organic, C paid / R organic, R paid (mutually compared) Metrics related to content volume concerned with content publication dynamics:

t t Intensity of the campaign – I = ΔR / ΔT (compared between campaigns or compared between time frames)

t t Intensity of attractiveness – I A = ΔL / ΔT (compared between campaigns or compared between time frames)

t t Intensity of persuasiveness – I P = ΔC / ΔT (compared between campaigns or compared between time frames) All proposed metrics can be mutually compared and further analyzed (e.g., the campaign could be attractive, however neither persuasive, nor effective; campaign could be less intensive but more effective; content type could be persuasive but not attractive etc.).

Grzegorz Chodak et al.

4. Experimental design The purpose of this experiment was to measure the difference between the interaction patterns of FB users, among different types of content. The experiment was conducted in collaboration with a Polish e-commerce store, which sells various board games, books, films and other products. The name of the company is not given due to the non-disclosure agreement and is simply called “E-Store”. The experiment was carried out on the Facebook fan page of the E-Store, which, at the time of initiating the experiment, had 4422 fans and 4439 fans at the end of experiment. The E-Store’s Facebook page is one of the two main advertising channels. Facebook posts are used as a medium to reach out to the customers and inform them about various products and novelties available online in the store. Facebook is a two-way medium of communication, hence it proves beneficial for the E-Store to obtain direct feedback from actual and prospective customers. This feedback is vital to understand the needs and opinions of the customers about the range of products. The E-Store also uses the FB fan page to disseminate special discount coupons, spread information about promotions and organize competitions with prizes for customers. In this particular experiment, nine board games were randomly selected out of numerous games available on from the E-Store. All of these board games were similar in type and genre and, for each board game, three types of posts were created: (i) Single Image + Short Description as Caption + Link to the product on E-Store, (ii) Video + Short Description as Caption + Link to the product on E-Store and (iii) Album of Images + Short Description as Album Caption + Link to the product on E-Store.

For each advertised board game title t (t = 1, 2, 3, ..., 8, 9), three types of posts: Video Vt, Image It and Album At, were published in the order V1, V2, V3, I1, I2, I3, A1, A2, A3, I4, I5, I6, A4, A5, A6, V4, V5,V6, A7, A8, A9, V7, V8, V9, I7, I8 th and I9, totaling 27 posts. The experiment started with the first post at 12:01am on 7 December 2018, with each proceeding post being published at an interval of 54 minutes. To give more diversity and have comparative analysis, the experiment was executed in two parts: Organic and Paid. At first, the posts spread organically for 96 hours and readings were observed for each post exactly after the time interval for each post (on 11th December 2018). Thereafter, each post was promoted with a budget of 10 PLN for the next 96 hours (2.5 PLN / day), at the end of which the observations were recorded again. Posts of one board game (t = 6) was skipped from paid promotion to observe the effect on its sales as well as on posts before and after this title. The rationale for such design of the experiment was: (i) not to flood the feed of fans with all the posts at the same time but to give them new content at equal time intervals; (ii) to give variety in type of content to the viewers; (iii) vary the time of posting to cater to fans online at different times of the day and (iv) to reach to fans as well as non-fans of the page. Through the experiment, we wanted to calculate the metrics as well as answer the following questions: Are there any differences in post reach and FB users’ reactions to ads with different type of content (image, video and album)? Is there an effect of the time of posting and number of fans online on the reach, reactions and engagement rate? We also had some additional aims: To analyze the relationship between the paid and organic impressions with respective number of paid and organic engagement (sum of likes, comments, shares and clicks). To analyze the relationship between the reach to fan and non-fans with the regard to paid reach and organic reach. 5. Experimental results At the beginning, the volume of sales of each board game marketed during the experiement through the E-Store were roughly analyzed to show the scale of sales. Data of sales achieved for each individual board game title during the experiment period (7th Dec 2018 to 15th Dec 2018) are shown in Table 1. Table 1: Data on the sales of board games from the E-Store during the period of experiment (7th to 14th Dec 2018) Title (t) 1 2 3 4 5 6 7 8 9 Total Sales Units Sold 2 4 5 6 0 2 1 0 2 22 An interesting observation from this data was that the title t=6 had 2 units sold during the experiment period. This title was not subjected to paid promotion and was only allowed to disseminate organically, yet its sales is close to the mean sales (2.44) of all the titles in the experiment.

Grzegorz Chodak et al.

Table 2: Recorded number of results for reach, reactions and clicks based on type of content (video, album, image)

t t t t t t t t t Title (t) R video R album R image L video L album L image C video C album C image 1 1520 1174 918 5 16 53 35 79 58 2 1588 1006 754 5 14 12 60 31 33 3 1462 1256 1059 5 13 16 38 86 66 4 1372 754 814 6 11 15 28 27 35 5 1252 1059 893 2 9 9 26 33 57 6 359 277 636 3 0 11 17 8 29 7 1331 1647 1153 0 23 34 17 130 89 8 1184 1124 1341 6 10 31 17 47 88 9 1351 1186 1386 3 15 99 10 55 117 Sum 11419 9483 8954 35 111 280 248 496 572 Mean 1268.8 1053.7 994.9 3.9 12.3 31.1 27.6 55.1 63.6 Std Dev 363.8 375 259.3 2.02 6.2 29.2 15.3 37.5 29.8 As it can be seen in Table 2, there are significant differences between Reach, Reactions and Clicks, among there being three types of posts. What is interesting to note is that videos obtained the highest Reach, whereas the lowest were Reactions and Clicks. It is an interesting result in the context of measuring the efficiency of social media activities. Average engagement for the post with an image and a caption is the highest among all the three types of posts studied in the experiment. At the same time, the standard deviation for reactions on such types of posts is high, which depicts that the engagement lacks consistency.

Figure 1: Moving average for fans online, organic reach, paid reach and total reach with respect to posting time During the whole day, at different times, there are varying numbers of FB users online and may change during the week as well. For businesses, this data provided by Facebook is very important to figure out when their fans are online and then schedule posts to reach out to the maximum number of FB users possible. To observe the effect of number of Fans online, on organic and paid dissemination, the authors decided to post different types of content at different times of the day. Figure 1 shows the plot for the moving average number of fans online, organic reach, paid reach, total reach and engagement. Interpretations of the plot show that the moving averages for number of fans online and organic reach have quite similar slopes but the total engagement does not seem to have direct correlations. 2500 2 per. Mov. Avg. (No. Of Fans Online) 2000 2 per. Mov. Avg. (Reach to Fans) 1500 2 per. Mov. Avg. (Organic Impression) 1000 500 0

Time of the Day Figure 2: Plot of moving average for fans online, organic impression to FB users and organic engagement

Grzegorz Chodak et al.

Figure 2 shows that the number of fans online do have an impact on the organic reach of the posts, which influences the organic reach, organic impression as well as engagement. However, for some posts, the patterns don’t match, which leaves the scope of quality of content. Table 3: Recorded number of engagement based on type of content (video, album and image) Title (t) 1 2 3 4 5 6 7 8 9 Sum Video 5 5 5 6 2 3 0 6 3 36 Reactions (Lt) Photo 53 12 16 15 9 11 37 31 93 383 Album 16 14 13 11 9 0 37 10 15 125 Table 3 shows the number of engagements recorded for different types of content for each advertised boardgame title. It can be observed that the engagement results are in cotradiction with the results for reach. Reach is the highest for video and least for photo, but engagement is highest of the photo and least for video. Table 4: Pearson Correlations between types of content considering total reach and reactions of posts

Total reach of posts Reactions to posts Pearson Pearson Correlations p-value Correlations p-value Video/Photo 0.316 0.407 -0.07553 0.846857 Photo/Album 0.661 0.052 0.50061 0.169868 Video/Album 0.698 0.036 -0.25886 0.501228 Table 4 shows the Pearson Correlations between types of content and reactions to posts. We are aware that for such a small sample size, statistical analysis, in particular the Pearson correlation calculation, is disputable. However, despite some doubts, we decided to carry out these calculations in order to show some relations between the variables. It is suprising that values of correlations are so low because one may expect that differences among products will be proportional independent from the form of posts. P-values show that only correlation between Video and Album post (taking into consideration total reach of posts) are statistically significant. The aim of this calculation was to show that, for the same products, posts presented in different forms received different reach and reactions from customers. Therefore, we can conclude that, in this experiment, the title of product did not determine the level of reach and reactions to posts independently from types of content.

8.00% 7.00% 6.00% Video Album Photo 5.00% 4.00% 3.00% 2.00% 1.00% 0.00% 1 2 3 4 5 6 7 8 9

Title of the Board Games (t)

Figure 3: Percentage of FB users who were reached through video, album and image type of content and they liked, commented or shared the post For the title t=6, dissemination was only organic and it can be seen that albums drew no reaction, photos drew the highest reaction but were lower as compared to the posts where organic and paid disseminations were carried out. The reactions observed for this title’s video post were highest as compared to all other title’s video posts. This shows that when videos are allowed to disseminate organically for longer periods, they tend to have a higher reaction rate.

Grzegorz Chodak et al.

16.00% 14.00% Video Album Photo 12.00% 10.00% 8.00% 6.00% 4.00% 2.00% 0.00% 1 2 3 4 5 6 7 8 9 Title of the Board Games (t)

Figure 4: Percentage of FB users who were reached through video, album and image type of content and they clicked on the post In terms of clicks on the posts, image and album type of content lead the way, although the average clicks for images is higher than that of an album. Once again, title t=6 shows different results as compared to other titles as it was the only post that was not subjected to paid dissemination. In this case, videos have higher clicks as compared to any other form of content, which indicates high impact of video content in organic promotions. Number of clicks on photos are close to that of videos in this case, as shown in Figure 4.

25.00% Video Album 20.00% Photo

15.00%

10.00%

5.00%

0.00% 1 2 3 4 5 6 7 8 9 Title of the Board Games (t)

Figure 5: Percentage of FB users who were reached through video, album and image type of content and they engaged (liked, commented, shared or clicked) with the post Figure 5 shows the overall engagement by the FB users, which is a combined effect of Figure 3 and Figure 4. In overall engagement as well, images lead the way followed by the album and the video type of content when the posts are subjected to organic and paid dissemination, whereas, for the organic type of content, images and videos have higher rate of engagement.

80.00% Fans Non-Fans Total Engagement 70.00% 60.00% 50.00% 40.00% 30.00% 20.00% 10.00% 0.00% 1 2 3 4 5 6 7 8 9 Title of the Board Games (t) Figure 6: Percentage of fans reached, non-fans reached and engagement

Grzegorz Chodak et al.

