how do you feel?1325718/fulltext01.pdf · to ekman, to have better awareness, because “usually we...

59
HOW DO YOU FEEL? Designing for Emotional Self-Awareness and Perceived Anonymity in an Audience Response System Charlotte Ristiniemi June 2019 Master Thesis, 30 ECTS Spring Term 2019 Master of Science in Interaction Technology and Design, 300 ECTS

Upload: others

Post on 05-Mar-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

HOWDO YOU FEEL?Designing for Emotional Self-Awareness and Perceived

Anonymity in an Audience Response System

Charlotte Ristiniemi

June 2019

Master Thesis, 30 ECTSSpring Term 2019

Master of Science in Interaction Technology and Design, 300 ECTS

Page 2: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Abstract

Humans’ emotions have the ability to take over, which might end up with re-sponses inappropriate to the situation. The solution to inappropriate responsesis to have a better awareness. Great team-work, calm employees, and ratio-nal decision making are all qualities that derive benefit from emotional self-awareness. However, studies show that only 36 percent can identify their emo-tions as they happen. This paper takes on the opportunity to raise emotionalself-awareness by designing a prototype that enables the users to reflect andanonymously share their emotion through an audience response system.

Forty-eight participants, in various group sizes, did within-subjects tests.They started by writing down their answer to the question: How do you feel?.They later answered through the prototype. Whether or not the participantsmanaged to be more specific through the prototype was measured, as well astheir perceived anonymity. The results revealed that the prototype was usefulin both helping the users to learn emotion definitions and further specify theiremotion. In regards to the perceived anonymity, it showed that the design wasfavoring a larger group size around 20 participants.

Keywords — Emotional Self-Awareness, Perceived Anonymity, Audience Re-sponse System, Interaction Design, Information Visualization, Emotion Visual-ization

Page 3: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Contents

1 Introduction 11.1 Mentimeter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.2 Aim and Objective . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Theoretical Framework 42.1 Emotions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.1.1 Basic Emotions . . . . . . . . . . . . . . . . . . . . . . . 42.1.2 Emotional Self-Awareness . . . . . . . . . . . . . . . . . 52.1.3 How to Become Emotionally Self-Aware . . . . . . . . . 62.1.4 Emotion Vocabulary . . . . . . . . . . . . . . . . . . . . 6

2.2 Designing for ESA and Perceived Anonymity . . . . . . . . . . 62.2.1 Interaction Design . . . . . . . . . . . . . . . . . . . . . 62.2.2 Design for Self-Reflection . . . . . . . . . . . . . . . . . 72.2.3 Designing an Emotion Vocabulary . . . . . . . . . . . . 82.2.4 Design for Perceived Anonymity . . . . . . . . . . . . . 8

3 Designing the Concept 93.1 In-Depth Interviews . . . . . . . . . . . . . . . . . . . . . . . . . 10

3.1.1 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . 103.2 Framing the Design Challenge . . . . . . . . . . . . . . . . . . . . 11

3.2.1 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . 123.3 Success Metric . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3.3.1 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123.4 User Journey Mapping . . . . . . . . . . . . . . . . . . . . . . . 13

3.4.1 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133.5 Iterative Prototyping . . . . . . . . . . . . . . . . . . . . . . . . . 14

3.5.1 Iteration One: Layout and Navigation . . . . . . . . . . . 143.5.2 Iteration Two: Navigation and Sharing . . . . . . . . . . . 163.5.3 Iteration Three: Navigation, Copy and On-boarding . . . 17

3.5.3.1 Usability Test Method . . . . . . . . . . . . . . . 183.5.3.2 Usability Test Result . . . . . . . . . . . . . . . 20

3.5.4 Iteration Four: Final UI Design . . . . . . . . . . . . . . . 213.5.5 Iteration Five: Emotion Visualization . . . . . . . . . . . 22

3.6 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243.6.1 Technologies . . . . . . . . . . . . . . . . . . . . . . . . . 253.6.2 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . 26

1

Page 4: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

4 Methodology 274.1 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . 274.2 Final Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

4.2.1 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . 274.2.2 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . 284.2.3 Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284.2.4 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

4.3 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294.3.1 Evaluating the ESA . . . . . . . . . . . . . . . . . . . . . 294.3.2 Evaluating the PA . . . . . . . . . . . . . . . . . . . . . . 29

5 Result 315.1 Emotional Self-Awareness . . . . . . . . . . . . . . . . . . . . . . 31

5.1.1 Contradicting Answers . . . . . . . . . . . . . . . . . . . . 325.1.2 Recurring Answers . . . . . . . . . . . . . . . . . . . . . . 33

5.2 Perceived Anonymity . . . . . . . . . . . . . . . . . . . . . . . . . 345.2.1 Interaction Design and PA . . . . . . . . . . . . . . . . . 34

5.3 Feedback from Participants . . . . . . . . . . . . . . . . . . . . . 36

6 Discussion 386.1 Emotional Self-Awareness . . . . . . . . . . . . . . . . . . . . . . 386.2 Perceived Anonymity . . . . . . . . . . . . . . . . . . . . . . . . . 396.3 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

7 Conclusions 427.1 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Appendices 48A Interview Questions . . . . . . . . . . . . . . . . . . . . . . . . . 48B Design Iteration 1 . . . . . . . . . . . . . . . . . . . . . . . . . . 49C Design Iteration 2 . . . . . . . . . . . . . . . . . . . . . . . . . . 51D Design Iteration 3 . . . . . . . . . . . . . . . . . . . . . . . . . . 52

2

Page 5: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Acknowledgements

I am particularly grateful for the assistance given by my supervisor at Men-timeter, Taylor Plante. His support and help with the design process and de-velopment have been incredibly appreciated. I want to express my very greatappreciation to my second supervisor Niklas Ingvar, CPO Mentimeter, who pro-vided me with valuable insights in Mentimeter, their users, what has worked,and what has failed throughout the years. Niklas’ support and help have beensignificant for many of the design decisions. I wish to acknowledge Emma Klintand Maja Jakobsson for providing help in the design process. My special thanksare extended to all employees of Mentimeter for providing an excellent environ-ment for me to do my research.

