Behavioral Demand Response – A Technology to Support the Smart Grids of the Future
USAMA SHAHID SIDDIQUI
Behavioral Demand Response - A Technology to Support the Smart Grids of the Future
USAMA SHAHID SIDDIQUI
Master in Energy for Smart Cities Date: November 03, 2020
Supervisors: Aram Mäkivierikko, Hossein Shahrokni Examiner: Fredrik Gröndahl
SCHOOL OF ARCHITECTURE AND THE BUILT ENVIRONMENT Swedish title: Beteendebaserad efterfrågeflexibilitet - ett sätt att stödja framtidens smarta elnät
TRITA: TRITA-ABE-MBT-20732
I detta examensarbete görs en omfattande analys med hjälp av kvalitativa och kvantitativa metoder för att undersöka ifall användandet av LocalLife – en lokal social nätverkstjänst som pilot-testas vid Kungliga Tekniska Högskolan – har förbättrat användarnas beteende och attityder kring hushållselsförbrukning samt hjälpt till att förbättra det lokala samhällslivet.
Bostadshus ger upphov till en betydande del av världens energiförbrukning.
På grund av den snabba tekniska utvecklingen har byggnaderna blivit mycket mer energieffektiva i Sverige på senare tid. Det finns dock fortfarande förbättringspotential när det gäller att spara el genom att ändra de boendes konsumtionsbeteende. Att få till en sådan beteendeförändring är dock inte enkelt. Enligt litteraturen påverkas beteendet av uppfattningar och normer.
Det finns dessutom en global trend där allt färre människor lär känna sina grannar, kallad “globalt uppkopplad men lokalt isolerad”.
Denna uppsats studerar därför en strategi att spara el som går ut på lokal sammanhållning och ökad kunskap om elanvändning hos de boende i studentlägenheterna på Malvinas väg på KTH campus. Studien utförs bland användare av LocalLife, ett lokalt socialt nätverk för hållbara stadsdelar – genom att blanda metoder såsom enkäter och intervjuer. Åtta LocalLife- användare studeras i detalj. Resultatet presenterar de mest lämpliga delarna från de relevanta ämnena som kan möjliggöra en bestående beteendeförändring öka chansen för att behålla användarna. Resultaten visar att deltagarna: 1) visade på en ökad energimedvetenhet; 2) upplevde en förbättring av det lokala samhällslivet; 3) kände sig motiverade att ändra sitt beteende för att spara el.
Residential buildings are one of the main stakeholders to electricity consumption. As there is fast-paced technological advancement in electricity conservation, the residential buildings infrastructure has become very electricity-efficient in Sweden. However, there is still room for improvement with regards to electricity conservation via behavioral change. Meaning, residents have the potential to reduce household electricity consumption by developing a conservative behavior. The road to such a behavioral development is not straightforward. According to literature, behavioral change is influenced by different beliefs and norms.
There also exists a global trend that fewer and fewer people are able to name a single neighbor, and it is aptly called “Globally Connected yet Locally Isolated”. In this master thesis the strategy to achieve electricity conservation is based upon local social cohesion, and the awareness of electricity, at Malvinas – a student residence at the campus of KTH Royal Institute of Technology. The study is carried out at LocalLife – a local social networking service for sustainable communities – implementing a mixed methodology of surveys and interviews. 8 LocalLife users are studied in detail. The result incorporates the most suitable features from the relevant topics that could enable long term change and retainment of users. The results showed that the participants: 1) indicated an increased energy awareness; 2) reported an improvement in local community-life; 3) felt motivated to change behavior to facilitate saving electricity.
Keywords
Social network, energy awareness, electricity savings, behavioral change, consumer engagement, effect of Covid-19, community life, Nvivo.
This master thesis was written as the final part of the master’s program in EIT- KIC: Energy for Smart Cities, batch 2020, at the Royal Institute of Technology.
I would like to dedicate my thesis to my parents (Shahid Amir and Simeen Shahid), who brought me up with love, care, respect, and a lot of sacrifice.
They saw dreams for me that they themselves could not fulfill. To them I owe my life, and utmost gratitude.
Also, I would like to express my gratitude for my main supervisor from the Royal Institute of Technology and LocalLife, Mr. Aram Mäkivierikko, for his continuous feedback, support, commitment, and dedication from start till the end of this thesis. It is due to his stand for excellent academic reporting that I made extra efforts into improving the thesis.
Few people whom I especially want to thank are Mr. Hossein Shahrokni at KTH and LocalLife – my co-supervisor – without his ideas, vision and problem-solving skills, the thesis process would have been more complicated, and others at LocalLife – Mahmood Bakhtawar, David Enarsson, Patrik Nyström. I feel honored to have worked with the dream team.
Lastly, I am immensely thankful to a handful of important friends in my life, who supported and motivated me to work smart, long and hard in studies whilst also embracing beautiful life moments. Among them are my brothers – Talha Shahid, Yamman Khan, Hassan Khan and Malhan Khan – and my friends USman Saleem and Tsadik Kidane.
“The people who are crazy enough to think they can change the world are the ones who do.”