Figure 6, shows that a greater number of non-fans can be reached through a combination of organic and paid promotions but it doesn’t guarantee higher rate of engagement. Organic promotions reach out to higher number of fans and result in an average number of interactions as compared with posts with both organic and paid promotions. 6. Calculation of metrics Table 5 shows the calculations of some metrics proposed in the paper based on data from the experiment. Due to the limitation of this paper, no detailed analysis of the results will be presented, but the obtained data can represent exemplification for the assessment of the effectiveness of the marketing campaign carried out in social media. Table 5: Calculated metrics for effectiveness, attractiveness and persuasiveness of the campaign with respect to type of content (video, album and image)

t t t t t t t t t Title Sales E video E Album E Image A Video A Album A Image P Video P Album P Image 1 2 0.057 0.025 0.034 0.0033 0.0136 0.0577 0.0230 0.0673 0.0632 2 4 0.067 0.129 0.121 0.0031 0.0139 0.0159 0.0378 0.0308 0.0438 3 5 0.132 0.058 0.076 0.0034 0.0104 0.0151 0.0260 0.0685 0.0623 4 6 0.214 0.222 0.171 0.0044 0.0146 0.0184 0.0204 0.0358 0.0430 5 0 0.000 0.000 0.000 0.0016 0.0085 0.0101 0.0208 0.0312 0.0638 6 2 0.118 0.250 0.069 0.0084 0.0000 0.0173 0.0474 0.0289 0.0456 7 1 0.059 0.008 0.011 0.0000 0.0140 0.0295 0.0128 0.0789 0.0772 8 0 0.000 0.000 0.000 0.0051 0.0089 0.0231 0.0144 0.0418 0.0656 9 2 0.200 0.036 0.017 0.0022 0.0126 0.0714 0.0074 0.0464 0.0844 Mean 0.0941 0.0809 0.0554 0.0035 0.0107 0.0287 0.0233 0.0477 0.0610 Std 0.0735 0.0912 0.0561 0.0022 0.0044 0.0201 0.0125 0.0190 0.0146 Dev The computed metrics reveal some possible concerns. Although images tend to engage more users (higher Attractiveness), the Effectiveness of video posts is higher (except Title 2). Attractiveness seems to be a possible determinant of sales volume, however, it indicates that a more detailed statistical analysis should be performed, accompanied by the clickstream examination on the E-Store website. The level of Persuasiveness is not a good indicator of sales and seems to be distorted for the albums because of the nature of interactions with that kind of post. Additionally, high standard deviation shows large, probable qualitative differences between the promoted titles. 7. Conclusions and future scope The obtained results clearly show how complicated the analysis of social media communication could be (Baym, 2003). Similar activities could have dissimilar effects and metrics have to be carefully designed to capture the relevant outcomes. The carried-out experiment shows that considerably different effects were obtained depending on the content of posts (picture, video, album) related to the same products. In the case of video, the highest level of reach, while in the case of photographs, the highest level of engagement was achieved. Additionally, the proposed metrics seem to point out that video posts may be generally more effective in terms of business aims (achieving greater sales). All these considerations call for extended research on the effectiveness of marketing communication in social media. Different media types require different approaches in the effectiveness evaluation. For example, eye-tracking research would be especially useful to understand the discrepancies in the way that users interact with text, photo or video based content. Moreover, to understand the purchase-related decisions, surveys and in-depth interviews would be recommended to capture the influence of content quality and valence. The authors are planning to perform further experiments in closely controlled environments in order to combine qualitative and quantitative analysis of the studied phenomenon. Further research could also attempt to determine the psychological profile of users through formal behavioural characteristics which diagnose basic, biologically conditioned dimensions of temperament which is one of the components of personality. In current research, social media environments create and deliver stimulation to users to react to the content. The scales could allow to better identify tendencies to quickly respond to emerging content, including content with different sensory value (i.e. text, visual and audio-visual means, weblinks),

Grzegorz Chodak et al. maintaining a high rate of activity in social media, ability to adequately respond to long-term or strongly appealing stimuli, sensitivity or emotional immunity. Acknowledgements This work was partially supported by the National Science Centre (NCN) in Poland under grant number: 2018/29/B/HS4/02857. References Barger, V. & Labrecque, L. (2013). “An integrated marketing communication perspective on social media metrics”, International Journal of Integrated Marketing Communications, Vol. 5, No. 1, pp 64-76. Baym, N.K. (2013). “Data not seen: The uses and short coming of social media metrices”, First Monday, Available at: https://ojphi.org/ojs/index.php/fm/article/view/4873/3752 [Accessed: 9th January 2019] Bucher, T. (2012). “Want to be on the top? Algorithmic power and the threat of invisibility on Facebook”, New Media & Society, Vol. 14, No. 7, 1164–1180. Chodak, G. & Suchacka, G. (2017). “An Experiment with Facebook as an Advertising Channel for Books and Audiobooks”, Information Systems Architecture and Technology, Vol 521, pp 221-233. de Vries, L., Gensler, S., Leeflang, P. (2012). “Popularity of Brand Posts on Brand Fan Pages: An Investigation of the Effects of Social Media Marketing”, Journal of Interactive Marketing, Vol 26, Issue 2, pp 83-91. Dolan, R., Conduit, J., Fahy, J., Goodman, S. (2017). “Social media: communication strategies, engagement and future research directions”, International Journal of Wine Business Research, Vol. 29 Issue: 1, pp.2-19, Hlavac, P., Simko, M. (2017). “Detecting genuinely read parts of web document, Proceedings of 12th International Workshop on Semantic and Social Media Adaptation and Personalization (SMAP)”, Bratislava / ed. by Maria Bielikova and Marian Simko : IEEE Computational Intelligence Society, Slovak University of Technology in Bratislava, pp 6-11. Kanuri, VK., Yixing, C., Shrihari, S. (2018). “Scheduling Content on Social Media: Theory, Evidence, and Application”, Journal of Marketing, Vol.82, No. 6, pp 89 - 108. Kapoor, K., Tamilmani, K., Rana, P., Patil, P., Dwivedi, Y., Nerur, S. (2018). “Advances in Social Media Research: Past, Present and Future”, Information Systems Frontiers, Vol. 20, Issue 3, pp 531-558. Kemp, s. (2018). “Digital in 2018”, We are social, Available at: https://wearesocial.com/blog/2018/01/global-digital-report- 2018 [Accessed 2nd January 2019] Kotarbinski, T. (2013). “Praxiology: An Introduction to the Sciences of Efficient Action”, Pergamon Press. Lamberton, C., Stephen, T. (2016). “A thematic exploration of Digital Social Media and Mobile Marketing: Research Evolution from 2000 to 2015 and an agenda for future Inquiry”, Journal of Marketing: AMA/MSI Special Issue, Vol. 80, pp 146-172. Lipsman, A., Mudd, G., Rich, M., Bruich, S. (2012). “The power of ‘Like’ – How brands reach (influence) fans through social media marketing”, Journal of Advertising Research, Vol. 2, No. 1, pp 40-52. Liu, L., Lee, M., Liu, R., Chen, J. (2018). “Trust transfer in social media brand communities: The role of consumer engagement”, International Journal of Information Management, Vol. 41, pp 1-13. Mangold, G., Faulds, D. (2009). “Social media: The new hybrid element of the promotion mix”, Business Horizons, Vol. 52, Issue 4, pp 357-365. Myers, WS. (2018). “Censored, suspended, shadow banned: User interpretations of content moderation on social media platforms”, New Media & Society, Vol. 20, No. 11, pp.4366-4383. Perrin, A. (2015). “Social Networking Usage: 2005-2015, Pew Research Center”, Pew Research Centre, Available at: http://www.pewinternet.org/2015/10/08/social-networking-usage-2005-2015/ [Accessed: 7th January 2019] Pszczolowski, T. (1976). “Zasady sprawnego działania. Wstęp do prakseologii”, Wiedza Powszechna, Warszawa. Salo, J. (2017). “Social media research in the industrial marketing field: Review of literature and future research directions”, Industrial Marketing Management, Vol. 66, pp 115-129.

81 Paper 6

G. Chodak, Y. Chawla, A. Dzidowski, K. Ludwikowska Y. Chawla, G. Chodak The Effectiveness of Marketing Commu- nication in Social Media Social Media Mar- keting for Businesses: Organic Promotions of Web-Links on Facebook Social Media Marketing for Businesses: Organic Promotions of Web-Links on Facebook

Yash Chawlaa,, Grzegorz Chodaka

aDepartment of Operations Research and Business Intelligence, Faculty of Computer Science and Management, Wrocław University of Science and Technology, 50-370 Wrocław, Poland

Abstract Eﬀectiveness of social media marketing is a topic of great interest for researchers as well as marketers. To enrich the literature, regarding the eﬀectiveness of various types of posts with a weblink, we design and conduct an experiment on Facebook (FB). This experiment was conducted in a real business environment, through the FB fan page of a Polish e-commerce store. The observations were analysed through simple linear regression and metrics adapted for this particular experiment from the literature. Results show that a web-link placed in the comments of an FB post, instead of the caption, is more lucrative. It is also shown that, based on the campaign aims, the metrics can give valuable information about the time of posting, as well as the interval between posts. Keywords: Social Media Marketing, Facebook Marketing, Organic Promotion, Web links, external links

1. Introduction Exponential growth in the number of internet users has lead the business to explore efficient ways of managing their presence in the electronic space. Businesses have adapted new business models that allow them to utilize the opportunities that the internet has to offer (Wielki, 2010). With a growing number of internet users, social media also gained traction rapidly and a significant increase in social media users is still noticeable across the globe. The number of social media users worldwide is estimated to be 2.82 billion. It is anticipated that, in 2020, the user base will reach 2.96 billion, which would further grow to 3.09 billion in 2021 1. The power of the social media ecosystem is amplified due to these huge numbers, but its high importance is due to the fact that it connects directly or indirectly online as well as offline elements of the economy (Hanna et al., 2011). This is especially the case for Facebook (FB), with more than 2.4 billion active monthly users (https://www.statista.com/statistics/264810/number-of-monthly-active- facebook-users-worldwide/), which has proven to be a very popular platform to market products,

Email addresses: [email protected] (Yash Chawla), [email protected] (Grzegorz Chodak) 1for more details see: https://www.statista.com/statistics/278414/number-of-worldwide-social-network-users/ Preprint submitted to Journal of Business Research This version: March 12, 2020 promote brands, manage relationships and lead discourse with customers (Chodak and Suchacka, 2017; Myers West, 2018). The numerous studies of social media research over the last two decades point out its growing importance for businesses (Kapoor et al., 2018), especially in the aspect of direct marketing in today’s digital economy (Unold, 2003). Therefore, activities on social media frequently become the basis of the marketing strategy of companies. In the present world, where each person can communicate about products, businesses or brands with thousands of peers, the impact of consumer engagement has been greatly magnified (Liu et al., 2018). Consequently, in the current culture, when planning the marketing content, managers are strongly dependent on the users and also the algorithms of social media platforms (Kanuri et al., 2018). However, the studies concerned with social media communication often do not take into account the effectiveness of communication for a variety of content, especially the links between position, levels of interaction and time of posting. There is also a lack of agreement about how to measure the effectiveness and which social media marketing indicators define communications performance in the best way (Lamberton and Stephen, 2016). This has attracted a lot of research in this field (Alalwan et al., 2017; De Vries et al., 2012a), however, significant gaps are still observed. This paper aims to fill some of these gaps. The purpose of this article is to analyze the effectiveness of different types of content in the context of link position in generating user engagement, through organic promotions on Facebook in a real business environment. In particular, we aim to: (i) Check the best position for placing a web-link on a Facebook fan page post, intended to be promoted organically (i.e. without paid promotions); (ii) To analyze various metrics for effectiveness of marketing communication in social media and draw out the engagement patterns among different types of post types, posted at different times of the day. The structure of the article is as follows: After a brief introduction, the manuscript in this Section 1, the literature review regarding social media media marketing, its effectiveness and challenges for conducting an experiment on social media in a real business environment is discussed in Section 2. After that, Section 3, discusses in detail the experiment setup and the methods used for analysing the empirical data recorded during the experiment. In the same section, metrics for analysing the results of a social media campaign and its constituent variables are also described. This was followed by the description of results & observations from the experiment in Section 4 and discussion on conclusions in Section 5. Lastly, the limitations and future scope of research have been discussed in Section 6, followed by the bibliography and appendix with the long tables.