A big thank you to David Hellman, Simon Asp, and Hanna Andersson frommy peer-review group who have been very supportive and helpful in my process,and writing. I would finally like to express my very great gratitude to myacademic supervisor Thomas Mejtoft for all guidance provided throughout thesemester.

Page 6: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Chapter 1

Introduction

Humans are emotional beings, and in the digital society we are living in rightnow, it is vital that we do not lose sight of our emotional nature. As renownedemotion scientist, Dr. Paul Ekman states, “It would be very dangerous if wedidn’t have emotions. It would also be a very dull life. Because, basically, ouremotions drive us — excitement, pleasure, even anger” [1]. Nevertheless, ouremotions have the ability to take over, and that might induce responses inap-propriate to the situation. The solution to inappropriate responses is, accordingto Ekman, to have better awareness, because “Usually we are not aware thatwe are emotional until afterwards, when we say something like, ’Oh, I lost myhead.’ Well, you didn’t lose your head; you just lost your awareness of whatyou were feeling at the moment” [1]. Emotional Self-Awareness (ESA) is theability to recognize and understand one’s own emotions and their effects [2].It is not only an ability, but also a tool that is confirmed to benefit decisionmaking [3], employees’ well being [4], job satisfaction [4], academic achievement[5], and team-work [6] to mention some. But ESA is also beneficial in the waythat it simply can help one to cope with their feelings [7]. This reflection, andintrospective, can create a calmer mind while having chaotic feelings.

To get this introspective, naming emotions is one way to go. Naming anemotion can be done by, for example, writing it down or saying it out loud, asthe emotion is occurring. Naming emotions will help to develop an emotionalvocabulary [8]. This simple tool of having an extended emotional vocabularyto specify our own emotions will help us to decrease distress [9] and build ourcapacity to be with and tolerate uncomfortable emotions [8]. The larger theemotional vocabulary is, the better help we get in identifying and regulate ouremotions [10].

Since ESA is of such great value, digital solutions that aim to achieve ESAare starting to appear. One particular study was done to measure how wellusers could reflect on their emotion through a mood tracker application andwhat therapeutic effects the application had [11]. The application was based oncognitive behavioral therapy, and the users reported their emotional states inthe application. The study confirmed the possibility of becoming emotionallyself-aware by using a mobile cell phone application. The study also reportedother benefits like coping abilities and improved mood.

Another project in regards to ESA resulted in the website Atlas of Emo-tions1. This website is exactly what the name suggests, an atlas of human

1

Page 7: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

emotions, or as the Dalai Lama says “a map of our emotions to develop a calmmind”1. The Atlas of Emotions was a realization of the fourteenth Dalai Lama’sidea, made together with Dr. Paul Ekman and Eve Ekman. To actualize thewebsite, designers at Stamen Design helped to turn the research into interactivevisualizations.

The above solutions are tapping into emotional awareness. Moreover, theyraise the question of how we best should design for ESA. Design principles todaymostly focus on making the journey through a user interface (UI) as efficient andintuitive as possible. Ultimately, these principles strive towards optimal flow byremoving all interaction obstacles and smooth the human-computer interactionto the point where users do not have to think about the interactions actively [12].These principles are indeed crucial for a lot of activities, but they are playing onhumans’ automatic response system, and when it comes to reflection, the userneeds to pause, reflect, and then move on. The underlying idea is that reflectivethinking needs principles that prevent autopilot, as current design principlesmight not be the best fit for reflection [12].

Some research has been done to promote conscious thought in digital design.They include design for reflection and design for friction. These principles arefurther explained in section 2.2.2. This paper will argue that, in order to becomeemotionally self-aware by using a digital product, the UI needs these principlesto make users reflect.

When it comes to sharing these emotions, things become more sensitive andcomplex. In today’s society, we are taught that showing emotions is a weaknessbecause they make us open for betrayal [8]. One might think of someone who iscrying as someone who does not have control over their emotions or themselves.One might even think of this person as embarrassing. However, research suggeststhat this comes from the fear of oneself being exposed in public, even justwith another person [8]. Hence, instead of portraying and showing our genuineemotions, it is easier to put up a facade, and pretend that everything is OK.For example, an emotion sharing with our colleagues may be very challengingif the goal is to be truly honest.

One way to minimize the fear is by keeping the sharing sessions anonymous.This can be done with Audience Response Systems (ARS) [13]. They allowboth small and large groups of people to vote on a topic or answer a question,usually, during some sort of presentation, lecture, or meeting. The audiencemembers use their remote device, for example a mobile phone, to respond.After the people in the audience select their response, the results are collectedand presented to everyone visually on the presenter’s screen. There exist ARSsthat will identify a user’s response, but there are also solutions where the usersremain anonymous. The latter solution is a prospect for emotion sharing in thisthesis research because of the desire for individuals to protect their emotionalexpression.

Emotional self-awareness, emotion sharing, and the anonymity from audi-ence response systems open up an opportunity in work meeting environments.Great team-work, calm employees, and rational decision making are all qualitiesthat are sought after in meeting situations. These are all qualities that derivebenefit from ESA. However, it is shown in studies that only 36 percent canidentify their emotions as they happen [14]. This means that only 36 percent

1http://atlasofemotions.org/

2

Page 8: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

of individuals can recognize their emotions and prevent them from bubbling upand possibly affecting the team productivity, the atmosphere in the room, anddecisions about to be made. It would be premature to say that unconsciousemotions are a critical problem in meetings today, but there is absolutely anopportunity to raise ESA to gain the benefits it brings.