— Steve Jobs
Sammanfattning i
Abstract ii
Acknowledgements iii
Table of Figures vi
List of Tables vii
Glossary viii
1. Introduction 1
1.1. Problem statement and research gap 2
1.2. Research questions 3
1.3. Aims and objectives 4
2. Background 5
2.1. Swedish Smart Grid 5
2.2. Behavioral demand response through LocalLife 8
2.2.1. LocalLife 8
2.2.2. Interface and features 8
3. Methodology 12
3.1. Field experiments 12
3.2. Literature review 13
3.3. Surveys 15
3.3.1. Types of surveys 15
3.3.2. Qualitative data 16
3.3.3. Quantitative data 25
3.4. Interviews 30
3.4.1. Framework modelling and coding 30
3.5. Overview of survey and interviews 32
4. Literature Study 33
5. Results 40
5.1. Qualitative analysis 40
5.1.1. Surveys 40
5.1.2. Interviews 43
5.2. Quantitative analysis 46
6. Discussion 48
6.1. Summary and implications of work completed 49
6.2. Limitations and future work 50
References 52
Appendix 57
Table of Figures
Figure 1: Interlinkage between Demand Side Management, Demand
Response and Energy Efficiency ... 6
Figure 2: The smart grid chart (Dai et al., 2012) ... 7
Figure 3: LocalLife logo ... 7
Figure 4: Electricity feedback on daily consumption compared to other households in neighbourhood (Left); Energy summary screen showing that a new pause hour is available to participate in (Right) ... 10
Figure 5: Detailed pause hour screen with participate button and tips on how to reach the savings goal (Left), Pause hour history (Right) ... 11
Figure 6: Overview of surveys and interviews ... 32
Figure 7: Theory of Reasoned Action (Fishbein and Ajzen, 1981) ... 34
Figure 8: Theory of Planned Behavior (Ajzen, 1991) ... 34
Figure 9: Number of respondents in respective themes in Midterm (MS) and Final (FS) surveys ... 41
Figure 10: Number of responses from qualitative analysis of surveys ... 42
Figure 11: Average normalized thematic progress using qualitative data from surveys ... 47
Figure 12: EA, BC, LC mapped on Theory of Planned Behavior ... 49
Figure 13: The feed, showing a post. ... 58
Figure 14 The building page, showing its feed ... 59
Figure 15: LocalLife interface - Neighbors page ... 60
Figure 16: Direct messages page ... 61
Figure 17: Marketplace ... 62
Figure 18: Nvivo node hierarchy: BC ... 64
Figure 19: Nvivo node hierarchy: EA ... 64
Figure 20: Nvivo node hierarchy: COE ... 64
Figure 21: Nvivo node hierarchy: EF ... 65
Figure 22: Nvivo node hierarchy: LC ... 65
Figure 23: Nvivo peripheral node ... 65
Figure 24: Progress of thematic score from surveys ... 65
Table 1: Malvinas field experiments and activities ... 12
Table 2: Categorical list for literature review ... 14
Table 3: Nvivo qualitative analysis coding for Midterm and Final surveys ... 17
Table 4: SS(EA): All shortlisted questions (numericalized) ... 26
Table 5: MS(EA) and FS(EA): All shortlisted questions (numericalized)... 26
Table 6: SS(BC), MS(BC), FS(BC): All shortlisted questions (numericalized) ... 26
Table 7: SS(LC): All shortlisted questions (numericalized) ... 27
Table 8: MS(LC): All shortlisted questions (numericalized) ... 27
Table 9: FS(EA): All shortlisted questions (numericalized) ... 28
Table 10: Nvivo themes for interview ... 31
Table 11: Key concepts in theoretical model of behavioral change ... 34
Table 12: Summary of social psychology parametric ... 36
Table 13: Previous studies on behavioral change ... 39
Table 14: Nvivo thematic coding (interviews) ... 45
Table 15: social psychology parametric versus main themes ... 45
Table 16: Thematic distribution of references in interviews ...46
Table 17: Types of LocalLife feed content ... 57
Table 18: Interviewee attributes ... 62
Table 19 (left): Midterm survey open-ended questions coded on Nvivo; Table 20 (right): Final survey open-ended questions coded on Nvivo ... 63
Glossary
GHG – Greenhouse Gases
DSO – Distribution System Operator TSO – Transmission System Operator LL – LocalLife
PH – Pause hour
BCP – Behavior Change Program DR – Demand Response
BDR – Behavioral Demand Response DSM – Demand Side Management SS – Signup survey
MS – Midterm survey FS – Final survey
EA – Energy awareness BC – Behavioral status LC – Local Community-life
1. Introduction
Energy efficiency and conservation have been deemed the perfect joint strategy to counter the increasing energy demand and the consequent environmental impact. As global energy consumption continues to rise, especially in the developing countries, a strong decoupling of economic growth and energy demand needs to be envisioned (International Energy Ageny, 2018) – which means economic prosperity, industrial development, urbanization, etc. without a dramatic increase in the energy demand and greenhouse gas emissions. To put into perspective, the total world primary energy demand is projected to grow by 22.5% between 2025 and 2040 in the current policies scenario simulated by International Energy Agency (ibid).
Contrary to modern belief, the share of fossil fuel in this scenario – if the world continues with the current policies – is almost the same in the starting and the ending years – 2025 and 2040 (ibid). To sum up, this concludes that a chain reaction detrimental to the planet has already started in that there will eventually be an increased demand of energy in the coming years. This increased demand will be met by an increased supply. The increased energy supply will be provided mainly by the fossil fuels because the renewable technologies have higher costs, intermittency, and have relatively lower energy density – making renewable technology even harder to fit into the developing economies where subsidizing renewable technologies is the last priority. This has set off alarms worldwide to collaboratively strategize climate action plan and reroute the energy systems towards a more sustainable pathway. Energy conservation – restraining the increase in energy demand (or Demand Response) – is one of the strategies that several nations have adopted very seriously, Sweden being one of them(City of Stockholm, 2016).
In Sweden the residential sector consumes 23% total final energy consumption (International Energy Agency, 2020b), out of which 26% is the share of electricity demand and the remaining 74% is taken up by the heat demand. If we focus on the share of residential electricity demand in total Swedish electricity demand, it was 20% in 2018 (Statistiska Centralbyrån, 2019). Although the residential sector does not represent a big share of electricity demand, but there is a crucial room for innovation related to low- cost energy efficiency and conservation measures.
The climate goal is to apply energy efficiency measures in the residential sector and carefully study the consumer behavior for optimization (Olga Kordas et al., 2015). Although there is significant potential to conserve energy in the residential sector via energy efficiency improvements, a major constraint in the struggle for reduced consumption remains the lack of motivation and awareness by the residential consumers. Policy makers and utilities are
consumers to conserve energy. This is the starting point of the concept of behavioral change– a social sciences breakthrough that incentivizes electricity consumers to conserve more. Depending on a multitude of variables and no matter how difficult, changing behavior in a community, through energy feedback has the potential to conserve energy (Allcott, 2011). However, experts are aware of many challenges in the path of building such dynamics where, firstly a community could be motivated to change its behavior for a long period and, secondly, the new electricity consumers joining the community could be motivated to follow the social norms of the community.
On top of this, the solution should be dispatchable, cost effective, and must promise noticeable savings.
LocalLife is a local social networking service – operating via website and mobile app – for sustainable communities aimed at engaging citizens on issues relating to the local community and sustainability at a student residence on KTH campus in Stockholm, called Malvinas. Being a local social network, LocalLife caters for local neighborhood needs whilst also attempting to save electricity at a community level through social and individual incentives.
This thesis will explore how a rich analysis, especially supported by Nvivo – a qualitative data analysis software that makes collecting, organizing, analyzing and visualizing unstructured or semi-structured data types very simple – amalgamating qualitative and quantitative methods, could help LocalLife understand better their user behavior and mindset. Three main themes are pursued:
1. Energy Awareness (EA): familiarity with the concept/fictitious value of energy, awareness of – and control over – electricity consumption of one’s household and other related indirect indicators.