2. Literature Background 2.1. Social Media Marketing and Effectiveness Evolution in technology has enabled social networking platforms to give users the medium to share content in various forms, such as text, visuals, audio-visuals, weblinks, etc. The general impact of different kinds of content on the brand’s page have been studied and shows that there is a variation in engagement levels among different types of content. (De Vries et al., 2012b). There is an unavailability of analysis in the literature on how individual elements of the communication process and communication strategies of its participants are influenced by mutual feedback in communication channels. These are fully bi-directional, highly dependent on the socio-cultural 2 context, variety of media forms, variable roles of participants, as well as a separate way of coding information. There is a need for establishing links between social media content and engagement, drivers of engagement and further conceptualization as well as measurement of Social Media content (Dolan et al., 2017). Customer engagement may be defined as a concept able to capture customers’ total set of behavioral activities toward a brand or company (Coulter et al., 2012). User engagement, measured by likes, comments, shares and clicks, is the basis on which the reach of the content usually is determined. Customer engagement creates enhanced organizational performance including, among others, increased sales, superior competitive advantage and profitability (Kumar et al., 2010; Hollebeek, 2011) and emotional connections (Chan and Li, 2010). Dehghani and Tumer (2015), presented that Facebook advertising significantly affected brand image and brand equity and both factors contributed to a significant change in purchasing intention. Consumers interactions through Facebook and the posted messages supports them on their purchasing decision as well as choosing of products in order to settle the purchasing decision (Di Pietro and Pantano, 2012). Efficiency of Facebook advertising may be analysed using modified methods taken from In- ternet marketing theory like composite efficiency index (CEI) (Vejacka,ˇ 2012), however, such methods don’t take into account specific conditions of social media. Oviedo-Garc´ıa et al. (2014), proposed metrics for customer engagement on Facebook called ratio of effective interest, which take into consideration the likes, comments, shares and other clicks, divided by number of posts in relation to average impressions. Shen et al. (2016), claimed that communication effectiveness, in the form of attitudes toward advertising and message-sharing intention is higher in an interactive advertising format than in a non-interactive format. In social media the higher the content engagement, the higher the reach is obtained (Lipsman et al., 2012). Social Media platforms keep updating and modernizing their algorithms to make the content more and more relevant for the users. Through an analysis of EdgeRank, the algorithm dedicated to structuring the flow of information and communication on Facebook’s ‘News Feed’, researchers have argued that the distribution of content is biased based on its type (Bucher, 2012). To obtain better insights on social media marketing effectiveness, the authors analyzed various metrics. All popular social media platforms provide some measurements that help to understand users’ engagement. Although researchers have raised the problem of needed expertise and adequacy of widely available metrics (Baym, 2013), the engagement is one of the most popular success factors for social media marketing. Based on those findings, authors decided to calculate metrics focusing on the effectiveness and interaction patterns among Facebook users. The effectiveness of communication is usually defined by the ISO 9000: 2005 standard, as the degree of achieving the planned goals. The meaning of effectiveness is based on the praxeology approach - the theory of efficient action (Kotarbinski, 2013). According to this meaning, the efficacy describes each component of good work” as constituting: effectiveness, favorableness and economy. The primary form of efficient action is effectiveness, described as a compatibility of action with the intended aim (Pszczołowski, 1967). In this meaning, the action can be effective when all performed activities enable reaching established goals. This can be measured as the level of reaching goals (what is not always easy to evaluate) or the degree of approach to reach them (defined as purposefulness). When the goal is reached partially, the action is also partially effective and, when the 3 goal is not achieved, the action is not efficacious, therefore, the effectiveness can be characterized by a different intensity (Chodak et al., 2019). Taking into consideration only the users presence on social media pages is not a sufficient indicator for marketing communications. Its marketing communication effectiveness should rather relate to what the users pay attention to. Assessing the effectiveness of marketing messages is crucial in many areas of communication research, from campaign design to theory testing. Detailed evaluation is very expensive in terms of time, financial as well as human resources but, at the same time, it is necessary to avoid future expenditures of campaign funds on weak messages (Kim and Cappella, 2019). The effectiveness measure is also connected with the area of social media usage. For example, Nawaz et al. (2017) analysed the effectiveness in healthcare, and Beshears (2017), in police. Kusumasondjaja (2018) analysed the roles of message appeals and orientation on social media brand communication effectiveness. Baruah (2012) pointed out the important feature of social media communication, namely instant messaging has created another method of interaction, firstly, the length of messages is shorter and the style of the interaction is more conversational. Peters et al. 2013 distinguish four dimensions of social media analysis: motives, content, network structure, and social roles and interactions. In that perspective, content has three distinct aspects: quality (e.g. interactivity, vividness, education, entertainment, information), valence (e.g. emotions, tonality, rating variance), and volume (counts and volumes). Since the purpose of the presented experiment was to measure the difference between the interaction patterns of FB users, the proposed metrics should focus on that area. It is also worth to mention that, according to Kim et al. (2019), effectiveness is the third most popular area in digital marketing communication research.

2.2. Experimenting on Social Media in Real Business Environment The experiment, whose results are described in this article, was conducted in a real business environment. Apart from the interesting results, there is also another scientific problem to be discussed - how to plan and carry out the experiments in a real business social media environment. The theory about experiment design is very wide, e.g. (Montgomery, 2017; Emery and Nenarokomov, 1998). There is also literature available about experiment design in real business (Gray, 2019), but there is a research gap in designing experiments in a real social media business environment, like on a Facebook fan page of a real business. The first problem is well known in the literature that experiments can’t be repeated many times, therefore, scientists should conclude from one experiment. Repeating an experiment many times in the same environment is impossible as the social media services change very dynamically. Considering a Facebook fan page – the number of followers may increase/decrease, the algorithms of post positioning may change and the competitors activities may influence the posts order. It is also hard to design experiments with a control group on the Facebook (as well as Twitter and Instagram) environment, as it is impossible to plan in which configuration and when the users from the control group will see the posts. Another important topic is that the sample, which takes part in the experiment, is not a random sample. Firstly, each business fan page has followers which are interested in specific products of the company. Secondly, the process of gathering the followers usually requires targeted marketing with a selection of users of certain age, sex and interests. Thirdly, the network effect causes the most active Facebook users with a higher number 4 of connections with other users to influence the experiment results more than those with a few connections. Another point is that Facebook users can not be aware of taking part in the experiment, other- wise, the reactions of users may be different. There is also a problem of the number of posts (or general marketing messages) which are created in the experiment. Too many posts may discour- age the users, but too few may cause the results from the experiment to not be worthy from the scientific point of view – for example, the statistical analysis is impossible. There are also practical problems which we met in this experiments. The products, which were advertised, should be on constant sale. If there is a longer time between designing, starting and finishing, the experiment for some products may become not available on the market, which distorts the results. The presented content for different types of posts should have the same form – the experiment designers must be aware that the ceteris paribus condition is met. This condition is important if we want to measure the influence of one factor on another without influence of other factors like quality of product photos, category of products, etc. Therefore, we suggest that the analysed group of products should be homogeneous. There is also the problem of getting access to an administration panel and databases to analyse the results of experiments. If databases include sensitive data, like customers addresses, the access to the database panel is restricted only to authorized users. This problem can be solved by signing an agreement with the company concerning sensitive data confidentiality. The second solution is an export by the company, excluding sensitive data, from the experiment which is valuable for the researchers. There is also the problem of the duration of the experiment. In a social media environment, where reactions from the audience and shares of posts influence the overall marketing effect, there is a need to wait with finishing the experiment. On the other hand, too long of a duration for the experiment may cause that additional external factors will blur the results. From our experience, a few days (less than 5) is too short a period to analyse Facebook reactions, therefore, we propose the optimum social media experiment duration to be one week to a maximum of two weeks. After that, time results should be gathered and analysed. Gathering the results is also a crucial point. It is advisable to collect data automatically using dedicated software, however, in some cases, data can be collected manually.

3. Experiment Design, Method and Metrics 3.1. Experiment Structure and Design The experiment was designed and conducted in collaboration with a Polish e-commerce store, that sells various board games, books, films and other products. The name of the company is not given due to the non-disclosure agreement and is simply called “E-Store” hereafter. Facebook (FB) page is one of the two advertising channels, the other being Google Ads, used by the E- Store for reaching out to customers. E-Store uses different types of posts and content on their FB page to inform customers about various products and novelties available at their store. FB is a two-way medium of communication, hence, it proves beneficial for the E-Store to obtain direct feedback from actual and prospective customers (?). This feedback is vital to understand the needs and opinions of the customers about the range of products available on the E-Store. The 5 E-Store also uses the FB fan page to disseminate special discount coupons, spread information about promotions and organize competitions with prizes for customers, which has been found to be effective (Radzi et al., 2018). The structure of the experiment was adapted from a previous study in the literature, that investigated the effectiveness of marketing communications for various types of promotional content such as graphics, videos, photo album and text (Chodak et al., 2019). We used the E-Store’s FB business page, to run the experiment, which enabled us to gather real-time empirical data from the engagement and actions of actual customers. During the course of the experiment, no other social media channels were active, apart from the E-Store’s FB page on which the experiment was conducted. The owner of the store randomly selected 24 books (P1-P24), which were similar in type and market ratings, out of numerous books available on from the E-Store. For each book, we created four posts (A1-A4) consisting of the weblink to the book’s details and buying option on the E-Store’s website with a short one line caption. The position of the link was changed in each type of post, as shown in Table 1.

Table 1: Four types of posts, based on position of the link, used for each book in the experiment

Type Structure Description A1 Link then Text In the caption, the link was placed ﬁrst and then the text A2 Text then Link In the caption, the link was placed after the text A3 Text Link Text In the caption, the link was sandwiched between the text. A4 Text, Link in In the caption, only text was placed with the preview im- Comment age. The link was placed in the post as a comment.

96 posts, in total, were published over 4 days at an interval of one hour between each post; as shown in Table A.5. The posts were scheduled to be published using the FB’s publishing tool, which enabled us to control the exact time of posting. The time for publishing the posts for each book title was the same each day and only the type of post changed. For example: for the book title P1, the time of posting was 00:00, and the post types were A1, A2, A3 & A4 on day 1, day 2, day 3 & day 4, respectively. Each post underwent organic promotion, without any interference from the authors, for 10 days. There were no posts on the E-Store’s FB, since 24 hours before the beginning of the experiment and also after the posting ended on day 4; all the observations were recorded on day 14. This was done to ensure that no other campaign had inﬂuence on the reactions, clicks or sales of the 24 book titles included in the experiment. Observations were recorded exactly 240 hours after each post was published, as shown by ”O.Post” in Table A.5. The design of the experiment was such that there were only two explanatory variables, type of post and time of the post, which would result in a diﬀerence in performance of the post.

3.2. Methods To analyse the effect of these two variables, we used the metrics adapted from Chodak et al. (2019), which were defined based on the typical measurements and the dimension of social media 6 content analysis. These metrics focus mainly on the volume, with less focus on the quality of content. For this particular experiment, this doesn’t create a bias, as all the products (in this case books) were of similar type and were best sellers in the market. Focus on volume also omits the valence dimension, as positive endorsement instead of rating is more of a basis for propagation of content on FB. Other negative reactions such as, hide post, hide all posts, report as spam and unlike page, are possible but we did not observe any such reactions in our experiment. Another important reason to use these metrics is because these volume related measures, although easiest to gather and analyse, can be treated as indicators of quality. The following variables were defined for the experiment in order to adapt the metrics and show the most effective way of placing weblinks on FB posts for organic promotion. For each advertised book title t (t = P1, P2....P24) and post type { } z (z = A1, A2, A3, A4 ) the following variables were defined: { }

Rt(z) – the number of FB users reached by a particular post type ”z”, of a particular book • title ”t”, respectively, in the experiment.

Rt - the total number of FB users reached (through all four post types), by a particular book • title ”t”, in the experiment.

Lt(z) – the number of reactions (in our case only Likes, Comments and Shares) by FB users • on a particular post type ”z”, of a particular book title ”t”, respectively, in the experiment.

Lt - the total number of reactions (on all four post types) by FB users, on a particular book • title ”t”, in the experiment.

Ct(z) – the number of clicks by FB users that were recorded on a particular type ”z”, of a • particular book title ”t”, respectively, in the experiment.

Ct - the total clicks (on all four post types) by FB users, on a particular book title ”t”, in the • experiment.

Tpost – the particular time of the day, when the posts of a particular book title ”t”, was • published.

St(z) - total sales of a particular book title ”t”, through a particular post type ”z”, in the • experiment.

St - total sales of a particular book title ”t” (through all four post types) in the experiment. • The following are the metrics that have been analyzed in this study. The proposed metrics are mutually comparable and analyzed further. For instance, it could be a case that a campaign is attractive but not persuasive or neither persuasive nor eﬀective; a campaign could be less intense but more eﬀective; content type could be persuasive but not attractive and so on.

Metrics related to goal attainment that focus on conversion rate: •

7 –E ﬀectiveness of the campaign (Et), related to the conversion rate which is compared between the book titles advertised.

S t Et = (1) Ct

–E ﬀectiveness of the post type (Et(z)), z = A1, A2, A3, A4 , mutually compared. { }

S t(z) Et(z) = (2) Ct(z)

Metrics related to content valence, which describes a post’s attractiveness and the users’ • willingness to engage (like, comment or share):

– Attractiveness of the campaign (At), compared between campaigns.

Lt At = (3) Rt

– Attractiveness of the content type (At(z)), z = A1, A2, A3, A4 , mutually compared. { }

Lt(z) At(z) = (4) Rt(z)

Metrics related to quality of the content, which illustrate the power of the message conveyed • through the content to get the desired action (link click):

– Campaign’s persuasiveness (Pt), compared between campaigns.