1.1 MentimeterThis thesis was written in cooperation with Mentimeter, which is a Stockholmbased company founded in 2014. Mentimeter offers an interactive presentationsoftware, also known as an ARS. This software lets the presenter engage theaudience in presentations like workshops, lectures, meetings, events, and con-ferences [15]. The presenter uses Mentimeter to ask a question to the audience,who respond anonymously, and the final results are shown on the presenter’sscreen. The voting can be done with a computer or tablet, but a mobile cellphone is the most common tool to use when responding. This type of anony-mous voting invites everyone to participate, not only the most extroverted peo-ple. Everyone gets the opportunity to reflect without the stress, nervousness,or potential embarrassment of being judged for their thoughts [16].

As stated, Mentimeter can be used in multiple environments and contexts.This means that the audience size varies a lot. The audience size can rangefrom 2 to 47’544 participants, but the average audience is 20 participants [17].Furthermore, the great value Mentimeter deliver, more than inviting everyoneto participate, is to talk about the results on the presenter’s screen. There is agreat value of bringing everyone’s opinion to the presenter’s screen, but whenyou talk about it, that is where the real value lies [18].

1.2 Aim and ObjectiveIt was stated in the introduction that studies confirm the possibility to becomeemotionally self-aware by using a mobile cell phone application. Hence, theaim of this research is ultimately to gain understanding about how interactiondesign, emotion visualizations, and added design friction can raise emotionalself-awareness by increasing the emotional vocabulary range. Furthermore, theaim is to find out what is needed in order to make people feel confident enoughto share their current emotion in a reflection session in a meeting context.

The objective is to develop a proof of concept that evaluates the participants’emotional vocabulary range and their perceived anonymity while reflecting ontheir emotions and sharing them through an ARS. The concept will be tested ina work meeting context where all participants first write down the feeling theyare having at the moment. Then, they get to reflect on their feelings throughthe ARS concept. Testing how specific the participants are in identifying theiremotions before and after will highlight changes in their ability to differentiateand identify with a better emotional vocabulary range. Finally, the participantswill answer a survey where their perceived anonymity will be measured.

3

Page 9: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Chapter 2

Theoretical Framework

This section presents a theoretical framework based on a literature study. Theframework contain two main parts: 2.1 Emotions, and 2.2 Designing for ESAand Perceived Anonymity.

2.1 EmotionsEmotions are exceptionally complex to examine in research, yet so fundamentalto the human experience. A great example of this is stated in the book Emo-tional Intelligence by Daniel Goleman [7]. The emotions that are released whenlosing a close relative always trigger sadness and despair, but how we choose toshow these emotions are culturally conditioned, for example, if we express orsuppress them. Long before our responses became conditioned, humans livedin very harsh circumstances where direct responses to an emotion could be life-saving. For example, the fear of a lurking predator in the bushes made humansrun and hide. This clearly shows how emotions basically are impulses to action,primitive action patterns that the evolution formed for humans to survive. To-day, when we do not have to listen for predators in the bushes, the usefulness ofour emotional responses are not always as important. This means, according toGoleman, that humans can either utilize their emotions in a situation or reasonabout them.

Nevertheless, our primitive responses still exist, and the stronger an emotionis, the less we use our ability to reason. For example, an overwhelming emotioncan disturb our working memory, or concentration, which makes it hard to thinkclearly on a specific task [7]. Yet, our emotions are essential in our daily lifebecause without emotions, our ability to make rational decisions are vigorouslyaffected [7].

There is a delicate balance here. People need emotions to function, but thereis also the need to be aware that emotions may interfere with rational thought.

2.1.1 Basic EmotionsThe basic emotions that scientists agree about are anger, fear, enjoyment, sad-ness, and disgust [19]. These emotions can be split up in specific emotions, seeFigure 2.1.

4

Page 10: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

This mapping of people’s emotions is used on Atlas of Emotions, and it willlay as a base for this paper as well. The specific emotions seen in Figure 2.1 arelaid out on a scale that shows the relative intensity of the emotions.

Figure 2.1: The basic emotions that scientist agrees about and their specificemotions. This is based on the emotions presented on the website Atlas ofEmotions2.

2.1.2 Emotional Self-AwarenessEmotional self-awareness is the ability to listen into oneself, one’s own emotions,understand them, how they feel, and what effects they have [2]. ESA is a wayto be true to oneself, and when one comes to this realization, a door opens upto become more emotionally intelligent (EI). According to Goleman [2], ESA isone of the learned competences that is nested in the profile of a person’s EI.The essence here is that ESA is a learned competence, i.e., there are methodsthat can help people to become more emotionally self-aware. These methodsare further explained in section 2.1.3. It would seem obvious to know one’s ownfeelings, but people repeatedly have the experience of being unconscious abouttheir feelings in the moment of a situation, only to afterward realize what wasfelt.

Humans can have conscious or unconscious emotions [7]. Unconscious emo-tions can profoundly affect our responses, even if we do not know they exist.Goleman illustrates this well [7]: take the example of a person that is irritatedby something earlier during the day and carries this irritation throughout theday. This person does not know that this irritation still exists and is shortin conversations and acts unpleasant towards people without a valid reason.

2http://atlasofemotions.org/

5

Page 11: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

When finally someone notices this person’s unpleasantness, the person will be-come aware of the underlying irritation and can change the mood. This lastbit of mood changing is called self-management, or self-regulation, and it is aneffect of becoming emotionally self-aware [20].

2.1.3 How to Become Emotionally Self-AwareIn order to become emotionally self-aware, one must spend enough time withtheir emotions by trying to identify how they feel physically, identify wherethey come from, and why they are there [14]. There are multiple ways todo this, but the thorough questions one must ask, and observations one mustnotice, to be in touch with one’s emotions can be quite overwhelming. However,Bradberry and Greaves [14] state that just thinking about self-awareness helpsto improve your skills in awareness and self-regulation. Their research suggeststhat simply labeling the emotion one is experiencing can achieve this [21]. It canhave a therapeutic effect, helping us to feel calm. It can also help us combinethinking and feeling, instead of just responding with our emotions. By namingour emotions, we are checking in with ourselves, which results in decreaseddistress [9]. A simple example of naming emotions is if you are stuck in traffic,get angry about it, and then name your emotion by saying “I am angry becauseI am stuck in traffic” [21].