2. Behavioral Change (BC): a change in behavior, routine and/or habits to facilitate energy savings in one’s household/neighborhood (Malvinas).
3. Local Community-life (LC): related to neighborhood connections, social life, livelihood, comfort and other concerns that empower community life at Malvinas.
1.1. Problem statement and research gap
Advanced governments proactively use policy instruments founded on campaigns to educate people to change their behavior (John et al., 2013). The campaigns involve the tradeoff between raising awareness, increasing knowledge stock, and economic benefits. However, research suggests that increasing awareness does not always correlate with corrective behavioral changes and that other factors also play a role. This model is correctly depicted
in the Theory of Planned Behavior (Ajzen, 1991), which we will describe in Section 4.1.
Hence, approaching the problem of reducing residential electricity demand is more complex and cannot be only countered by increasing electricity consumption awareness. Also, forcing people to change behavior can be costly and results in insufficient reductions and low retainment, making it a high- cost low-benefit strategy (Ellabban and Abu-Rub, 2016).
Substantial progress has been made in providing citizens with visualization and feedback technologies via monitors, apps and in-home displays. However, those that initially engage seem to stop their engagement after a limited amount of time – a phenomena also called “the honeymoon effect”.
Citizens are at the heart of the Demand Response programs in that these programs are only successful upon a high engagement by the citizens. While there is a significant focus on technology development and evaluation, they are also targeted to facilitate the satisfaction and engagement of the citizens.
The citizens are predominantly occupied with questions concerning family, friends, work, stability, safety, and health. Values of altruism and ethics drive some of their decisions, while financial and time constraints drive other decisions. Most of them are unaware about how the electric power grid works, since electricity - for most citizens - has always worked perfectly, “like magic”, and they have never needed to learn the complexity of the supply-chain of electricity. Therefore, requesting new actions or engagement on their behalf in the context of smart grids is a considerable challenge, and more so to sustain.
Ultimately, there is a significant research gap related to electricity savings via a combination of behavior change and energy awareness.
A research gap also exists in energy-efficient student residences, where electricity is extremely cheap (Sönnichsen, 2020) and household electrification factor is 100%. Geographically, there is also a research gap in such residences, especially with regards to policy making in the Northern Hemisphere, which are interwoven in the social fabric identified with a high sustainability index(Mont et al., 2013).
1.2. Research questions
This thesis study will focus on verifying the hypothesis that LocalLife, within 5 months of service at Malvinas (one full semester at KTH), has improved or affected the energy awareness, behavioral change, and local community-life.
1. How has LocalLife affected the energy awareness of users at Malvinas in the 5-month service?
2. Whether LocalLife has motivated the users at Malvinas to change their behavior to conserve electricity by participating in the Pause Hours (refer to Section 2.2.2.2)? Are the users motivated to change their behavior even without the Pause Hours?
3. In what sense has LocalLife actively affected local community network among the residents? Did LocalLife somehow serve a way to build new connections with neighbors and jointly perform sustainable tasks?
1.3. Aims and objectives
The aim of this thesis project is to further develop user-based behavioral research for LocalLife at KTH Malvinas. Respecting the aim, the following set of objectives are laid out:
1. Conduct field experiments at Malvinas that engage users and affect their energy awareness (EA), behavioral change (BC) and local community-life (LC).
2. Conduct a firm literature study on previous experiments related to energy savings behavior change and theoretical background in social psychology to help build a framework to explain LocalLife user behaviors.
3. Conduct a series of surveys and detailed interviews to understand a LocalLife user’s lifestyle, routine, perceptions, and values.
4. Build a qualitative analysis framework on Nvivo (qualitative analysis software) and analyze interview transcripts and open-ended survey questions using an established social sciences approach – Grounded Theory.
5. Perform a rich thematic quantitative analysis on survey responses.
6. Consolidate data from surveys and interviews into holistic valuable insights to track energy awareness, local community network and behavioral change of users.
7. Seek areas of improvement related to the - on-app and other – provides by LocalLife and identify the challenges and limitations in conducting this thesis study. These limitations could be considered for future research.
2. Background
2.1. Swedish Smart Grid
Prior to 1996 the Swedish electricity market was a dominated by a monopoly.
The Swedish electricity market then undergone reforms which revolutionized the monopolistic market whilst maximizing the welfare of the society. The reforms also brought liberalization to the electricity market at all levels of the electricity supply chain – generation, transmission and distribution (Ribbing, 2018). Modern technological development, especially with regards to data sciences, it is now possible to implement real-time electricity pricing.
Noteworthy changes can be observed on the consumer-side, where electricity consumers can now choose from a large pool of electricity suppliers. These suppliers offer varied services and have opened opportunities for new market players like electricity traders. This has substantially changed the market dynamics from being linear to a very complicated and yet transparent system.
Furthermore, Sweden is also experiencing a rapid growth of renewable energy in the electricity generation mix. Solar and wind power drive up the renewables share (International Energy Agency, 2020a). Wind and solar power are replacing coal and oil as well as slowly ousting nuclear power.
Although nuclear power is cleaner than fossil fuel power but the safety of spent radioactive fuel, regulations, terroristic risks are not the reasons for nuclear decline. It is because the levelized costs of electricity of non-nuclear generation are lower than that of nuclear power – making nuclear generation a lower-profit and risky business to invest in, especially during the times when investors are boarding the bandwagon renewable penetration due to government subsidies (Proost and Pepermans, 2016). Increased renewables in the generation mix means increased intermittency and fluctuations from the supply end. The system has until now relied upon their large share of flexible hydropower (Pumped Hydro Storage) to provide room for penetration of intermittent sources. In Sweden there is a lot of “ordinary” hydropower but few pumped storage (Karin Byman, 2016). This opens even more doors for curtailing electricity demand. However, with the growth of the percentage of renewables in the electricity mix, electric vehicles, energy-efficiency measures in the industrial sector, and the adoption of domestic electricity generation, the need for demand flexibility is expected to increase in the foreseeable future (Swedish Energy Markets Swedish Energy Markets Inspectorate, 2017).
Inevitably, there is a solid business case for ancillary and flexibility services to help maintain balance in the electric grid.
A study conducted by Swedish Energy Markets Swedish Energy Markets Inspectorate (2017) has estimated the potential for demand side flexibility in all consumption sectors (residential, industrial, services, properties, other).