Ct Pt = (5) Rt

– Persuasiveness of the content type (Pt(z)), z = A1, A2, A3, A4 , mutually compared. { }

Ct(z) Pt(z) = (6) Rt(z)

Metric related to the volume of content concerned with its publication dynamics: •

– Intensity of the campaign (It), compared between time frames or between campaigns. ∆R I = t (7) t ∆T

– Intensity of the type of post (It(z)), z = A1, A2, A3, A4 , compared between time { } frames or between campaigns. ∆R (z) I (z) = t (8) t ∆T(z) 8 A – Intensity of campaign’s attractiveness (It ), compared between time frames or between campaigns. ∆L IA = t (9) t ∆T – Intensity of type of post’s attractiveness (IA(z)), z = A1, A2, A3, A4 , compared t { } between time frames or between campaigns. ∆L (z) IA(z) = t (10) t ∆T(z)

P – Intensity of campaign’s persuasiveness (It ), compared between time frames or between campaigns. ∆C IP = t (11) t ∆T – Intensity of type of post’s persuasiveness (IP(z)), z = A1, A2, A3, A4 , compared t { } between time frames or between campaigns. ∆C (z) IP(z) = t (12) t ∆T(z)

4. Experiment Results and Observations 4.1. Observed data and initial data analysis Recorded data before initiating the experiment shows that the E-Store’s FB page had 5032 fans which increased to 5045, as per the post experiment data records. Cumulatively, the E-Store sold 101 books during the experiment. The day-wise sales record of each book is presented in Table 2. The book P12, ran out of stock and was unavailable after its ﬁnal unit sale on the 3rd day. The stock of each book was veriﬁed before initiating the experiment but, for P12, there was a bulk purchase of 11 units on Day 1, which caused this unavailability.

Throughout the campaign, in this experiment, the E-Store reached out to 17,546 FB users,which resulted in 664 reactions and 258 link clicks. The largest overall reach (1,284 FB users) and link clicks (26) were found to be for the book title (P21), which had it’s posts at 8 pm in the evening. The maximum reactions, 97, were observed on the post at 7 am in the morning, which was of book title (P8). The Figure 1, shows the overall reach, reactions and clicks on the posts of each book title at the end of the experiment. In terms of the type of posts, A4 contributed to an overall 44.48% reach, 48.34% reactions and 62.02% clicks in the campaign. This showed that it was the most dominant type of post out of the four used in this experiment. To further see the eﬀect of including the A4 type of post in the campaign, we calculated the mean and standard deviation (SD) of reach, reactions and clicks for each book title. For each title, two means and SD were calculated. First, which included all the four types of posts (A1-A4) for each book title and second, for only three types (A1-A3), excluding the dominant A4 type of posts. The calculations in Table A.6 show that, for each individual book 9 Table 2: Day-wise sales recorded for each book during the experiment

Book/Day 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Units Sold Et P1 1 2 3 0.6000 P2 0 0.0000 P3 1 1 2 2 2.000 P4 2 1 1 1 3 5 0.5000 P5 1 1 1 1 1 5 0.6250 P6 1 2 1 4 8 8.0000 P7 1 1 1 3 7 2 1 15 1.1429 P8 1 1 1 1 2 0.1905 P9 1 1 0.1111 P10 0 0.0000 P11 1 6 1 1 1 1 8 1.2222 + P12∗ 11 1 Out of Stock 12 N.A. P13 0 0.0000 P14 0 0.0000 P15 2 2 0.6667 P16 1 2 3 0.1765 P17 1 1 0.0526 P18 0 0.0000 P19 3 1 4 2.0000 P20 1 4 5 0.7143 P21 0 0.0000 P22 1 1 0.0833 P23 1 1 2 0.2000 P24 3 1 2 1 1 1 8 0.7857 Total number of books sold during the campaign 101

∗This title was out of stock in the E-store, after the sale of its final unit on Day 3. + Et(P12) is not calculated as the product was not available in store for the whole duration of the experiment . title, the inclusion of post type A4 increased the mean of reach, reactions and clicks. There were only four cases found where, the mean of the reactions, was slightly higher when post A4 was not included. This was for the means of reactions of P1, P13, P20 and P22, where the difference between the means was 0.08, 0.25, 0.08 and 0.42, respectively. There was no direct explanation of these two exceptions, as both the means of reach and number of clicks were higher for these titles when we included the A4 type of posts. According to us, it might have something to do with the time duration between the time of posting and the first reaction on these posts, but we could not ascertain it for sure. From the data in Table A.6, we observed that more reach did not necessarily mean more reactions or clicks. Hence, we carried out simple linear regression, with the dependent variable as the overall reach (Rt) of each book title (where all four posts were included) and the regressors as the total reactions (Lt) and clicks (Ct). We found that reactions (Lt) had a statistically significant and positive relationship with reach (Rt), whereas the relationship of clicks was statistically insignificant. We repeated the regression by excluding the post type A4 and obtained similar results. Results of both the regressions are shown in Table 3. Regression analysis, in both cases, showed that more reach of a post resulted in more interactions but did not

10 Figure 1: Combined reach, reactions and clicks of posts for each book, during the lifetime of the experiment Note: The value of ”Reach” is scaled to 1/10th, in this graph, so as to improve visibility of the reach and reactions bars. necessarily transform into more link clicks.

Table 3: Results of simple linear regression with dependent variable as Rt and regressors as Lt and Ct (all four types of posts considered)

Coeﬃcient Std. Error t-ratio p-value

For Dependent variable Rt and Regressors as Lt &Ct (all four types of posts considered) const 470.597 43.4852 10.82 0.0000

Lt 6.47050 1.63771 3.951 0.0007

Ct 7.57849 6.22431 1.218 0.2369

For Dependent variable Rt and Regressors as Lt &Ct (A4 type posts excluded) const 260.407 22.3275 11.66 0.0000

Lt(withoutA4) 5.07253 2.31923 2.187 0.0402

Ct(withoutA4) 17.8711 9.12456 1.959 0.0636

We also performed linear regressions with the dependent variables as reach of each type of post (Rt(A1)... Rt(A4)) and their corresponding reactions and link clicks. The results of these regressions are shown in Table 4. Individually, for post type A1, A2 & A3, total reach was positively 11 and significantly related with both their corresponding reactions and link clicks. This is interesting because, in terms of their collective sum, there was no significant relation of the reach with link clicks. For post type A4, only a significant relation between reach and reactions was found.

Table 4: Results of simple linear regression with dependent variables as Rt(A1), Rt(A2), Rt(A3) & Rt(A4), and regressors as their corresponding Lt(An) and Ct(An)

Coeﬃcient Std. Error t-ratio p-value

For Dependent variable Rt(A1) and Regressors as Lt(A1) & Ct(A1) const 89.5194 10.1230 8.843 0.0000

Lt(A1) 7.22855 1.51810 4.762 0.0001

Ct(A1) 16.8512 5.03877 3.344 0.0031

For Dependent variable Rt(A2) and Regressors as Lt(A2) & Ct(A2) const 85.9984 10.0510 8.556 0.0000

Lt(A2) 5.27508 2.11231 2.497 0.0209

Ct(A2) 17.0753 6.67204 2.559 0.0183

For Dependent variable Rt(A3) and Regressors as Lt(A3) & Ct(A3) const 70.7781 9.38505 7.542 0.0000

Lt(A3) 5.63590 2.10307 2.680 0.0140

Ct(A3) 19.0862 8.12237 2.350 0.0286

For Dependent variable Rt(A4) and Regressors as Lt(A4) & Ct(A4) const 208.237 24.5274 8.490 0.0000

Lt(A4) 5.86555 1.42167 4.126 0.0005

Ct(A4) 5.77798 4.22092 1.369 0.1855

4.2. Performance of the campaign, based on metrics For all the book tiles and post types, various metrics are deﬁned in sub-section 3.2. The detailed results obtained for each metric are illustrated in the following subsections.

4.2.1. Effectiveness The effectiveness of the campaign was calculated based on equation 1, where the sale of each particular book title ”t” (St) was divided by its respective number of clicks (Ct) and is shown in Table 2, along with the sales of each book title. Due to the small number of sales for individual post types, we did not have any conclusive evidence regarding the effectiveness of the post types calculated based on equation 2, hence, their details are not included in the article. The effectiveness of the campaign (Et) gives us two main results. First, it shows that, even though the collective reach 12 (of all four book titles) of a certain book title was less as compared to some other posts, it can have higher Et. For example, let us consider the sales of three book titles, P7, P11 and P19. They had sales of, 15, 8 and 4, respectively, whereas their Et were 1.1429, 1.2222 and 2.000, respectively. From this, we can deduce the second factor, which was the time of posting. As the book titles were all from a similar category, all being best sellers (same level of quality based on market opinion) and all having the four types of post, the time of posting can be one of the measurable factors that differs between them. Hence, Et can also be used to determine the best time of posting in case the objective of the campaign is to be the most effective. In the current experiment, the best time of posting in descending order is, 5 am, 2 am or 6 pm, 10 am and so on. These times also show another important detail when compared with insights provided by Facebook to the administrators of a Facebook page. The number of users online at a certain time, shown under the ”Posts” section on ”Insights”, did not seem to be significant for the effectiveness of the campaign. According to the insights for the E-Store, the maximum users were online at 8 pm and 9 pm, which were not the recommended times of posting based on the calculation in this study.

4.2.2. Attractiveness The attractiveness of the campaign of each book title was calculated based on equation 3, and the attractiveness of each post for each book title was calculated based on equation 4. The obtained results have been plotted on the graph in Figure 2. Attractiveness of the campaign was the highest for book title P8, corresponding to 7 am as the time of posting. Followed by (in decreasing order of campaign attractiveness) P12 (11 am), P16 (3 pm) and P21 (8 pm). On average, the post type A4 was the most attractive (average At(A4) = 0.0348), with the rest of the three types with similar attractiveness scores. Averages of At(A1), At(A2) & At(A3), were 0.0253, 0.0277 & 0.0281, respectively. Two of the four cases (P13 and P22), where the means of reactions was higher when post type A4 was excluded, can be related to this remotely. Both these book titles had their posts after the highly attractive P12 and P21 but this eﬀect can also be by chance, hence, we emphasize that we did not have any clear evidence of these exceptions. Individually, the A4 type of post was most attractive for 9 book titles, mostly between the posting times of 3 am to 5 pm. The rest of the three posts were most attractive for 5 book titles each and scattered during the day. We also observe a steep increase and decrease in the attractiveness of all the campaign as well as the post types, in a very similar pattern. From this, we can conclude that the posting made between 12 am to 2 am, around 5 am, 10 am, 5 pm to 7 pm, would be less attractive to the E-Store’s audience. While, the peaks in the graph, in Figure 2, show the times where posts can be highly attractive to their audience and more reactions can be obtained.

4.2.3. Persuasiveness Persuasiveness of the campaign and of each post type, for each book title, were calculated based on equations 5 and 6, respectively. The results of the calculations are presented in the Figure 3. P04, P17, P12, and P21 had the most persuasiveness of the campaign (in decreasing order), while P06 had the least. The pattern of peaks observed for attractiveness was also observed here. The A4 type of post was found to be the most persuasive for 15 out of the 24 book titles in the experiment. Among the average of the persuasiveness of all types of posts, A4’s persuasiveness was more than double of A2, A3 or A4, individually. This shows that users were more likely to 13 Figure 2: Attractiveness of the campaign and each post type, for respective book titles.

Note: At = Attractiveness of the campaign of respective book title (At), At(A1)..At(A4) = Attractiveness of respective post type for the respective book titles (At(A1)..At(A4)) click on the links with the A4 type of post. The average persuasiveness of A3 was found to be lower than A1, even though A3 was found to the most attractive for 6 book titles, whereas A1 was found for 2. The A2 type of posts were found to be the least persuasive.

Figure 3: Persuasiveness of the campaign and each post type, for respective book titles.