2.1.4 Emotion VocabularyIf naming our emotions is a way to become emotionally self-aware, there is aneed for a vocabulary. This vocabulary consisting of emotions is just as essentialas the vocabulary you need in order to learn a new language [22]. If you do nothave the vocabulary, it will be challenging to use the language. The other wayaround, a broad emotion vocabulary range will enable the possibility to be morespecific in identifying the emotion that is occurring at the moment. This is whya broad emotion vocabulary range is vital for ESA. Though, the vocabularycannot be seen as an indicator of ESA, but rather something that enables ESA.

2.2 Designing for ESA and Perceived AnonymityWhen designing an application that aims to increase a user’s emotion vocab-ulary range, there are a couple of important facts to think about. The designshould try to make the users actually reflect, but also offer an emotion vocabu-lary that promotes learning definitions of different emotions. Furthermore, thedesign should make the users feel anonymous enough, so they are comfortablein choosing the emotions they have and dare to share it through the ARS.

2.2.1 Interaction DesignInteraction design is basically about designing the interaction that happensbetween users and products [23]. Interaction design should enable the users’objective when using the product. The product can involve elements like aes-thetics, motion, sound, space, and more. A great way to look at interactiondesign is, for example, navigation. Which navigation pattern is the best for

6

Page 12: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

the users objective? Is it to have a scroll pattern or multiple pages? Anothergreat example is the layout, where you choose to put elements like buttons andtext to make it intuitive to the user. There is a suggestion that there are fourdimensions to interaction design [24]. These dimensions concern:

1D: Words and text which for instance can communicate ”tone ofvoice”, represent sections in a menu, or convey the action a button willfire if clicked.

2D: Visual representations such as typography, buttons, diagrams, oricons.

3D: The physical object and environment or space which the usersinteract. This could be a smartphone or computer for the physical object,and the environment could be a crowded subway or a work environment.

4D: Time which represents content that changes over time, such as ani-mations, sound, or video.

With these dimensions in mind, there are a lot of questions an interaction de-signer could ask [25]. Some examples may be:

– What can a user do with their mouse, finger, or a stylus to directly interactwith the interface?

– What about the appearance (color, shape, size, etc.) gives the user a clueabout how it may function?

– What information do you provide to let a user know what will happenbefore they perform an action?

– Are there constraints put in place to help prevent errors?

– What feedback does a user get once an action is performed?

– Are edges and corners strategically being used to locate interactive ele-ments like menus?

– Are common formats used?

2.2.2 Design for Self-ReflectionThe objective when designing the interaction design for self-reflection is thatthe reflection can be seen as the outcome. The goal is to make the user take amoment and reflect. One way to achieve this is by design friction [26]. Whenfriction is designed into a product, it can disrupt the mindless automatic in-teractions. The designed friction can result in a pause which, according to S.Bagnara [12], makes it possible to achieve reflection. The pauses, invoked byfriction, can be explained as a breakdown where the tool the user is interact-ing with becomes impossible to use. An example of this could be a mandatoryon-boarding. The user has to go through the steps in order to use the actualproduct. Friction that results in pauses is often discouraged to reduce the riskof user frustration and disengagement. Although, research suggests that pausesalso opens up the opportunity for the user to distance her/himself from the flow

7

Page 13: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

of interaction [12][26]. If we look at the on-boarding again; maybe the extrasteps were slightly annoying, but in the end made the user understand why, orhow, the product could be used.

This paper will rely on renowned design principles and carefully introducefriction when needed. How this design friction was implemented can be read insection 3.5.3.1.

2.2.3 Designing an Emotion VocabularyAs for the vocabulary, the objective is to make the user discover and learnabout the different emotions, what their label is, and how they are defined. TheUI should be highly usable, and the user should intuitively and efficiently findthe emotions with their associated definition. Hence, there is no need to addfriction. Additionally, more than having a clear navigation pattern, one wayto structure UIs with loads of information is to follow the Visual Information-Seeking Mantra (VISM): overview first, zoom and filter, then details on demand[27].

Overview: Gain an overview of the entire collection.

Zoom: Zoom in on items of interest.

Filter: Filter out uninteresting items.

Details-on-demand: Select an item or group and get details when needed.

How the interaction design was designed in a discoverable and usable way toenable learning can be read in the design process 3.5.1 and 3.5.3.2.

2.2.4 Design for Perceived AnonymityA general definition of anonymity is “the state of being unidentified or unknown”[28]. This will be referred to as actual anonymity. Perceived Anonymity (PA)on the other hand, is something else. PA can be defined as “the extent to whichindividuals perceive that their personal identity is unknown to others or thatthey are unidentifiable as an individual” [28]. Research has shown that PA caninfluence behavior [28], even though the relationships between anonymity andbehaviors are not fully understood.

When sharing an emotion through an ARS, it is vital that the perceivedanonymity is high. It does not matter whether the ARS offers actual anonymitysince the perception of being anonymous is likely better at explaining behaviorsthan the actual anonymity [28]. This means that the design should highlightall aspects that make the users feel anonymous to prevent possible changes inbehavior that can lead them to not sharing their true emotion.

More about the anonymity aspects that became important for this thesisprototype, and how they were implemented can be read in the interview section3.1.1, and iterative prototyping process 3.5.

8

Page 14: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Chapter 3

Designing the Concept

This section thoroughly describes the process of designing and developing theprototype that was used in the final study.