Their study concludes that a demand flexibility potential of 300MW – equivalent to a full-sized combined cycle gas turbine plant (Energy Information McGrath, 2019) – is present in the residential sector. In other words, this means that instead of supplying 300MW electricity to the consumers, the consumers can save 300MW electricity by applying voluntary Management (Eissa, 2018), Demand Response (Ellabban and Abu-Rub, 2016) and Energy Efficiency. Worrell et al. (2003) and Popescu et al. (2012) discuss the energy efficiency measures for industrial and residential consumers, and develop a critical viewpoint how Energy Efficiency is partaken at the policy level. The minute definitions of these categories may vary in different parts of the world. But the overarching definitions and the interlinkage between the three categories are universally accepted.
Figure 1 illustrates the interlinkage. DSM includes all demand-reducing measures. Put differently, it includes both the drivers for its execution, demand response and energy efficiency. DSM seeks out a balance between energy demand and supply both on the side of grid operators, utilities, and consumers. While DR does it from consumer’s side, DR encourages consumers to reduce their energy demand in the short term, while DSM includes not only these in DR, but also long-term or permanent energy efficiency measures.
As far as Demand Response programs are concerned, economic incentives for the consumer are common in these programs (via time-based or price-based mechanisms (Ellabban and Abu-Rub, 2016)). LocalLife differs in this regard, as it provides social incentive and no economic incentive to save electricity. It is because the residents at Malvinas do not pay separately for their electricity consumption. Dai et al. (2012) constructs a simplified schematic of the smart grid architecture. It is shown in Figure 2. Nemati et al. (2014) elaborates a well-rounded overview of the Swedish smart grid and the challenges therein.
It is to be noted that the red lines in Figure 2 represent bi-directional communication channels and blue lines represent the bi-direction power flow
Figure 1: Interlinkage between Demand Side Management, Demand Response and Energy Efficiency
and that all the parties are highly intelligent and can take decisions based on the intricate flow of information via the communication channels.
Furthermore, in this figure behavioral demand response exists mainly at the nodes of commercial and residential customers. The industrial customers have no potential for behavioral demand response because of automatized and already optimized processes therein (International Energy Agency, 2017). To put into perspective, when magnified, the industrial customers comprise of industrial machines and critical loads that are tightly knit within the supply chain, in that reducing production of an industry would have severe production losses and supply-chain disruption among other alarming consequences (ibid). Economically explaining, a pause in the supply of an industrial good causes the market prices of the good to increase – compromising consumer surplus. As with hospitals, industrial customers are also untouched in terms of implementing behavioral demand response. This is the reason why, instead of behavioral change, industrial customers are only subjected to installing energy-efficient systems and machinery which reduces energy intensity (ibid).
Figure 2: The smart grid chart (Dai et al., 2012)
Figure 3: LocalLife logo
2.2. Behavioral demand response through LocalLife
2.2.1. LocalLife
LocalLife is a local social networking service for communities, critical about sustainability, aimed at engaging citizens on issues relating to the local community. Being a local social network, LocalLife caters for local neighborhood needs, which are identified throughout iterative field research and dialogue with the residents. The inherent social nature of LocalLife has enabled the network to address specific local needs. This service has possibly allowed for a higher user-retention for sustainability issues. LocalLife provides a natural context for personalized energy feedback, which reports back to individual residents how much electricity they have consumed and how their consumption compares to that of similar households. The personalized feedback also gives collective comparisons with the local neighborhood.
LocalLife was implemented and evaluated together with 152 student households joining the local social network at the KTH Live-In Lab student residence (commonly known as Malvinas). All apartments have smart meters (Zheng et al., 2013) for electricity, hot tap water, and sensors for temperature and CO2 level. Almost all apartments are occupied by single students and are 20 sq. meters in size.
The residents were recruited to the LocalLife network and were then exposed to a variety of social and locally themed campaigns and activities (summarized in Table 1). In parallel to the activities, the concept of “Pause Hour" (see Section 2.2.2.2) was introduced to the residents (a behavioral demand response strategy). The PH was launched during certain hours when there was, assumably, a peak-load in the electricity grid. Within the scope of this pilot, LocalLife did not set the PH launch times based on forecasted grid data.
Rather, the peak times were preselected based off of the historical electricity- demand profile. Residents were encouraged to reduce their energy consumption during the PH. Through LocalLife the potential for such environmental feedback to be noticed and acted upon was tested and evaluated in the context of a frequently used social network.
2.2.2. Interface and features
LocalLife has features having to do with social networking, but also a set of sustainability features that sets it apart from other social networks. The features are described in Appendix 6.2.A.1 except for electricity feedback and Pause Hour, which are described below.
2.2.2.1. Electricity feedback
The most relevant sustainability feature – the electricity feedback – was designed to use the social influence that is inherently available in the LocalLife local social network (Mäkivierikko et al., 2019a).
The Live-In Lab residents can see electricity feedback (under Energy consumption) based on their daily, weekly or monthly consumption compared to the other households in Malvinas (Figure 4).
2.2.2.2. Pause Hours
The residents can also participate in so-called Pause Hours, during which they should attempt to reduce their electricity consumption. The pause hours are intended to occur during the time of peak consumption of the grid, thus decreasing the risk of overloading the grid (InteGrid, 2020). Provided that such peak times are known in advance, the residents are notified and asked to participate in a Pause Hour the day before it occurs, but such notification can also be given at short notice, usually one hour before. A notification to participate is also sent out 30 minutes prior to the start of the Pause Hour.
When a new Pause Hour is available, it is shown in the My energy page in the app (at the bottom of the right image in Figure 4) including a user-specific savings goal calculated based on the resident’s previous consumption during the same hour for the past 10 similar days. For the current thesis, the calculation of energy savings goal and some other variables are omitted as they are irrelevant.
When the resident clicks the yellow pause hour card in the right image in, a detailed page is shown in the left image of Figure 5, where the user can choose to participate in the pause hour, see the savings goal, and see examples on how the given savings goal can be achieved. This page is where the user lands after clicking the link in the email or push notification that asks the user to participate. Those who have chosen to participate in a pause hour are reminded five minutes before it starts by email and push notifications.
When the electricity consumption during the pause hour for all participating residents is available (usually during the next day), it can be seen whether the user has reached the savings goal and thus passed or failed the pause hour.
The user is notified and is taken to the pause hour history page, where the actual savings compared to the savings goal and to other participating users can be seen in the right image in Figure 5.