Note: Pt = Persuasiveness of the campaign of respective book title (Pt), Pt(A1)..Pt(A4) = Attractiveness of respective post type for the respective book titles (Pt(A1)..Pt(A4))

14 4.2.4. Intensity The intensity of the campaign, for each book title, was calculated using the equation 1 and the intensity for each type of post was calculated using equation 8. It is very important to ensure that campaigns are not highly intensive because it can lead to flooding the news feed of the user and may lead to negative reactions. Results presented in the figure 4 show that, overall intensity of the campaign, ranged between 1.2468 for P1 and 4.11 for P21. For intensity as well, we saw a similar peak of patterns in persuasiveness and attractiveness, but here, for the intensity of the campaign, it was observed as an upward trend. This is because there were more users online on Facebook during the later part of the day. For different types of posts, it can be seen that the intensity of post type A4 was the highest as compared to the other three types posts apart from two instances, around 6 am and 12 noon. At these two times, the A1 type of post had more intensity. Know the intensity of the campaign and a certain post time can be fruitful in determining the time interval between posting. This would allow the performance of the posted content to achieve its maximum possible efficiency.

Figure 4: Intensity of the campaign and each post type, for respective book titles.

Note: It = Intensity of the campaign of respective book title (It), It(A1)..It(A4) = Intensity of respective post type for the respective book titles (It(A1)..It(A4))

For the intensity of the campaign and post types, two additional metrics, intensity of attractiveness and intensity of persuasiveness, were also calculated using the equations 9, 10, 11 and 12, respectively. The results for intensity of attractiveness for the campaign and for each type of posts of each book title are presented in Figure 5. Here again, we observed a similar peak pattern and that the post type A4 was the most dominant. Higher intensity of attractiveness, for the campaign, was observed for the posts made at 7 am, 11 am, 1 pm, 3 pm and 8 pm. These times also corresponded to the higher intensity of attractiveness of post type A4. The results for intensity of persuasiveness of the campaign and the post types, for each book title, is shown in figure 6. The results for intensity of persuasiveness is quite similar to that of 15 Figure 5: Intensity of attractiveness of the campaign and each post type, for respective book titles. A Note: ItA = Intensity of Attractiveness of the campaign of respective book title (It ), It(A1)..It(A4) = Intensity of A A Attractiveness of respective post type for the respective book titles (It (A1)..It (A4)) the intensity of attractiveness of the campaign. The post type A4 had the highest intensity of persuasiveness, while the rest of the post types were more or less similar. Comparing the figures 4, 5 and 6, it is evident that the intensity of attractiveness and intensity of persuasiveness, for both the campaign, as well as the type of posts, were directly affected by the intensity of the campaign and of the post types, respectively. The desired combination had a lesser intensity but higher intensity of attractiveness and persuasiveness. Through these results, for different times of the day, different types of posts can be selected or the campaign can be designed to achieve the desired combination.

5. Conclusion and Discussion The fabric of social media marketing is very dynamic and keeps marketing managers on their toes to help the business progress in competitive markets. There are numerous challenges, which scientists have to face, in order to design an efficient experiment for determining the effectiveness of marketing communications on social media in a real business environment. The experiment dealt with four specific type of posts, which dealt with different possible placements of links for organic promotion on a company’s FB page. We found that the most lucrative post type was A4, where the link was posted as a comment on the post and the preview of the book title, along with a short textual caption, formed the post. This accounted for maximum reach, reactions, as well as clicks on the post. Our analysis showed that while, for other post types (A1, A2 A3) the number of clicks depended upon the reach of the posts while, for post type A4, it did not. This made the link clicks though the A4 types of posts unpredictable but, for all cases, A4 accounted for the most link clicks than any other type of post. Hence, it would be the most effective option for organic promotions of a web link on FB. According to us, this could be because of one or both of the following reasons. First, the FB algorithm rates the organic appearance of posts with web 16 Figure 6: Intensity of Persuasiveness of the campaign and each post type, for respective book titles. P Note: ItP = Intensity of Persuasiveness of the campaign of respective book title (It ), ItP(A1)..ItP(A4) = P P Attractiveness of respective post type for the respective book titles (It (A1)..It (A4)) links (A1, A2 & A3) lower than the posts which do not have a web link (A4, as it’s link is in the comments). Secondly, the link posted in a comment, along with the post, could be treated as an immediate reaction by the FB algorithm, resulting in a higher ranking or probability of the post to be displayed to other users. The first reason makes sense from FB’s point of view also, as any user or business trying to promote web links on FB do so for drawing customers from FB to their website, which generates revenues for them. Hence, the links are more valuable to marketers, so lower organic dissemination means that marketers would invest in paid promotions of the link. The metrics defined and analysed for the different post types and book titles, give various pieces of information. These metrics would be useful for scientists, as well as managers, to calculate the effectiveness of a campaign. Firstly, based on the aim of a campaign, the time of posting can be decided through the 24 hour graphs (Figure 2 - 6). We emphasize here that these results could and would vary for each business page, based on their geographic location, location of their audiences, exact time of posting, the text in caption, segment of the market and so on. Hence, we recommend that, for getting personalized times and results, a similar experiment can be run and the observations can be compared with the ones in this study. Secondly, the peak and valley pattern of the graphs for various metrics is similar. Observing the time interval between the peak points in the graphs, gives information of the minimum duration that should be maintained the between posts. The schedule of posting, for a particular experiment or campaign, can be synchronized to the peaks in the graphs, which should result in a more uniform horizontal lined graph. A perfectly horizontal line would mean that each post is performing to its utmost efficiency and the interval between the post is good. Another interesting result that we obtained by the impact of the number of users online, shown in FB insights, did not have any significant impact on the results of the metrics. There was an effect in the intensity metric, which depends on the number of users online

17 but, for the rest, there was no uniform eﬀect observed.

6. Limitations and Future Scope of Research The experiment designed in this study provides a basis for the theoretical, as well as practical research experiments, which overcomes a number of challenges discussed in the literature. It opens new doors for research on the effectiveness of marketing communication on social media and also real market experiments, giving valuable insights to the marketers. In the present state, the study has some limitations. The experiment was designed and conducted for a specific online store with specific products. Due to numerous factors that would define the outcome of the results on social media, this experiment can be run in different geographical locations, for different market segments, on different social media platforms, and so on. Although, we expect the results, in a certain market segment in a country / region, to not differ too much, it would be interesting to replicate this experiment for another E-Store with similar products, audience and segment.

Appendix A.

18 Table A.5: Schedule of publishing the posts and recording the observations

Day 1 22 November 2019 Time 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 P. Code P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 P16 P17 P18 P19 P20 P21 P22 P23 P24 T. Post A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 Day 2 23 November 2019 Time 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 P. Code P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 P16 P17 P18 P19 P20 P21 P22 P23 P24 T. Post A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 Day 3 24 November 2019 Time 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 P. Code P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 P16 P17 P18 P19 P20 P21 P22 P23 P24 T. Post A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 Day 4 25 November 2019 Time 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 P. Code P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 P16 P17 P18 P19 P20 P21 P22 P23 P24 19 T. Post A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 Posts are allowed to organically spread from 26 November 2019 to 1 December 2019 (Day 5 to Day 10) Day 11 2 December 2019 Time 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 P. Code P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 P16 P17 P18 P19 P20 P21 P22 P23 P24 O. Post A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 Day 12 3 December 2019 Time 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 P. Code P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 P16 P17 P18 P19 P20 P21 P22 P23 P24 O. Post A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 Day 13 4 December 2019 Time 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 P. Code P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 P16 P17 P18 P19 P20 P21 P22 P23 P24 O. Post A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 Day 14 5 December 2019 Time 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 P. Code P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 P16 P17 P18 P19 P20 P21 P22 P23 P24 O. Post A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 A4 A1 A2 A3 Note: P. Code = Product Code; T. Post = Type of Post; O. Post = Observations recorded for the type of post. Table A.6: Reach, reaction and clicks, along with their mean and standard deviation (SD) among the four types of posts, for each book title

Reach Reactions Clicks 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 Title (Rt) (Rt) Mean Mean SD SD (Lt) (Lt) Mean Mean SD SD (Ct) (Ct) Mean Mean SD SD P1 389 213 97.25 71.00 53.58 13.11 5 4 1.25 1.33 0.50 0.58 5 0 1.25 0.00 2.50 0.00 P2 568 260 142.00 86.67 112.92 27.47 6 2 1.50 0.67 1.91 1.15 6 0 1.50 0.00 3.00 0.00 P3 475 271 118.75 90.33 60.08 23.86 8 3 2.00 1.00 2.45 1.73 2 1 0.50 0.33 0.58 0.58 P4 554 227 138.50 75.67 127.63 27.30 15 4 3.75 1.33 4.86 0.58 16 2 4.00 0.67 6.73 1.15 P5 579 252 144.75 84.00 124.62 33.96 20 11 5.00 3.67 3.65 3.06 8 4 2.00 1.33 2.31 2.31 P6 407 219 101.75 73.00 60.32 22.34 4 1 1.00 0.33 1.41 0.58 1 0 0.25 0.00 0.50 0.00 P7 988 657 247.00 219.00 138.85 155.62 41 29 10.25 9.67 9.03 10.97 14 8 3.50 2.67 2.65 2.52 P8 1050 619 262.50 206.33 144.35 111.02 97 50 24.25 16.67 16.96 9.29 21 13 5.25 4.33 2.50 2.08 P9 667 297 166.75 99.00 139.29 39.51 12 5 3.00 1.67 2.83 1.15 9 1 2.25 0.33 3.86 0.58 P10 496 241 124.00 80.33 93.03 39.27 4 1 1.00 0.33 1.41 0.58 5 0 1.25 0.00 2.50 0.00 P11 651 405 162.75 135.00 56.27 11.36 28 17 7.00 5.67 4.24 4.04 9 5 2.25 1.67 1.26 0.58