IDEO’s field guide to human-centered design profoundly inspired the designprocess [29], but since the prototype required development, the process wasslightly altered. Nevertheless, the process takes the designer through three mainphases: Inspiration, Ideation, and Implementation. Different methods are usedin every phase. In Figure 3.1, an illustration of the process, and their methodscan be seen. The x-axis illustrates the time, and the amplitude illustrates theconvergence of mind and ideas. This design process enables the designer tofrequently shift gears between highly abstract thinking to concrete observationsor ideas. Furthermore, some methods from Google Venture Design Sprint [30]were used to complement IDEO’s design methods.

Figure 3.1: The design process

InspirationIn the inspirational phase, the designer learns how to understand peoples’ needsand problems better. To truly solve the right problem and fulfill the users’needs, the designer needs to explore the users’ context and environment. Thisis done with methods such as interviews and by framing the design challenge.

9

Page 15: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

IdeationIn this phase, the designer zeroes in on the critical insights gathered from theprevious phase. These insights are then used to generate multiple ideas, findopportunities for design, and finally test and refine these designs. This processutilizes iterative prototyping and refinement methods. By the end of this phase,a final design is decided.

ImplementationThe implementation phase is where the developments start. It is guided by in-quires such as which technologies to use, and how to structure the environmentto make the prototype work best. This phase is the one that differs the mostfrom IDEO’s original implementation phase.

3.1 In-Depth InterviewsIn order to get an overview of the users’ needs and problems, in-depth inter-views were conducted. A total of four people, two women, and two men, wereinterviewed. All of the participants were working in the tech industry, wheremeetings often occur. The full questionnaire can be seen in Appendix A. Thepurpose of the interview was to get a deeper understanding of the followingquestions:

– Do people share their feelings in meetings today and, if so, are they com-fortable sharing them?

– What do people feel about sharing their emotions even though it is anony-mous?

– Can people share their true emotions?

After the interviews, the aim was to find themes [29]. This meant thatkey takeaways of every interview were written down as statements on stickynotes. All the statements were then grouped in themes if they followed similarpatterns or had something in common. These themes would ultimately becomeopportunities for design.

3.1.1 FindingsAs the four interviews were conducted, the answers started to recur and follow-ing themes were identified:

– Intense feelings lead to avoidance: It is hard to share the most intensefeeling regardless if it is anonymous or public.

– Professional persona: It is different to share feelings in a private settingversus in a professional environment. People tend to show the best versionof themselves at work.

– Follow the (happy) herd / Disinclined to spread bad energy: Herdmentality is a well known cognitive bias that can be defined as “the align-ment of thoughts or behaviors of individuals in a group (herd) through

10

Page 16: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

local interactions rather than centralized coordination” [31]. In a profes-sional setting, it may be connected to the fact that people do not wantto spread or be around any bad energy. They do not want to be the onesharing ”negative” feelings if everyone else shared ”positive” feelings.

– Fear that the emotion will not remain anonymous / fear of beingjudged: When people share something anonymously, and everyone seesthe details of that sharing, it might not be as anonymous as one mightthink. If it sticks out (e.g., one person shares that they are angry whileeveryone else shares that they are happy), then it might spark curiosityand people may become curious to find out who shared it. If the sharingbecomes public, people are then afraid they might be judged.

– Team size matters: One person thought it was more comfortable toshare emotions in a small team since you probably know everyone in theroom. Though, another person thought that it is easier in a bigger teambecause it is more likely that one can remain anonymous.

– It is intimidating to show vulnerability: The consensus is that it isfrightening to share emotions.

– The domino effect: If one person starts talking, then a domino effectmay kick in, leading other people to feel inclined to share their feelings.

The key takeaways from the interview were that anonymity is paramount,and showing the label of the emotions on the presenters’ screen might not be wellsuited for anonymity reasons. If the labels are visible on the presenter’s screen,there might be a chance that people figure out the identity behind that emotion.This fear was recurring in the interviews multiple times; “...if you share some redflags, it will spark curiosity” and “...but people are trying to guess. ’I wonderwho is scared in here. That person is very quiet, it is probably that person’. It isnice to say that things are anonymous, but there are probably many thoughts inpeoples’ mind anyway”. If the users feel like their anonymity is uncertain, theymay not be truthful in their sharing. Furthermore, if the emotion label is shownon the presenter’s screen, follow the herd mentality might become prevalent, andthe truthfulness might be compromised yet again.

3.2 Framing the Design ChallengeFrom all the insights from the interviews, the design challenge was defined.Having a clearly defined design challenge was necessary to have a clear direction,organize thoughts about the solution, and in moments of ambiguity, help clarifywhere the design should be pushed [29]. The framing was done by going throughthe following steps:

1. What is the problem you are trying to solve?

2. Try to frame it as a design question, i.e., “How might we...”

3. What is the impact you are trying to have?

4. What are some possible solutions to your problem?

11

Page 17: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

5. Write some of the context and constraints that you are facing.

6. Does your original design question need to be refined or changed?

3.2.1 FindingsTo frame the design challenge, the interviews, previous literature study aboutemotions, and designing for ESA were of great value. The interviews pointed outthat anonymity should be highly prioritized. The theory about emotions anddesigning for ESA helped to point out the importance in designing a vocabularythat promotes both reflection, and learning about the different emotions. Thedesign challenge boiled down to a design question that would lay as a base forthe design direction:

What might a design for an emotional vocabulary look like, that helps peopleto improve their emotional vocabulary range, and that helps people to reflect ontheir emotions, which they truthfully can share anonymously through an ARSwith comfort?

3.3 Success MetricAfter defining the design challenge, a success metric was constructed [30]. Thiswas done in order to specify the goals further, how to measure the results and,help to decide how the final study should be performed. The success metric wasconstructed by answering the following questions:

– Goal: What are you trying to help users do? What problem are youtrying to solve?