The Pause hours serve the primary objective to saving electricity in a joint, community effort – targeting behavioral change. Whereas the purpose of the energy consumption feature is to bring competitiveness to saving electricity and to increase specific energy related awareness among the LocalLife users.
Based on these two methods, LocalLife can assess personal preferences, pro- environmental behaviors and other factors.
Figure 4: Electricity feedback on daily consumption compared to other households in neighbourhood (Left);
Energy summary screen showing that a new pause hour is available to participate in (Right)
Figure 5: Detailed pause hour screen with participate button and tips on how to reach the savings goal (Left), Pause hour history (Right)
3. Methodology
3.1. Field experiments
The first step to Behavioral Demand Response was to conduct field experiments/activities that were focused mainly on increasing energy awareness, changing behavior and improving neighborhood ties.
The main field experiments are listed in Table 1. To keep this thesis concise, further explanation of the field experiments and activities would be unnecessary. However, it is important to understand the diversity and rigor of the effort that were made to engage LocalLife users. It was these experiments that later translated into changes in energy awareness, behavioral change and local community-life as we will see in Results section.
Table 1: Malvinas field experiments and activities
Experiment/activities Details Outdoor signups;
Movie and game nights; corridor potluck dinners,
Aimed to recruit Malvinas residents and to create energy awareness.
Food coupons Food coupons were given to LocalLife users who showed engagement in the activities.
Pause Hour:
individual and social incentives
On passing certain Pause Hours, the LocalLife users were given individual incentives such as food coupons, and on one occasion – towards the end of the semester – the Grand Pause Hour was conducted. The winning building at Malvinas of the Grand Pause Hour was given a boat-ride across Stockholm (or equivalent).
LocalLife
brochure LL brochures were inserted in the post boxes of Malvinas residents. The brochures contained information about energy and instructions about signing up on LocalLife platform.
Recruiting building ambassadors
Four building ambassadors were hired as building ambassadors. They were Malvinas residents and LocalLife users who performed multiple tasks related to getting more residents to signup on LocalLife and provide valuable Malvinas community insights.
Behavioral
Change Program It was a fast-paced series of experiments that were performed back-to-back for 4 weeks. With 4 treatments groups – each treated differently for the entire period of the program. The treatment groups were:
1. G1: only treated with general information on the LocalLife platform (example: posts in general feed)
2. G2: on top of general information, they were treated with social incentives to participate in the Pause Hour to save electricity (example:
Zoom Pizza – an online pizza baking activity between neighbors at Malvinas)
3. G3: on top of general information, they were giving individual incentives to pass the Pause Hour (example: Foodora coupons)
4. G4: on top of general information, they were sent a customized energy newsletter via email.
3.2. Literature review
A literature review for this thesis was necessary to understand the intricacies about the research topic. Previous similar experiments, governance of the modern Swedish smart grid, sustainable development and psychological models in communities were studied meticulously. Also, to explain LocalLife user behaviors it was quintessential to study literature about social psychology, which played a pivotal role in building an Nvivo framework to thematically analyze qualitative data. Additionally, the framework served dual-purpose – qualitative analysis and setting ground for quantitative data, which was later analyzed using Microsoft Excel. To be precise, the literature review helped achieve objectives 2, 4, 5 and 6 from Section 1.3.
A rigorous approach for literature review was adopted based on two steps:
wide and deep. The wide approach was used to study the different literature
and get an overview of the crux of the literature. Whereas the deep approach was used to study in-depth the methodology and findings of the literature. The deep approach usually follows the wide approach.
A critical 3-pass approach by Keshav (2007), in combination with a meta-data search on the available databases, was used to traverse literature – yielding time savings and necessary critiques. Furthermore, to guide the analysis the book Research Methods in Psychology (2011) was used. Although plenty in this thesis study, Table 2 shows the categorization of the literature pertaining especially to the fields of energy technology and social psychology.
Table 2: Categorical list for literature review
Material type List
Research papers (Mäkivierikko et al., 2019a) (Mäkivierikko et al., 2019b) (Zheng et al., 2013) (Ahvenniemi et al., 2017) (Azar and Al Ansari, 2017) (Bittle et al., 1979) (Brauer et al., 1995) (Burger, 2001)
(Crutchfield, 1955) (D’Oca et al., 2017) (Dai et al., 2012) (Dolan and Metcalfe, 2015) (Fishbein and Ajzen, 1981) (Harding and Hsiaw, 2014) (Harkins, 1987) (Janis, 1971) (Lauria et al., 2019) (Nilsson et al., 2018) (Postmes, 2001) (Ribbing, 2018)
(Snyder, 2019) (Zajonc and Sales, 1966)
KTH theses (LUO, 2018) (Haglund, 2017) (Kaczmarek, 2015) (Elliott, 2012)
Books (The Intergovernmental Panel on Climate Change, 2014) (Alan Bryman, 2012) (John et al., 2013) (Proost and Pepermans, 2016) (Smith et al., 2011) YouTube videos (Black, 2016) (Green and Green, 2014a) (Green
and Green, 2014b) (Laskey, 2013)
Blogs and articles (The Intergovernmental Panel on Climate Change, 2014) (Olga Kordas et al., 2015)
Reviews by
organizations (IEA, ExxonMobil, etc.)
(ExxonMobil, 2020) (International Energy Ageny, 2018) (United Nations, 2015) (Statistiska
Centralbyrån, 2019) Databases (IEA, Statista,
etc.) (International Energy Agency, 2020b) (International Energy Agency, 2020a) (Sönnichsen, 2020)
For research papers in the field of social psychology, the papers that addressed experiments, were only considered from the last ten years (2010 onwards) to build up from the most recent research.
The following keywords used to filter research papers:
1. Energy feedback
2. Consumer engagement 3. Energy behavior
4. Social networks 5. Smart grids
6. Demand response
7. Demand side management 8. Grounded theory
9. Theory of planned behavior 10. Theory of reasoned action
3.3. Surveys
Surveys are the essential tools of the qualitative analysis toolbox. They were conducted to obtain information from LocalLife users. Each survey created was utilized to assess attitudes, perception and behavior and to receive their opinion about the needed improvements on the app. This feedback served unequivocal doorway to improvements in LocalLife strategy. Furthermore, surveys empowered LocalLife users, increasing their retainment and connection with LL. Surveying was supposedly the excellent methodology in finding out what it was that satisfied LocalLife users.
LocalLife conducted three batches of surveys, which were carefully designed and rolled out at different times. The important questions in the preceding survey were not omitted from the following surveys. However, all the surveys were designed to assess at least the three main themes of LocalLife – energy awareness, local community-life; energy savings behavior. The objective of the survey was to evaluate the status and progress of the main themes (EA, BC, LC).