20 P12 1070 699 267.50 233.00 92.53 75.50 69 45 17.25 15.00 7.27 7.00 23 9 5.75 3.00 5.56 1.00 P13 643 409 160.75 136.33 119.91 134.13 11 9 2.75 3.00 2.87 3.46 7 5 1.75 1.67 2.36 2.89 P14 1194 551 298.50 183.67 241.91 93.04 62 32 15.50 10.67 12.18 9.07 14 6 3.50 2.00 3.11 1.00 P15 710 412 177.50 137.33 88.13 44.38 21 9 5.25 3.00 4.99 2.65 3 2 0.75 0.67 0.96 1.15 P16 871 426 217.75 142.00 154.56 37.47 55 18 13.75 6.00 15.56 1.73 17 3 4.25 1.00 6.55 1.00 P17 770 393 192.50 131.00 133.40 63.24 33 11 8.25 3.67 9.60 3.51 19 7 4.75 2.33 5.50 3.21 P18 708 374 177.00 124.67 105.21 13.05 10 2 2.50 0.67 3.70 0.58 8 0 2.00 0.00 4.00 0.00 P19 595 372 148.75 124.00 58.61 38.43 6 4 1.50 1.33 1.00 1.15 2 1 0.50 0.33 0.58 0.58 P20 542 354 135.50 118.00 43.93 32.51 5 4 1.25 1.33 0.96 1.15 7 3 1.75 1.00 2.06 1.73 P21 1284 687 321.00 229.00 212.44 130.03 76 33 19.00 11.00 19.75 14.18 26 10 6.50 3.33 6.81 3.06 P22 816 473 204.00 157.67 105.96 62.94 33 26 8.25 8.67 4.11 4.93 12 7 3.00 2.33 2.94 3.21 P23 739 497 184.75 165.67 85.05 93.09 10 6 2.50 2.00 3.00 3.46 10 6 2.50 2.00 1.91 2.00 P24 780 433 195.00 144.33 116.69 70.87 33 17 8.25 5.67 6.13 4.04 14 5 3.50 1.67 3.79 1.15 1For all four types of posts (P1-P4), 2For only three types of posts (P1-P3) References Alalwan, A.A., Rana, N.P., Dwivedi, Y.K., Algharabat, R., 2017. Social media in marketing: A review and analysis of the existing literature. Telematics and Informatics 34, 1177–1190. Baruah, T.D., 2012. Effectiveness of social media as a tool of communication and its potential for technology enabled connections: A micro-level study. International Journal of Scientific and Research Publications 2, 1–10. Baym, N.K., 2013. Data not seen: The uses and shortcomings of social media metrics. First Monday 18. Beshears, M.L., 2017. Effectiveness of police social media use. American Journal of Criminal Justice 42, 489–501. Bucher, T., 2012. Want to be on the top? algorithmic power and the threat of invisibility on facebook. New media & society 14, 1164–1180. Chan, K.W., Li, S.Y., 2010. Understanding consumer-to-consumer interactions in virtual communities: The salience of reciprocity. Journal of Business Research 63, 1033–1040. Chodak, G., Chawla, Y., Dzidowski, A., Ludwikowska, K., 2019. The effectiveness of marketing communication in social media, in: ECSM 2019 6th European Conference on Social Media, Academic Conferences and publishing limited. pp. 73–81. Chodak, G., Suchacka, G., 2017. An experiment with facebook as an advertising channel for books and audiobooks, in: Information Systems Architecture and Technology: Proceedings of 37th International Conference on Informa- tion Systems Architecture and Technology–ISAT 2016–Part I, Springer. pp. 221–233. Coulter, K.S., Gummerus, J., Liljander, V., Weman, E., Pihlstrom,¨ M., 2012. Customer engagement in a facebook brand community. Management Research Review . De Vries, L., Gensler, S., Leeflang, P.S., 2012a. Popularity of brand posts on brand fan pages: An investigation of the effects of social media marketing. Journal of interactive marketing 26, 83–91. De Vries, L., Gensler, S., Leeflang, P.S., 2012b. Popularity of brand posts on brand fan pages: An investigation of the effects of social media marketing. Journal of interactive marketing 26, 83–91. Dehghani, M., Tumer, M., 2015. A research on effectiveness of facebook advertising on enhancing purchase intention of consumers. Computers in Human Behavior 49, 597–600. Di Pietro, L., Pantano, E., 2012. An empirical investigation of social network influence on consumer purchasing decision: The case of facebook. Journal of Direct, Data and Digital Marketing Practice 14, 18–29. Dolan, R., Conduit, J., Fahy, J., Goodman, S., 2017. Social media: communication strategies, engagement and future research directions. International Journal of Wine Business Research . Emery, A.F., Nenarokomov, A.V., 1998. Optimal experiment design. Measurement Science and Technology 9, 864. Gray, D.E., 2019. Doing research in the business world. Sage Publications Limited. Hanna, R., Rohm, A., Crittenden, V.L., 2011. We’re all connected: The power of the social media ecosystem. Business horizons 54, 265–273. Hollebeek, L.D., 2011. Demystifying customer brand engagement: Exploring the loyalty nexus. Journal of marketing management 27, 785–807. Kanuri, V.K., Chen, Y., Sridhar, S., 2018. Scheduling content on social media: Theory, evidence, and application. Journal of Marketing 82, 89–108. Kapoor, K.K., Tamilmani, K., Rana, N.P., Patil, P., Dwivedi, Y.K., Nerur, S., 2018. Advances in social media research: Past, present and future. Information Systems Frontiers 20, 531–558. Kim, J., Kang, S., Lee, K.H., 2019. Evolution of digital marketing communication: Bibliometric analysis and network visualization from key articles. Journal of Business Research . Kim, M., Cappella, J.N., 2019. Reliable, valid and efficient evaluation of media messages. Journal of Communication Management . Kotarbinski, T., 2013. Praxiology: an introduction to the sciences of efficient action. Elsevier. Kumar, V., Aksoy, L., Donkers, B., Venkatesan, R., Wiesel, T., Tillmanns, S., 2010. Undervalued or overvalued customers: capturing total customer engagement value. Journal of service research 13, 297–310. Kusumasondjaja, S., 2018. The roles of message appeals and orientation on social media brand communication effectiveness. Asia Pacific Journal of Marketing and Logistics . Lamberton, C., Stephen, A.T., 2016. A thematic exploration of digital, social media, and mobile marketing: Research evolution from 2000 to 2015 and an agenda for future inquiry. Journal of Marketing 80, 146–172.

21 Lipsman, A., Mudd, G., Rich, M., Bruich, S., 2012. The power of “like”: How brands reach (and influence) fans through social-media marketing. Journal of Advertising research 52, 40–52. Liu, L., Lee, M.K., Liu, R., Chen, J., 2018. Trust transfer in social media brand communities: The role of consumer engagement. International Journal of Information Management 41, 1–13. Montgomery, D.C., 2017. Design and analysis of experiments. John wiley & sons. Myers West, S., 2018. Censored, suspended, shadowbanned: User interpretations of content moderation on social media platforms. New Media & Society 20, 4366–4383. Nawaz, M.S., Bilal, M., Lali, M.I., Ul Mustafa, R., Aslam, W., Jajja, S., 2017. Effectiveness of social media data in healthcare communication. Journal of Medical Imaging and Health Informatics 7, 1365–1371. Oviedo-Garc´ıa, M.A.,´ Munoz-Exp˜ osito,´ M., Castellanos-Verdugo, M., Sancho-Mej´ıas, M., 2014. Metric proposal for customer engagement in facebook. Journal of research in interactive marketing . Pszczołowski, T., 1967. Zasady sprawnego działania: wstep do prakseologii. Wiedza Powszechna. Radzi, N.A.A., Harun, A., Ramayah, T., Kassim, A.W.M., Lily, J., 2018. Benefits of facebook fan/brand page marketing and its influence on relationship commitment among generation y: Empirical evidence from malaysia. Telematics and Informatics 35, 1980–1993. Shen, G.C.C., Chiou, J.S., Hsiao, C.H., Wang, C.H., Li, H.N., 2016. Effective marketing communication via social networking site: The moderating role of the social tie. Journal of Business Research 69, 2265–2270. Unold, J., 2003. Basic aspects of the digital economy. Acta universitatis lodziensis folia oeconomica 167. Vejacka,ˇ M., 2012. Facebook advertising and its efficiency on the slovak market . Wielki, J., 2010. The impact of the internet on the development of web-based business models. Journal of Internet Banking and Commerce 15.

22 180 BIBLIOGRAPHY Paper 7

Y. Chawla, G. Chodak Recommendations for Social Media Ac- tivities to Positively Inﬂuence the Economic Factors 328

Recommendations for Social Media Activities to Positively Influence the Economic Factors

Yash CHAWLA, Grzegorz CHODAK

Wroclaw University of Science and Technology, Wroclaw, Poland {yash.chawla,grzegorz.ghodak}@pwr.edu.pl

Abstract. Social Media has become an important part of people life today, all over the world. The number of social media users has grown rapidly during last decade and are projected to keep growing. In this article the statistics concerning social media usage and the global characteristics of this phenomenon were presented. According to researchers, the number of social media users will exceed 3.02 Billion by 2019. Such a huge number of active users, have a strong influence on various economic areas. Social media has changed, the way companies carry out marketing and branding activities, the flow of information or news in the global scenario, international trade, public awareness, transparency of government or public administration, employability or talent acquisition etc. These socio-economic or micro-economic factors inevitably influence the economic factors of each country. In this article, the different aspects of growing number of social media users were analysed. Also, we recommend a set of guidelines for individuals, businesses and government which would positively impact the efficiency of social media usage. These recommendations, brings about a balance between the social, personal and professional aspects of individuals. For businesses and governments, they yield a better connectivity with the citizens / customers. A combination of both these outputs results in positive influence on economic factors.

Keywords: Social Media, Facebook, Macroeconomics, Economic Growth

1 Introduction

The increasing popularity of social media during last decade is one of the most interesting phenomenon of XXI century. Over 2 billion Facebook users and hundreds of millions of users in other social media services [11] have a direct or indirect influence, on various aspects of life [1, 22], and also the Economy. People who spend time on social media engage in relationships, watch advertisements, recommend products / brands, and indulge in lot of other things that affect the economy at both microeconomic and macroeconomic levels [5, 15, 24]. In the literature there were only a few papers which analyzed the impact of social media on economy from macroeconomic perspective [5, 17]. In our paper we want to fill this gap in the literature and make basis for further research in this area. 329

The main purpose of this article is to recommend important points that should be applied by social media users, i.e individuals, businesses / firms & government, for positively influencing the economic factors. The growth of internet through mobile devices and the vast usage of social networking sites by users around the world have direct or indirect effect on various economic factors. The recommendations for activities stated in this article can positively influence these economic factors, resulting in GDP growth of the country. The structure of the article is as follows. In the first section, current statistics concerning Internet and social media usage are shown. There-after the mutual relations between social media and economy are analysed. In the section after that, recommendations, for effective usage of social media, are enlisted for individuals, businesses and government. The last section includes short conclusions.

2 The current scenario of Internet & social media

2.1 Internet usage

Internet usage has increased significantly during last decade and mobile internet services were one of the key factors which contribute to this growth [1]. Mobile data traffic increased by 4000 times from 2005 to 2015. If we consider the period between 2000 to 2015, then the increase in mobile data traffic is a staggering 400 million times. This has been possible because of the increased affordability and availability of smart phones as well as the development in mobile communication technology over the last decade. Internet usage on smartphone is estimated as 97% of the total mobile data traffic. According to the usage patterns, a prediction by CISCO puts the monthly global mobile data traffic at 30.6 exabytes and number of mobiles connected per capita at 1.5. Three fourth of the mobile data traffic would be videos by 2020. Further, with the implementation of 4G, the average global mobile speed will surpass 3 Mbps by 2017. At that point the monthly mobile data traffic is estimated at around 9.9 exabytes. These predictions illustrates that there will be an increase of over 300% in the next three years [6].

2.2 Development and Growth of Social Media

In 1979, Tom Truscott and Jim Elis from Duke University created the Usenet, which was a system that enabled the users to have a discussion by posting public messages over the internet. This was one of the first signs of digital social networking. In 1997 an early social networking site named “Open Diary” was founded by Bruce and Susan Abelson. It was around the same time, the word “weblog” was first used. This formed the bases of the word “blog” a year later, as one of the bloggers jokingly divided the word in a phrase “we blog”. With the further evolution of internet speeds and affordability, social networking sites like MySpace and Facebook were founded. “Social Media” became a popular term henceforth [1]. Currently, majority of the time spent on the internet is on social media, which includes blogs, virtual games, social 330

worlds, social networking sites, collaborative projects, building communities etc. [1, 22]. Future evolution of the world wide web is going to be connected with social media, affecting every individual and business directly or indirectly, as the user base of social media expands rapidly [2, 25]. According to a study carried out at the Pew Research Centre in America, over 65% of the American adults were using social media in 2015 as compared to 7% in 2005. Although young adults aging between 18 to 29 years are in majority, the growth of social media users was observed in all the age groups. Based on gender, in 2015, it was found that there was a modest gap of 2% between men & women, with women leading the way. As far as the classification of the users based on socio-economic factors are concerned, it was observed that individuals with higher education level and household income, used social media more. Nevertheless, since past few years it was noted that over 50% of the users from lower income houses or less education had initiated using social media actively. In terms of users by geographic location: rural, suburban and urban; it was observed that the patterns were consistent over the decade of study with the percentage of users standing at 58, 68 & 64 respectively [23]. Analysing the global social media users base, a survey by eMarketer published in July 2017 shows a steady rise in the number of users worldwide. (Fig. 1).

3 2,5 2 1,5 1 0,5 0 2010 2011 2012 2013 2014 2015 2016 2017

Number of Users in Billions

Fig. 1. Year wise number of social media users globally on all platforms [16]

According to Eurostats, considering all the 28 countries in the European union, the frequency of Social Media usage by residents can be divided into everyday users (26.2%), every week users (13.3%), once a month users (2.8%), several times a month users (5.2%), not in the last twelve months (50.5%) and at least once a year (1.9%). These figures are very recent as they were updated on 19th September 2017. Country wise distribution of the users in European Union, on the same basis as above, is presented in the Figure 2. It can be seen that the highest percentage value of every day users exists in Norway (45.5%), followed by Malta (42.6%) & Ireland (41.8) [7].

331

120 100 80 60 40 20 0 Italy Spain Malta Latvia Serbia France Poland Ireland Greece Cyprus Croatia Austria Iceland Estonia Finland Sweden Norway Portugal Bulgaria Belgium Hungary Slovenia Slovakia Romania Denmark Lithuania Switzerland Netherlands Luxembourg Germany(until… Czech Republic United Kingdom Former Yugoslav…

Every day Users Every Week Users Once a month users Several times a month users Not in the last 12 months At least once a year

Fig. 2. Demographics of European social media usage on basis of time spent [7]

The projections for usage of social media in the coming years, show a steady rise in the number of users. In 2018, the user base is projected to grow by another 0.16 billion and reach to 2.62 billion social media users. This number would further grow to 2.77 billion, 2.9 billion and 3.02 billion in 2019, 2020 and 2021 respectively [16].