– Signal: What change in user behavior or opinion would indicate you havebeen successful in your goals?

– Metric: How will you measure the size of any change in user behavior oropinion?

3.3.1 FindingsThe questions were answered in regards to the insights of the interview and thedesign challenge. The answers to the success metric questions boiled down tothe following:

– Goal: To design an emotion vocabulary that helps people to improvetheir emotional vocabulary range, and that helps people to reflect on theiremotions, which they can then truthfully share anonymously with comfort.

– Signal: (1) People can be more specific when they describe their feelings.(2) People feel anonymous in sharing their emotions through the ARS.

– Metric: (1) Improved emotional vocabulary: i.e., checking if peopleidentified a more specific emotion label after they used the tool. (2)Anonymity: measured with the help from a framework on perceived anonymity.The framework consists of a series of questions where the user can answeron a 7-point Likert scale.

12

Page 18: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

3.4 User Journey MappingUser journey mapping is a common Design Sprint method that attempts toforecast a user’s experience step by step as they encounter the problem spaceor interact with the product [30]. User journey mapping was used to identifywhere possible design opportunities or problem areas may appear.

3.4.1 FindingsThe user journey mapping pointed out two main areas. The first area regardedthe opportunity to introduce design friction. As stated in Design for Self-Reflection 2.2.3, the design friction should not intervene with the interactionof the vocabulary. Therefore, putting the friction as an on-boarding may be apossibility, see Figure 3.2, box 2.

The other area regarded the anonymity aspects. The map pointed out whereit may be possible to make the user feel more comfortable in their anonymity.It was stated in Design for Anonymity 2.2.4, and In-depth Interviews 3.1.1,that the PA is of utmost importance to make the user feel confident in sharingthe true emotion without changing the answer out of fear. Therefore, the usershould feel secure that the label of the emotion will not be submitted, but onlythe visualization. Thus, the visualization should be visible as the user selectsthe emotion in the UI, see Figure 3.2, box 4. Furthermore, the user should becertain that when they click the submit button, only the visualization will beshared, see Figure 3.2, box 6.

Figure 3.2: The journey mapping

13

Page 19: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

3.5 Iterative PrototypingWhen the user problems and needs were identified, the design challenge framed,the success metric defined, and the journey mapping sketched, the iterativeprototyping stage started. This involved five stages where the combined resultbecame the final prototype. This method was inspired by IDEO’s rapid proto-typing method [29]. The idea was to determine what to prototype, create theprototype, test it, and refine it. This was iterated until the design was fit totest the design question in the best way. Figure 3.3 illustrates the loop and itsfive stages.

Figure 3.3: The iteration loop and the five stages

3.5.1 Iteration One: Layout and NavigationSince there were 47 possible emotion labels and five basic emotions, a clear in-teraction design for the navigation structure was essential. It should be intuitiveand easy for the users to find their felt emotion.

The first decision was to use the basic emotions as categories to get anoverview of the emotion collection, and only show the specific emotions after abasic emotion was clicked, i.e. have a two-layer deep navigation. This designproposal was suggested with the VISM (mentioned in section 2.2.3) in mind,i.e., first get an overview. The view where the user picks the basic emotionwas the same for all variations (see Figure 3.4). Out of that, two differentnavigation patterns were designed to go back and forth between the layers. Thetwo navigation patterns were the following:

– Multiple views: When the user clicked a basic emotion, a new viewopened for the specific emotions. Backward navigation was made with a”go back” button, see Figure 3.5

– Expand/Collapse: When the user clicked a basic emotion, the spaceright underneath that button expanded to show the specific emotions. The

14

Page 20: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

specific emotion space collapsed if the user toggled the current selectedbasic emotion, see Figure 3.6

Figure 3.4: The first view of basic emotions

Figure 3.5: Multiple views navigation Figure 3.6: Expand/Collapse

Regarding the layout, a layer variation was made in the Expand/Collapsenavigation idea. The emotion visualization was not visible in the same viewas the specific emotions. Therefore, the visualization was layered on top of theview together with the submit button, see Figure 3.7.

Other layout proposals concerned where to put the emotion visualization.The first design variation placed the visualization on top of the view, becauseof the high priority of the visualization. The other design variation placed thevisualization on the bottom of the screen to keep it close to the submit button,see Figure 2 in appendix B.

All the design variations went through a design critique with three colleaguesat Mentimeter, the chief product officer, an art and visual developer, and a datavisualization developer. The purpose was to evaluate the navigation, the inten-tion with the design, and whether or not the design was within Mentimeter’sdesign guidelines.

15

Page 21: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Figure 3.7: Layered emotion visualization and submit button

First off, a decision on where to put the visualization was made. It was allunanimous that it should be put toward the bottom. Since the visualizationwas the only thing to be submitted, it must be close to the submit button.

The next thing to evaluate was the two different navigation variations. Thefirst variation with Multiple Views was not approved since every question on theuser’s phone should represent one slide on the presenter’s screen. The secondvariation, Expand/Collapse, got the comment that is was too cluttered to keepall the basic emotions visible as one basic emotion was selected.

As for the layered visualization and submit button, it did not follow theMentimeter design pattern for submit buttons, and made the scrollable viewvery tight, and was therefore dismissed.

However, the concept of having the basic emotions as clickable boxes gotpositive feedback. Therefore, it was kept. It was also commented that thesecond variation, Expand/Collapse, was a good idea, but it could be better tofollow the VISM structure 2.2.3 and filter out the unimportant basic emotions.Moreover, a scroll navigation pattern was also suggested.

Another essential detail was to include definitions of the emotions sincenot everyone knows what each emotion means (details-on-demand from VISM).These definitions could assist the user in learning about emotions. The lastcomment was to make the submission of the emotion have an optional tone inorder to make the users feel safe in sharing their emotions.

All views of the design variations can be seen in Appendix B.