3.3.1. Types of surveys 3.3.1.1. Signup survey (SS)
It was a mandatory survey which appeared after signing up on LocalLife (both on website and on the app). It was designed to be brief and be able to evaluate to an extent the current status of the LocalLife user based on three main themes. The SS survey only contained closed-ended questions (quantitative
data). As a result, we will omit SS from qualitative analysis in the later sections of this report.
3.3.1.2. Midterm survey (MS)
It was an optional survey, conducted in March 2020 – hence the name. It was designed to assess the three main themes of LL, in addition to more insights regarding the performance of users and how they perceived LL. MS survey contained open- and closed-ended questions – enabling it to be utilized for qualitative and quantitative analyses.
3.3.1.3. Final survey (FS)
It was conducted in May/June 2020. At that time students were finishing their semester studies at KTH. It was also an optional survey, containing open- and closed-ended questions.
3.3.2. Qualitative data
The qualitative data from the surveys corresponded to the open-ended questions which were asked to get further explanation regarding the previous questions.
The open-ended survey questions were manually coded on Nvivo into respective themes using a Grounded Approach for qualitative analysis as explained by Smith et al. (2011), in that instead of keeping the themes (nodes) constant, more themes were created as the survey responses were traversed.
These extra themes provided insights to understand the complexity of the energy perception, attitudes, and behavior of the individual. On top of that, it also provided feedback that was later used to improve the LocalLife app, strategy, and operations. To be precise, besides the main themes (EA, LC and BC), more themes related to energy control, sustainability motivation, past energy behavior, security at residence and other issues were created (see Appendix 6.2.A.3 for full list of themes/nodes). These themes will be represented by positive number, showing the number of occurrences. Table 3 shows the description of the shortlisted themes.
The table contains survey-respondent codes in ‘XXYY’ format, where ‘XX’ is the survey type (MS or FS), and ‘YY’ is the two-digit respondent number (example: 08).
It is to be noted that only MS and FS contained open-ended questions, SS did not – this being the reason SS will be omitted from Nvivo qualitative analysis in the proceeding sections.
Table 3: Nvivo qualitative analysis coding for Midterm and Final surveys
Survey Themes Description
BC Negative No change in behavior in the context of energy conservation was identified. Hence, every addition to this theme represents that LocalLife has not been able to affect behavior change, with respect to energy conservation, of the survey respondent.
However, it is necessary to state that it might be possible that a certain respondent’s behavior was affected by LocalLife, but the effect was indirect, negligibly small, or not related to energy conservation. It may also be possible that the survey respondent neglected changes in his behavior, thinking they are not relevant, that contributed to energy conservation and did not state them explicitly in the survey. In such special cases, it will still be considered that there is no behavioral change (BC Negative will increase by 1 point).
Example:
MS08:
Question: “If ‘yes’ for taking part in Pause Hour, during the pause hours have you done anything to reduce your consumption e.g. held off of cleaning/cooking/
brewing/heating/gaming; gone outside for a walk/activity;
spent time with neighbours (beyond the movie night). If so, what?”
Response: “No”
BC Neutral Not conclusive whether a behavior change with regards to energy conservation has happened due to the services provided by LocalLife. This theme accounted for: 1. neutral
individual responses, 2. for when the overall a neutral position of the respondent was depicted after evaluating all the answers given by them.
Example:
MS22:
Question: “Please motivate why/why not to avoid cooking/vacuum cleaning between 18:30-19:30 if you knew that it would be good for the environment due to less polluting electricity production”
Response: “If I'm too hungry I'll have to cook, if not then it's ok”
BC Positive A straightforward influence/change in behavior for energy conservation was observed. It also counted single word responses, without a sentence-body, that depicted a positive influence on behavior change.
Example:
MS05:
Question: “Please motivate why/why not to avoid cooking/vacuum cleaning between 18:30-19:30 if you knew that it would be good for the environment due to less polluting electricity production”
Response: “environment”
EA Insight LocalLife has helped users learn new insights about their consumption. For instance, knowing that a privately owned space heater consumes several times more electricity than household lighting consumption.
In this context – to store insights related to energy awareness – EA Insight was created as an extension to EA Positive. It demonstrated the perspective of a respondent in terms of the lessons they learnt about energy/electricity via LocalLife.
Alternatively, EA Insight was created to store valuable details of the reasons of a positive influence on energy awareness of the respondent.
Contrast EA Insights with EA Positive, and it can be noticed that their definitions are complimentary to each other and overlap – increasing the risk of double-counting. To cater for this risk, EA Insights will not be counted in the final calculations, but it will be useful compile a pool of lessons, that influenced energy awareness, LocalLife was able to impart to its users.
Example:
MS21:
Question: “If yes (referring to the previous question “Have you visited the My Energy section in LocalLife to learn more about your household electricity consumption?”) what new insights regarding your household electricity consumption have you learned from the energy feedback (excluding pause hours)?”
Response: “Apartments facing north side consume more electricity for lighting up the room.”
EA Negative No change in energy awareness was identified. Hence, every addition to this theme represents that LocalLife has not been able to affect/influence energy awareness of the survey respondent.
Example:
MS24:
Question: “If yes (referring to the previous question “Have you visited the My Energy section in LocalLife to learn more about your household electricity consumption?”) what new insights regarding your household electricity consumption have you learned from the energy feedback (excluding pause hours)?”
Response: “Unfortunately measuring of power consumption broken after 2 days and I have resigned to get it fixed while I live here now.”
In this case, the respondent’s electricity meter was faulty.
Therefore, the electricity feedback that the respondent continuously received, on LocalLife, showed zero electricity consumption. This response was categorized as EA Negative.
EA Positive An observed straightforward influence on energy awareness.
Example:
FS30:
Question: “Did you gain any new insights regarding your household electricity consumption from the pause hours”
Response: “Yes, I did.”
EF Easy Electricity feedback feature was easy to understand. The questions in this theme addressed whether logically and/or aesthetically the electricity feedback was easy to understand with regards to the respondents’/neighborhood’s electricity consumption, especially during the Pause Hours.
Example:
FS03:
Question: “What did you think about the pause hour concept, feedback and notifications. Was it easy to understand? Do you have any suggestions that could help us improve the pause hours?”
Response: “It was easy to understand”
EF Hard Driven from the same intuition as EF Easy but the opposite.
Example:
FS10:
Question: “In case you found the energy feedback difficult to understand, what could be improved? What kind of feedback would be helpful for you?”