2.3 The spectrum of social media and the current user base

Users are active on more than one social media site, having the variety of different possibilities. Largest social networking website is decisively Facebook. A report by Kepios in September 2017 shows the statistics of the top 20 social networking websites in the world (Fig. 3). As it can be seen there are four social networking website with the number of users over 1 billion and next four with the number of users over 0.5 billion [11].

332

2 500

2 000

1 500

1 000

500

0 yy QQ LINE Viber Skype QZone Tumblr Twitter WeChat LinkedIn YouTube Pinterest Snapchat Telegram Facebook Instagram WhatsApp Sina WeiboSina Baidu Tieba FB MessengerFB

Number of active users in millions

Fig. 3. User base of top 20 social networking websites in 2017 [11]

3 Social media & Economy

The way of disseminating information as well as the way in which individuals or society perceives the information, both can affect the macroeconomic factors. Su- Heng et al. (2013) [5] as well as Jana Nunvářová and Pavel Bachmann (2017) [17] have extensively proved the same in their work on Social Networks & Macroeconomic Stability. In their work they describe three components of the economy which are also used for the calculation of Gross Domestic Product (GDP) by the expenditure approach. Consumption – the expenditures by individuals or households, Investment – the expenditure by the firms / businesses and Government spending, being the three components. Individuals these days prefer to read news on social media and also it has also become their prime source for personal updates. Optimizing the use of social media by individuals, businesses / firms and government in an effective way, would bring about a positive impact on economic growth [5, 17]. Factors of the economy, such as the employability of the citizens, are directly affected by the rising of social media. In this information age, the students need to be equipped with the most accurate knowledge about their fields as well as be connected to the right people globally. Managing one’s career as well as getting proper direction at proper time by professionals is also an area which is aided by professional social networking sites dedicated to the labour market, like LinkedIn. Social media provides a very effective platform supporting the flow of information on the labor market. This allows more effective entry into the labour market for students and a higher mobility on the labour market for employees for both young as well as experienced professionals. Social media may also decrease the cost of recruiting workers for 333

businesses. Vladlena Benson et al. (2013) have highlighted this impact in their work on Social Career Management [3]. Social media also has an impact on the trade, especially e-commerce, because SM marketing allows very accurate targeting of advertising audience at lower cost. This increases the effectiveness of marketing efforts by the businesses [9]. Several companies use Facebook, Instagram and other social networking platforms to their specific group of customers. Also, small niche businesses use social media services (e.g. Facebook, YouTube) to spread wide scale awareness about their products, what was previously difficult to carry out in such scale. These microeconomic activities, influence the aggregated supply and demand. Effective use of social media by businesses enables building relationships with customers and increases brand awareness [8]. Today's consumers look for product details in the internet and discuss about them using social media before making the actual buying decision. [18]. Sharing the information about products in Internet connected with social influence may cause Bandwagon effect or even speculative bubble. Sustainable business is a very important part of a stable economy. It contributes to the overall development of economy as well as generate employment, which takes the macroeconomic factors towards stability. Cristian Bogdan Onete, et al. (2013) in their work on social media in sustainable business development, highlighted that social media can be a good supporting tool for development of goods & services that can become sustainable [4]. For businesses it is really important to be connected to the customers and social media gives them a very convenient medium to do so [24]. Seonjeong Lee (2016) published his article about the same research area, but for the hotel industry. In his study he had found that the social media activities of the businesses’ that influenced customer’s psychological needs and impact on a sense of well-being, proved to be really positive for the brand usage intent of the customers. In the study it was also found that involving the customers in content creation and encouraging or rewarding them for sharing their experiences through social networking sites, attracted customers to engage more with the brand [12]. It is also worth mentioning that excessive & ineffective usage of social media proves to harmful not just socially or psychologically but also physically. The negative effects of social media abuse, such as the possible health problems with the cervical spine (this is due to the excessive use of smartphones and the unnatural position of the bent neck), social alienation, mental problems social media addition etc. All these problems result in the increase of health care expenses.

4 Recommendation for effective use of social media

It is almost impossible to ask social media users, especially the youth, to decrease the time they spend on social media. This is because social media has got integrate in their lifestyles. Hence, increasing the effectivity or positive outcome of this time spent on social media, would be one of the best case scenarios. The recommendations below, for social media users, can improve the efficiency of its use as well as it 334

effectiveness. The suggestions are divided into three groups: for individuals, for businesses/firms and for government.

4.1 For individual users

In this section we propose set of suggestions which can improve the efficiency of social media usage by individuals. First of all individual users need to be really vigilant in differentiating the fake news from the truth. This can be done by verifying the source of information back to its origin. If the time / effort doesn’t permit verification, then further circulation of information or news by the individual should be avoided. Social media news feed algorithms are becoming smarter by the day and show only the most relevant posts made by various users to individuals based on their activities. Following the right path as suggested above would aid the algorithms also to identify the fake news patterns. Moreover, individuals need to build up a strong social media profile, right from the time they enter into college [1, 3, 20, 22]. Individuals need to have a very clear vision of who they are, who they want to be and define their goals in life, before proceeding towards projecting themselves. Once this is decided, individuals need to ensure that the content they are sharing is in line with their defined goals & objectives. This is one of the most important factors which is missing from the social media activities of users today. Asking questions is always good, hence before posting, individuals needs to always ask himself / herself, that the content they are sharing would be beneficial to them or to their friends / family members / colleagues or in general to anyone who views it. If yes, then how? If no, then the content shouldn’t be shared. Our lives are a combination of three roles, Personal, Professional & Social. Therefore, it is very important to have a fine balance between all the three roles through the activities on social media. While interacting with content posted by others, the goals / objects set by an individual should be adhered to. One should interact with only those posts / content which are in line with their objectives. Entertainment or amusing content, shouldn’t cause a deviation from the same. It is very important to train the social media profiles, to show the most relevant information that is useful to the user. This can be achieved by keeping the activities aligned as suggested above. One needs to be really clear and understandable with their comments, their suggestion or any other content they share on social media. A share has the highest impact (always put views on the post shared from others, avoid sharing without your own views in the caption), then comes the comment and then a like or a reaction. A general observation while using social media is the annoying tags that we get on photos totally irrelevant to us. Never tag people, if the post is not relevant for them. Repeated tagging of people in un-relevant content for the purpose of increasing the reach, actually increases the spam score and decreases the reach of the posts gradually. Being punctual, regular, consistent and precise with sharing content goes a long way in creating a better impact on the audience.

335

4.2 For businesses / firms

Businesses and firms have been taking up social media marketing to great effect for over half a decade now. Since past couple of years, the integration of social media in the customer relationship management has also evolved to a great deal. Today the terms such as re-marketing, influencer marketing, highly precise targeting marketing etc. have become popular among businesses due to the high number of users being active in social media [13, 18, 21]. Now the businesses need to take the use of social media further by using it to influence positive change, for training & development of internal human resources, utilizing the internal resources as influencers, stress more on organic & content rich marketing, building up personal repot with customers etc. A few top end companies have already initiated doing this. But, the majority of the users of social media are in the developing countries. This makes it important for the businesses in those regions as well as businesses targeting those regions to be effective in reaching out there. Some of the companies which are trying to connect with consumers through social media are: Oreo – introducing engaging content, Netflix – making an effort to understanding their audience, Pampers – reaching its specific target market, Dove – creating inspiring content, GoPro – engaging customers in content creation, Royal Dutch Airlines – addressing customer grievances, Always – engaging users for social change etc.

4.3 For Government

Social media has had a lot of impact on the political outcomes in the recent times. Government policy plays a very important role in sustaining the stability of macroeconomic factors in the each country. Maintaining & balancing the cash flow, managing the policies of imports & exports, supporting local businesses etc. all influence the macroeconomic factors. Here the role of social media for the government is vital [10, 13, 15]. Social media trends affect the economic factors such as the stock market [14], consumer behavior in expenditures and many other factors. It is very important for the governments to have social media integrated into public governance. Literacy about social media is still not considered as important subject in the education system at lower levels. Integration of social media knowledge in the high school curriculum is of utmost importance, if a country wants to have socially literate & responsible citizen in the times to come. Currently social media journalism /citizen journalism doesn’t require a license, which has boosted the growth of fake news markets. Proper licensing for such e-portals should be made mandatory and initiator of fake news content should be punished by law. Fighting corruption is one of the major concerns for governments of developing countries. Social media is a very powerful tool for the same [10]. Citizen journalism through social media in a monitored way would prove to be really impactful in the fight against corruption. Government should also monitor the flow of forex out of the country due to the marketing expenditure carried out by the business on search engines as well as social media. 336

5 Projected outcomes of the above recommendations

The above recommendation would bring out an integration of individuals for a better society, integration of society, businesses & government for a better economy. For individuals the recommendations listed would improve their social & professional presence. This would give them better career opportunities as well social integration. Personal life is an aggregate of the social & professional life, hence it would be positive for all the three aspects of an individual’s life. As the users become more alert as well as aware, they would make better social, professional, political and commercial decisions. The business would adapt more organic methods of promoting their business, involving more and more citizen in the same. Currently, the largest social media sites as well as search engines are based in the United States. Hence for rest of the countries, when business spending for marketing on these sites, basically forex flow out of the country into the United States. Increase in more organic marketing would decrease this flow of forex out of the country. For governments, the capability of reaching out to majority of the citizens as well as businesses through social media would decrease the gap between them. Hence a better informed government and a more sustainable government system would evolve. As the three important components of an economy, i.e. individuals, businesses & government, are becoming more effective through these recommended social media activities, therefore the Economy is bound to be influenced positively.

6 Conclusions

Growth of social media is projected to hit new heights in the next four years. Another billion people are projected to become active on social media, apart from over 2 billion people who are already frequent users of it. These large numbers, make the effective use of social media - a necessity. Individuals need to equip themselves, to have the power to differentiate the right from the wrong. They also need to pave their social, personal and professional lives, to get the maximum desired output, which is aided by social media. Moreover, the businesses as well as governments have to use social media effectively in-order to keep the macro-economic factors stable, taking the country towards growth. In the article, due to the limited volume we pointed out only selected aspects of influence of social media on economy, and we gave brief recommendation on how to use it in more effective way. The directions of further research that we intend to undertake include elaborate description of the macroeconomic indicators that social media influence the most.