3.5.2 Iteration Two: Navigation and SharingBack to the drawing board. Two new navigation variations were designed.One design was a variation of the Expand/Collapse version, but with the VISMstructure applied. This meant that all the basic emotions would filter out as thespecific emotions became visible. Moreover, the emotion description (details-on-demand) would be visible when a user clicked a specific emotion. The secondidea was to have a scroll pattern where more content would be appended belowwhat was clicked.

Both of these designs also had some changes in their copy, and an addedcapability to finish the reflection without sharing the emotion to the presentersscreen. The underlying change was the optional tone of sharing. The idea wasto inform the users that they are ”done reflecting”, and can choose if they want

16

Page 22: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

to share it or not. Three variations were made: Modal (see Figure 3.8), Statechange (see Figure 3.9), and New view (see Figure 3.10).

Figure 3.8: Modal Figure 3.9: State change Figure 3.10: New view

These new designs also went through a design critique. The purpose of thiscritique was to decide which version would be part of the usability study in theupcoming iteration.

Both of the navigation versions received affirming feedback, but the scrollversion got the feedback that it might be hard to create a smooth scroll acrossdevices, and scrolling which is not smooth might confuse users. Therefore, theversion with toggle buttons and filter functionality was chosen.

As for the final response after the users clicked ”I’m done reflecting”, onlythe Modal was dismissed because a duplication of the emotion visualization mayslow down the interface, and because modals in such contexts are unconventionalfor Mentimeter.

See all views of the design variations in Appendix C.

3.5.3 Iteration Three: Navigation, Copy and On-boardingFrom the second iteration where one navigation pattern was decided, smalladjustments were tested in a usability test. The versions included:

– Toggle navigation and multiple selections, see Figure 3.11

– Toggle navigation with clearing button and single selection, see Figure3.12

– Toggle navigation with clearing button, single selection, altered copy, andon-boarding, see Figure 3.13, 3.14 and 3.15

The test that was performed on these prototypes aimed to investigate theusability of the design. The test also evaluated if the on-boarding/friction madethe users more mindful about their interactions. Another topic that was inves-tigated was whether the intention of the reflection was clear since one part ofthe on-boarding explained in short terms why being emotionally self-aware isbeneficial.

17

Page 23: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Figure 3.11: Only toggle and multipleselection

Figure 3.12: Clearing button and singleselection

Figure 3.13: First copy version Figure 3.14: Second copy version

Figure 3.15: On-boarding right before the reflection starts

3.5.3.1 Usability Test Method

Inspiration from Nilsen Norman Group was used to build up the test method[32]. A meeting room was used as the location, and the test was carried outin one-on-one sessions. The question ”How do you feel?” was displayed like aMentimeter slide on a TV screen in the room, see Figure 3.16. The visualizationin the middle was animated such that its edges were slowly moving, like a mem-brane suspended in water. The test had a total of nine participants working in

18

Page 24: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

different departments of Mentimeter. These participants had no prior knowl-edge about the project other than the topic being researched. The participantswere told to perform three tasks:

1. Choose dislike as the emotion

2. Choose hopelessness as the emotion instead

3. Finish the flow like you are done reflecting

Figure 3.16: The question displayed on the TV screen.

As they performed the tasks, they were told to think aloud, which is a com-mon tool to use during usability tests [33]. Thinking aloud is vocalizing one’sstream of consciousness during the test. After the participants performed thetasks, they got to answer questions about the interface:

– What do you think that the abstract illustration was?

– Did you see or read the description of the emotions?

– Did you understand that the ”I’m done reflecting” (or ”Go to next”)button would not send your answer to the big screen?

– What do you think that you were sharing to the big screen on the laststep when you clicked the button ”Yes, I would like to share it”?

19

Page 25: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

3.5.3.2 Usability Test Result

All the documentation from the tests were summarized into bullet points, quotes,or statements. These quotes and statements were then categorized into themes.Every theme had a couple of quotes or statements and a suggested design pro-posal. The following was observed:

– Basic emotion: This regards the screen where all the basic emotionswere positioned, see Figure 3.4. Some participants stated that these emo-tions are quite negative and intense. But after clicking on a basic emotion,everyone understood that the basic emotions only were categories. There-fore, they stated that they wanted to have some prior indication of thespecific emotions. The new design proposal was that there should be anindication that more will come after clicking a basic emotion.

– Toggle navigation and toggle navigation with clearing button:This regards going back and forth between the basic emotions and thespecific emotions, see Figure 3.11 and 3.12. The conclusions were that thetoggle navigation pattern was hard to learn but easy to remember. Whenthe clearing button was there on the other hand, it was more intuitive thefirst time, but some people suspected the clearing button would canceleverything. The new suggested design proposal was to keep the possibilityof toggling the basic emotions, but that a back button is also needed. Inaddition, the basic emotions should be selected at all times, even if aspecific emotion is selected.

– ”I’m done reflecting” and ”Go to next” button: Mentimeter users,and perhaps software users more generally, are used to one-click submit,and 50 percent clicked this button on autopilot. Starting with the par-ticipants that got the button copy ”I’m done reflecting”: If they clickedit, they often thought later on that they had submitted the reflection,i.e. the word and the visualization. However, if they read the text underthe button, they could see that this action would not submit anything,which made them say, “What will happen?”. The participants that gotthe button copy ”Go to Next” were also confused where the action mighttake them. The new design proposal was either to change the design ofthe button or have two buttons under each other were one is ”Submit”and one is ”Finish without submitting”.

– ”Yes I would like to share it” and ”Yes, submit” button: Whenthe button had the word ”share” in it, almost everyone thought they weresharing to Facebook or another social platform. One hundred percentthought that this action would submit the visualization, but the peoplethat did not read the small text (around 50 percent), thought that the ac-tual word would be submitted as well. People generally liked the optionaltone of submitting; “I can focus on my reflection”. The new design pro-posal suggested to use the word ”submit” even though it is quite formal.Another suggestion was to make the button tell what will be submitted,which is the visualization.