Response: “It would be better that it tracked usual usage and showed the changes from usual in all plug points which can help to track exactly which device is using energy a lot as the cooking doesn't show the energy consumption. Also, it can help us replace the faulty devices by tracking the power consumption changes.”
In the example, FS10 explains why the energy feedback feature was difficult to understand. Therefore, it was coded to EF Hard.
EF Neutral Electricity feedback feature was neither difficult nor easy to understand. This theme was also created to categorize the responses which were inconclusive and needed further speculation.
Energy
competitiveness
This theme contained responses illustrating a competitive/comparative way to perceive the electricity feedback and the Pause Hours. To categorize the details of competitive electricity saver and to learn their perspective about the latent neighborhood competition to save the most electricity during the Pause Hours.
Example:
MS12:
Question: “If yes (referring to the previous question “Have you visited the My Energy section in LocalLife to learn more about your household electricity consumption?”) what new insights regarding your household electricity consumption
have you learned from the energy feedback (excluding pause hours)?”
Response: “It is nice to compare with neighbours and typical consumption.”
LC Negative This theme addressed the responses where it was evident that LocalLife did not improve the respondent’s neighborhood connections/network.
Example:
MS01:
Question: “Has LocalLife in some way helped you learn to know new neighbours or to know your existing neighbours better (e.g. using features in the app, attending events such as potluck dinners or movie nights)? If so, how?”
Response: “No”
LC Neutral This theme enlisted the individual responses, or the overall depiction from the responses of a respondent, that were uncertain whether the local community-life were influenced.
With regards to socialization, it is important to note that resident’s own inclination to connect to neighbors was the main driver to socialize. To put this into context, LocalLife organized Movie Night for Malvinas residents, aiming to empower neighborhood connections. The residents that did not join the Movie Night because they did not want to socialize, were not influenced by LocalLife’s connection- building event. But they did not participate. So, LocalLife is not accountable.
Example:
MS22:
Question: “Has LocalLife in some way helped you learn to know new neighbours or to know your existing neighbours better (e.g. using features in the app, attending events such as potluck dinners or movie nights)? If so, how?”
Response: “No because I was not available”
In this example the respondent provided a conditional answer, which raises further questions like “had the respondent been available, would they have socialized?”.
These type of answers – involving uncertainty – are coded to Social Neutral just to be safe.
LC Positive An observed straightforward influence on increased local community-life. This definition is broad, but holistically LC Positive serves as a happiness indicator about their community life – entailing social interactions and connections.
Example:
FS30:
Question: “Has LocalLife in some way helped you learn to know new neighbours or to know your existing neighbours better (e.g. using features in the app, attending events such as potluck dinners or movie nights)? If so, how?”
Response: “By organizing corridor events and other initiatives via LocalLife app”
Past energy behavior
Energy saving behavior in the past (prior to joining LocalLife).
This theme gives a good sense of the sorts of measures the respondents had been taking in the past.
Example:
FS33:
Question: “Have you done any other energy saving efforts than the ones listed above (Turned off/unpIugged/avoided to charge electric appliances when not needed (computers, mobile phones, etc; Bought energy efficient home appliances; Set a correct temperature in fridge/freezer and defrosted them regularly; Reduced the use of additional heater; Reduced use of hot tap water). If so, what”
Response: “Did not put hot things in the fridge”
It is to be noted that each of the survey themes were used for midterm and final surveys. The difference between them, however, was that the midterm survey themes were coded to show the status of the respectable theme at that time tracked since the beginning of the signup. The final survey themes are coded to show a progress in the period between midterm and final survey.
3.3.3. Quantitative data
The numerical quantitative data retrieved from the surveys was normalized and then averaged per theme. It is also to be noted that because SS, MS and FS surveys had different number of questions pertaining to each theme, an additional weighting was done at the end (as shown by Equation 1). The following sections list the numericalized survey questions to the respective scales. In these sections the available choices/range for each question are shown in the parathesis after the body of the question.
3.3.3.1. Energy awareness (EA) 3.3.3.1.1. Signup survey (SS)
Table 4: SS(EA): All shortlisted questions (numericalized)
Question Scale
How important do you think it is to save energy in your household?
(Range: “not very important” =1, “very important” =7)
1 – 7
How much knowledge do you have over saving energy in your household?
(Range: “very little” =1, “a great deal of knowledge” =7)
3.3.3.1.2. Midterm and Final survey (MS and FS)
Table 5: MS(EA) and FS(EA): All shortlisted questions (numericalized)
Question Scale
How important do you think it is to save energy in your household? (Range:
“not very important” =1, “very important” =7)
1 – 7
How much knowledge do you have over saving energy in your household?
(Range: “very little” =1, “a great deal of knowledge” =7)
Has the feedback on your energy consumption (excluding pause hours) helped you to better understand your household electricity consumption?
(Range: “not at all” =1, “yes, a lot” =7)
Has participation in pause hours helped you to better understand your household electricity consumption? (Range: “not at all” =1, “yes, a lot” =7)
3.3.3.2. Behavioral status (BC)
3.3.3.2.1. Signup Midterm and Final survey (SS, MS and FS)
Table 6: SS(BC), MS(BC), FS(BC): All shortlisted questions (numericalized)
Question Scale
Have you done anything to save energy in the past? (Never=0, Occasionally=1, Often=2)
Turned off lights when not needed
0 – 2
Turned off/unpIugged/avoided to charge electric appliances when not needed (computers, mobile phones, etc)
Bought energy efficient home appliances
Set a correct temperature in fridge/freezer and defrosted them regularly
Lowered the use of heater if you have one
Lowered the consumption of hot tap water
Would you avoid cooking food and vacuum cleaning once a week during 18:30-19:30 if you knew that it would save you money due to a cheaper electricity price? (Range: “least likely” =1, “most likely” =7)
1 – 7
Would you avoid cooking food and vacuum cleaning once a week during 18:30-19:30 if you knew that it would be good for the environment due to less polluting electricity production? (Range: “least likely” =1, “most likely”
=7)
To what extent do you save energy in your household today? (Range: “not a lot” =1, “a lot” =7)
3.3.3.3. Local Community-life (LC) 3.3.3.3.1. Signup survey (SS)
Table 7: SS(LC): All shortlisted questions (numericalized)
Question Scale
How many neighbours did you know by name where you previously lived? (not including your family)
1 – 15
3.3.3.3.2. Midterm survey (MS)
Table 8: MS(LC): All shortlisted questions (numericalized)
Question Scale
How much do you like to live in Malvinas? (Range: “not at all” =1, “very much” =7)
1 – 7
I strongly identify as a Malvinas resident (Range: “strongly disagree” =1,
“strongly agree” =7)
Do you feel that you can trust the people living in Malvinas? (Range:
“strongly disagree” =1, “strongly agree” =7)
Do you feel safe when you take a walk in the neighbourhood - also during the night? (Range: “very unsafe” =1, “very safe” =7)
How strong are the ties with your neighbours in Malvinas? (Range: “very weak ties” =1, “very strong ties” =7)
Is it common in Malvinas for neighbours to talk to each other when you meet? (Range: “not common at all” =1, “not common at all” =7)
How many neighbours did you know by name where you previously lived?