References

1. Kaplan, A.M., HaenleinM.: Users of the world, unite! The challenges and opportunities of Social Media, In Business Horizons, vol. 53, Issue 1, 2010, pp. 59–68, ISSN 0007-6813, DOI: 10.1016/j.bushor.2009.09.003. 337

2. Asur, S., Huberman, B. A.: Predicting the Future with Social Media, 2010 IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology, Toronto, ON, 2010, Pages 492-499. DOI: 10.1109/WI-IAT.2010.63. 3. Benson, V., Morgan, S.e, Filippaios, F.: Social career management: Social media and employability skills gap, In Computers in Human Behavior, vol. 30, 2014, pp. 519-525, ISSN 0747-5632, https://doi.org/10.1016/j.chb.2013.06.015. 4. Bogdan Cristian Onete, Dina Răzvan, Vlad Denisa Elena, Social media in the development of sustainable business, Amfiteatru Economic, vol. XV, Special No. 7, November 2013, pp, 659-670, ISSN: 1582-9146 (Print); 2247-9104 (Online). 5. Chen Shu-Heng, Chang Chia-Ling, Wen Ming-Chang.: Social Networks and Macroeconomic Stability. Economics: The Open-Access, Open-Assessment E-Journal, 8 (2014-16): 1–40. DOI: 10.5018/economics-ejournal.ja.2014-16 6. Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2015–2020 https://www.cisco.com/c/dam/m/en_in/innovation/enterprise/assets/mobile-white-paper- c11-520862.pdf , Last accessed: 2017/10/26 7. Frequency of communication via social media by sex, age and educational attainment level , Eurostat Report of Social Media Usage 2015, updated 2017, http://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=ilc_scp13&lang=en, Last accessed: 2017/10/27 8. Hudson, S., Li Huang, Roth, M. S., Madden, T. J.: The influence of social media interactions on consumer–brand relationships: A three-country study of brand perceptions and marketing behaviors, In International Journal of Research in Marketing, vol. 33, Issue 1, 2016, pp. 27-41, ISSN 0167-8116, DOI: 10.1016/j.ijresmar.2015.06.004. 9. Ioanid, A., Militaru, G., Negoita, O. D., Dumitriu, D.: Managing business using social networks: The relation between a company’s social media activities and the results obtained. Bucharest, Issue 7, Oct 2015, pp. 211-219. Niculescu Publishing House. 10. Ionescu, L.:, E-Government & social media as effective tools in controlling corruption in public administration. Economics, Management & Financial Markets, vol. 11, Issue 1, pp. 66-72. 11. Kepios: Global social networks ranked by number of users 2017, Digital in APAC 2017, https://www.statista.com/statistics/272014/global-social-networks-ranked-by-number-of- users/, Last accessed on: 2017/10/27 12. Lee, S.: The Effect of Social Networking Sites’ Activities on Customers’ Well-Being, Journal of Hospitality & Tourism Research, 201X, Vol. XX, No. X, Month 2016, Pages 1– 20, DOI: 10.1177/1096348016675926. 13. Li Qing, Wang TieJun, Li Ping, Liu Ling, Gong Qixu, Chen Yuanzhu: The effect of news and public mood on stock movements, In Information Sciences, vol. 278, 2014, pp. 826- 840, ISSN 0020-0255, DOI: 10.1016/j.ins.2014.03.096. 14. Lin Shen, Ren Da, Zhang Wei, Zang Yongjie, Shen Dehu: Network interdependency between social media and stock trading activities: Evidence from China, In Physica A: Statistical Mechanics and its Applications, vol. 451, 2016, pp. 305-312, ISSN 0378-4371, DOI: 10.1016/j.physa.2016.01.095. 15. Liu Ling, Wu Jing, Li Ping, Li Qing: A social-media-based approach to predicting stock comovement, In Expert Systems with Applications, vol. 42, Issue 8, 2015, pp. 3893-3901, ISSN 0957-4174, DOI: 10.1016/j.eswa.2014.12.049. 16. Number of social network users worldwide from 2010 to 2021 (in billions) - eMarketer, http://www.statista.com/statistics/278414/number-of-worldwide-social-network-users/, Last accessed: 2017/10/29 338

17. Nunvářová, J., Bachmann, P.: The influence of social networks on macroeconomic stability. In: Pavel Jedlička, Petra Marešová, Ivan Soukal, International scientific conference Hradec Economic Days 2017, Hradec Králové, Czech Republic, pp 636-642. ISSN 2464-6059 (Print), ISSN 2464-6067 (Online), ISBN 978-80-7435-664-3, University of Hradec Králové, Czech Republic (2017) 18. Parzonko, A.: Wpływ mediów społecznościowych na zachowania konsumentów. Handel Wewnętrzny, (6 (359)), 122-131. (2015). 19. Relling, M., Schnittka, O., Sattler, H., Johnen, M.” Each can help or hurt: Negative and positive word of mouth in social network brand communities, In International Journal of Research in Marketing, vol. 33, Issue 1, 2016, pp. 42–58, ISSN 0167-8116, DOI: 10.1016/j.ijresmar.2015.11.001. 20. Rosso, A.: Integrating your social media activities. Collector, 80(5), 22-27. Retrieved from https://search.proquest.com/docview/1651962021?accountid=46407(2014). 21. Saboo Alok R., Kumar V., Ramani Girish: Evaluating the impact of social media activities on human brand sales, In International Journal of Research in Marketing, vol. 33, Issue 3, 2016, pp. 524–541, ISSN 0167-8116, DOxI: 10.1016/j.ijresmar.2015.02.007. 22. Scott Carol F., Bay-Cheng Laina Y., Prince Mark A., Nochajski Thomas H., Collins R. Lorraine: Time spent online: Latent profile analyses of emerging adults' social media use, In Computers in Human Behavior, vol. 75, 2017, pp. 311-319, ISSN 0747-5632, DOI: 10.1016/j.chb.2017.05.026. 23. “Social Networking Usage: 2005-2015.” Pew Research Center. October 2015. http://www.pewinternet.org/2015/10/08/social-networking-usage-2005-2015/, Last accessed: 2017/10/26 24. Stephen Andrew T., Lehmann Donald R.: How word-of-mouth transmission encouragement affects consumers' transmission decisions, receiver selection, and diffusion speed, In International Journal of Research in Marketing, vol. 33, Issue 4, 2016, pp. 755- 766, ISSN 0167-8116, DOI: 10.1016/j.ijresmar.2016.08.003. 25. Zhang Xiaohang, Wang Cenyue, Li Zhengren, Zhu Ji, Shi Wenhua, Wang Qi: Exploring the sequential usage patterns of mobile Internet services based on Markov models, In Electronic Commerce Research and Applications, vol. 17, 2016, pp. 1-11, ISSN 1567- 4223, DOI: 10.1016/j.elerap.2016.02.002. Appendix B: Co-authorship declarations (Papers 1-7) Wroclaw, Poland, 12.03.2020

Anna Kowalska-Pyzalska, Ph.D. Associate Professor, Department of Operations Research and Business Intelligence, Faculty of Computer Science and Management, Wroclaw University of Science and Technology, Poland

I declare that in my work: Y. Chawla (60%), A. Kowalska-Pyzalska (25%), B. Oralhan (15%), Attitudes and Opinions of Social Media Users Towards Smart Meters’ Rollout in Turkey, Energies, 2020, 13(3), 732, my participation consisted reviewing the design of the survey, reviewing the literature, drafting, editing and reviewing the manuscript. I estimate my contribution to be 25%.

Oświadczam, że w mojej pracy: Y. Chawla (60%), A. Kowalska-Pyzalska (25%), B. Oralhan (15%), Attitudes and Opinions of Social Media Users Towards Smart Meters’ Rollout in Turkey, Energies, 2020, 13(3), 732, mój udział polegał na konsultacji projektu badania, opracowaniu przeglądu literatury, opracowaniu, edycji i recenzji manuskryptu. Mój wkład oceniam na 25%.

______

Signature / Podpis

Wroclaw, Poland, 12.03.2020

I declare that in the work: Y. Chawla (50%), A. Kowalska-Pyzalska (35%), P. Silveira (15%), Marketing and communications channels for diffusion of electricity smart meters in Portugal (Accepted for publication in the journal: Telematics and Informatics), my participation consisted reviewing the design of the survey, reviewing the literature, analyzing the data, drafting, editing and reviewing the manuscript. I estimate my contribution to be 35%.

Oświadczam, że w pracy:: Y. Chawla (50%), A. Kowalska-Pyzalska (35%), P. Silveira (15%), Marketing and communications channels for diffusion of electricity smart meters in Portugal (zaakceptowany do publikacji w czasopiśmie: Telematics and Informatics), mój udział polegał na konsultacji projektu badania, opracowaniu przeglądu literatury, analizie danych, opracowaniu, edycji i recenzji manuskryptu. Mój udział procentowy oceniam na 35%.

______

Signature / Podpis Setubal, Portugal, 10.02.2020

Prof. Paulo Duarte Silveira Department of Marketing and Logistics, ESCE- College of Business and Administration, Instituto Politécnico de Setúbal, Portugal

I declare that in the work: Y. Chawla (50%), A. Kowalska-Pyzalska (35%), P. Silveira (15%), Marketing and communications channels for diffusion of electricity smart meters in Portugal, my participation consisted in review of survey questionnaire, collection of data, English-Portuguese language translations, and contribution to literature review. I estimate my contribution to be 15%.

Oświadczam, że w pracy:: Y. Chawla (50%), A. Kowalska-Pyzalska (35%), P. Silveira (15%), Marketing and communications channels for diffusion of electricity smart meters in Portugal, mój udział polegał na przeglądzie kwestionariusza ankiety, gromadzeniu danych, tłumaczeniach na język angielski i portugalski oraz na recenzowaniu literatury. Mój udział procentowy oceniam na 15%.

______

Signature / Podpis Wroclaw, Poland, 12.03.2020

I declare that in the work: Y. Chawla (50%), A. Kowalska-Pyzalska (50%), Public Awareness and Consumer Acceptance of Smart Meters among Polish Social Media Users, Energies 2019, 12(14), 2759, my participation consisted reviewing the design of the survey, reviewing the literature, analyzing the data, drafting, editing and reviewing the manuscript. I estimate my contribution to be 50%.

Oświadczam, że w pracy: Y. Chawla (50%), A. Kowalska-Pyzalska (50%), Public Awareness and Consumer Acceptance of Smart Meters among Polish Social Media Users, Energies 2019, 12(14), 2759, mój udział polegał na konsultacji projektu badania, opracowaniu przeglądu literatury, analizie danych, opracowaniu, edycji i recenzji manuskryptu. Mój udział procentowy oceniam na 50%.

______

Signature / Podpis Wroclaw, Poland, 12.03.2020

I declare that in my work: Y. Chawla (60%), A. Kowalska-Pyzalska (25%), W. Widayat (15%), Consumer Willingness and Acceptance of Smart Meters in Indonesia, Resources, 2019, 8(4), 177, my participation consisted reviewing the design of the survey, reviewing the literature, drafting, editing and reviewing the manuscript. I estimate my contribution to be 25%.

Oświadczam, że w mojej pracy: Y. Chawla (60%), A. Kowalska-Pyzalska (25%), W. Widayat (15%), Consumer Willingness and Acceptance of Smart Meters in Indonesia, Resources, 2019, 8(4), 177, mój udział polegał na konsultacji projektu badania, opracowaniu przeglądu literatury, opracowaniu, edycji i recenzji manuskryptu. Mój wkład oceniam na 25%.

______

Signature / Podpis Malang, Indonesia, 26.11.2019

Dr. Widayat Widayat, Department of Management, Faculty of Economics and Business, University of Muhammadiyah Malang, Malang, Indonesia

I declare that in my work: Y. Chawla (60%), A. Kowalska-Pyzalska (25%), W. Widayat (15%), Consumer Willingness and Acceptance of Smart Meters in Indonesia, Resources, 2019, 8(4), 177, my participation consisted in collection of data, English-Indonesian language translations and contribution to literature review. I estimate my contribution to be 15%.

Oświadczam, że w mojej pracy: Y. Chawla (60%), A. Kowalska-Pyzalska (25%), W. Widayat (15%), Consumer Willingness and Acceptance of Smart Meters in Indonesia, Resources, 2019, 8(4), 177, mój udział polegał na gromadzeniu danych, tłumaczeniach na język angielski i indonezyjski oraz na recenzowaniu literatury. Mój udział procentowy oceniam na 15%.

Dr. Widayat, MM

Signature / Podpis

Wroclaw, 10.03.2020

Dr. Kamila Ludwikowska, Department of Management Systems and Organisational Development, Faculty of Computer Science and Management, Wroclaw University of Science and Technology

I declare that in the work: G. Chodak (35%), Y. Chawla (35%), A. Dzidowski (20%), K. Ludwikowska (10%), The effectiveness of marketing communication in social media. In: Proceedings of the 6th European Conference on Social Media, ECSM 2019: University of Brighton, UK, 13-14 June 2019 / Ed. by Wybe Popma and Stuart Francis. Sonning Common: Academic Conferences and Publishing International Limited, cop. 2019. pp. 73-81., my participation consisted in reviewing and writing literature background, and reviewing the manuscript. I estimate my contribution to be 10%.

Oświadczam, że w pracy: G. Chodak (35%), Y. Chawla (35%), A. Dzidowski (20%), K. Ludwikowska (10%), The effectiveness of marketing communication in social media. In: Proceedings of the 6th European Conference on Social Media, ECSM 2019: University of Brighton, UK, 13-14 June 2019 / Ed. by Wybe Popma and Stuart Francis. Sonning Common: Academic Conferences and Publishing International Limited, cop. 2019. pp. 73-81., mój udział polegał na recenzowaniu i pisaniu tła literatury oraz recenzowaniu manuskryptu. Mój udział procentowy oceniam na 10%.

______

Signature / Podpis