(not including your family)
1 – 15
How many neighbours in Malvinas do you know by name today (mid-March 2020)?
How often do you spend time with any of your neighbors? (“daily/almost daily” =5; “A few times per week” =4; “A few times per month” =3; “A few times per year” =2, “never” =1)
1 – 5
How many corridor events have you attended at Malvinas since January 2020? (none=0, ‘1-3’ =1, ‘3-5’ =2, ‘more than 5’ =3)
0 – 4
If you did not have a Whatsapp group before the mid of February, do you have the WhatsApp group now? (No=0, Yes=1)
0 – 1
3.3.3.3.3. Final survey (FS)
Table 9: FS(EA): All shortlisted questions (numericalized)
Question Scale
How much do/did you like to live in Malvinas? (Range: “not at all” =1, “very much” =7)
1 – 7
I identify as a Malvinas resident (Range: “strongly disagree” =1, “strongly agree” =7)
How trustable are people living in Malvinas? (Range: “not trustable” =1,
“very trustable” =7)
When taking a walk in the neighborhood, I feel safe (Range: “strongly disagree” =1, “strongly agree” =7)
How strong are your ties with the neighbors in Malvinas? (Range: “very weak ties” =1, “very strong ties” =7)
How common is it for you and your neighbors in Malvinas to greet each other and/or have a quick chat when you meet (e.g., in the hallway)?
(Range: “not common at all” =1, “not common at all” =7)
How often do you spend time with any of your neighbors (i.e., hang-out together or do something together)? (“daily/almost daily” =5; “A few times per week” =4; “A few times per month” =3; “A few times per year” =2,
“never” =1)
1 – 5
How many corridor events have you attended at Malvinas since January 2020? (none=0, ‘1-3’ =1, ‘3-5’ =2, ‘more than 5’ =3)
0 – 3
Has LocalLife helped you to meet new neighbors or get to know your neighbors better? (“no”=0, “Yes, I met new neighbors”=1, “Yes, I got to know my neighbors better”=2, “Yes, I met new neighbors and feel like I know my neighbors better”=3)
How many neighbors in Malvinas do you know by name today (mid-May 2020)?
1 – 20
How many neighbors did you know by name where you previously lived?
(not including your family)
1 – 15
The numericalized responses from the questions are treated in a 3-step process:
Step 1: normalization on a 1-5 range
Step 2: applicable to EA and LC: weighted average into a final thematic score on a scale of 1-5 using Equation 1.
Step 3: applicable to BC: simple average into a final thematic score on a scale of 1-5.
Equation 1: weighted average to obtain the final thematic score
At = 0.5 * ∑ 𝐴𝑛𝑠𝑛1 𝑐,𝑖
𝑁𝑐 + 0.5 * ∑ 𝐴𝑛𝑠𝑛1 𝑢𝑐,𝑖
𝑁𝑐′
Where:
At = final thematic score t = theme (EA, LC)
𝐴𝑛𝑠𝑐,𝑖 = normalized survey responses that are common among SS, MS and FS
𝐴𝑛𝑠𝑢𝑐,𝑖= normalized survey responses that are unique: only present in MS and FS
𝑁𝑐 = total number of common questions 𝑁𝑐′ = total number of unique questions
As seen in Table 4 to Table 9, there were instances where, for example, the score of EA for SS was dependent on 2 questions, while the scores of EA for MS and FS were dependent on 4 questions – out of which 2 questions were repetitions from the SS. As a result, a sound comparison between the scores of EA for SS, MS and FS could not be established, since the scores were calculated from different number of questions (variables). To cater this, weighted averaging (Step 2) was performed, where a 50% weightage was assumed for common as well as unique questions. In our example for the calculation of EA, the common questions are shown in Table 4, whereas the
the same reasoning for the calculation of LC. However, the reason BC is excluded from Step 2 is because the number of questions related to BC (9) is same across all surveys.
3.4. Interviews
In addition to the evaluation of LocalLife ’s influence on household energy consumption and awareness, social well-being and behaviors, qualitative insights were collected through a series of semi-structure interviews with active users of the LocalLife. Interviews took place during February and May of 2020. Considering the ongoing Corona-epidemic, all interviews were conducted and recorded digitally via Zoom, lasted around 30 to 60 minutes and were later transcribed to prepare for thematic content analysis on Nvivo.
Two recruitment approaches were combined: first, at the end of a survey, interested users could choose to be contacted for and participate in a qualitative interview. Second, three series of users were specifically contacted to participate in an interview. These users either (a) very actively participated in LocalLife and Pause Hours, (b) showed average participation in LocalLife and Pause Hours or (c) seldomly used LocalLife and (almost) never participated in pause hours. In total, out of 10 residents who were asked to participate, 9 took part in an interview for which a slightly adapted interview guide focusing on residents’ use of LocalLife and experience with the app, interaction with their neighbors and area, attitude towards energy and the environment and household habits (specifically regarding Pause Hours) was developed. Some of the interviewees had already participated in the first batch of interviews which enabled more intense discussions and allowed to gain deeper insights into user attitudes and behaviors. The interviewee details are listed in Appendix 6.2.A.2.
3.4.1. Framework modelling and coding
The functionality of Nvivo to provide flexibility in coding open-ended questions and building a hierarchical structure of themes, was fully utilized in assigning transcripted interview responses to relevant themes and later analyzing them. The data gathered from the interviews was very diverse and detailed, as the interviewer’s objective was to dive deeper into the concerned topics – for example, the details about the habits the interviewees had developed by participating in the Pause Hours.
Prior to the task of coding, a hierarchical framework was built on Nvivo that could correctly categorize not only the main themes (EA, BC and LC), but also the social psychology reasons behind behavioral change. The motive was to find an underlying explanation – by repetitive reading and granular analysis