– Emotion description: The texts used got the feedback that it “promoteslearning”, “I want to see what these emotions mean” and “good as a

20

Page 26: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

guideline”. Around 50 percent read or saw this description. These emotiondefinitions were kept the same.

– Emotion visualization: Some quotes of interest were; “if I select anemotion, it will probably change the visualization” and “the visualiza-tion represents the emotion”. However, some could not easily understandwhat the colors and shapes represented in detail. Nevertheless, everyonethought it represented the emotion in some way, which also created a senseof connection to what was showing on the TV, and therefore implied whatwould be shared. A good portion of the participants also thought thattheir submission would morph the visualization on the TV. The visualiza-tion sparked curiosity because the likes of it had never been seen before inMentimeter’s products. Since mostly positive data were collected aboutthis visualization, the conclusion is that for something so different, it canremain unchanged because there is little precedent on how to visualizeemotions. The confusion may partly be because the users have never seenanything like it before.

– “Why should I do this?”: “What is the reason for doing this, or evenchoosing a specific emotion, if it will not contribute to the big visualiza-tion?” and “Why am I doing this in the first place?” were two quotesabout the UI design and whole reflection scenario. This skepticism maycome from the users not understanding how beneficial it is being emotion-ally aware. These participants did not get to test the on-boarding version.The feedback suggests that it may be useful to have a small on-boarding,telling the user why emotional awareness is important.

– Design friction/on-boarding: Two participants got to test the designwith an on-boarding that had three screens with information. It washard to decide if the on-boarding enabled reflection for the users, but oneinteresting thing happened. “Being aware of your emotions can calm yourmind” was written on one screen. One of the participants stated, whilechoosing a specific emotion, “The text in the beginning makes me realizethat choosing a specific emotion is good for me”. This highlights the valuethe on-boarding can give the users that questions why they are even doingthe reflection. A design proposal may, therefore, be to possibly give thisinformation on the presenter’s screen or to keep a short on-boarding. But,the feeling was that having three screens with text was too long becauseit will always stay the same for everyone, and there is no realistic scopeto track anonymous users and tailor a new on-boarding every time.

3.5.4 Iteration Four: Final UI DesignAfter mapping out the information from the usability test, a final design wascreated from the suggestions from the usability test, see Figure 3.17. This designversion went through a design critique like the versions before. It was agreedthat a short on-boarding should be included with multiple small statementsthat could help the user understand why this reflection is of such value. As forthe basic emotions where the users wanted some indication of categories, smallarrows were included in the box. The previous text above the basic emotionsstating I feel..., was changed to I feel some level of..., to further direct the users

21

Page 27: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

to understand that they are not choosing their first thought of, for exampledisgust, which may be quite the intense feeling. The choice to use two buttonsin the end was chosen for the reason to follow the ”one click-submit” pattern.The copy on the submit button was also changed to ”Submit visualization” tofurther imply what would be shared.

Figure 3.17: The final UI design.

3.5.5 Iteration Five: Emotion VisualizationA representation of the emotions was needed since no labels should be shared.However, the visualization did not have to be descriptive enough to be un-derstood by itself. The importance of the visualization was to make the userunderstand that this visualization is what will be shared. Moreover, togetherwith the defining text of the emotion, the visualization should add the feelingof intensity of the specific emotions. The emotion visualizations were designed

22

Page 28: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

with a couple of variables in mind:

– Basic emotion category

– The intensity of specific emotion

Atlas of Emotions1 came to be a big influence for the final result. In theiremotion visualizations, they worked with color, shape, and animation. Thecolor, type of animation, and shape differentiated the basic emotions. Theintensity of the specific emotions was visualized by a diagram, where the mostintense emotions were high and placed to the right, and the least intense weresmall and placed to the left.

The color, size, and animation attributes were taken into consideration, anda framework was built up for the prototype in this research. Color came todifferentiate the basic emotions and their specific emotions, see Figure 3.18.Speed of animation and size of visualization came to represent the intensity ofthe specific emotions. A fast animation and a bigger size represented an intensespecific emotion. Furthermore, two new intensity attributes were added to thevisualization. They were: amount of layers and opacity. All these intensityattributes can be seen in Figure 3.19. The final decision was to let the emptystate be without color, and only have a grey border in the same uneven circleshape.

Figure 3.18: The colors that differentiated the basic emotions.

Early in the design process, a placeholder for the emotion visualization wasdesigned very quickly, see Figure 3.8 or 3.17 for reference. This placeholderwas used for the user testing so that the test participants could see somethingthat might be close to the result. This placeholder visualization had differentshapes for the basic emotions, for example wobbly for disgust, blurry for sadness,and spiky for anger. These shape attributes were later disregarded since theyonly enhanced the emotion in the reflection state when the user was browsingemotions. However, if all these different shapes would be shared to the bigscreen, the randomness might be too great, resulting in a disrupted rhythm.Therefore, the uneven circle shape was decided to be uniform throughout theemotions.

As for the presenter’s screen, the visualization was decided after a designcritique. A couple of examples were showed, and the outcome was to have theempty state looking similar to the empty state in the users’ phone. This meant

1http://atlasofemotions.org/

23

Page 29: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Figure 3.19: The intensity attributes for the specific emotions.

a slowly animating uneven circle with a border, see Figure 3.20. Additionally,as the emotions pop onto the screen, they would be kept inside of this circle,floating around the space, see Figure 3.21.

3.6 ImplementationThe purpose of implementing the design was to enable the test participantsto reflect with the help from their own phone. When they would submit theemotion, it would pop up on the presenter’s screen, which inevitably would feelmore realistic then faking the visualizations. The realness would be vital whenevaluating the perceived anonymity.

24

Page 30: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Figure 3.20: The empty state on the presenters screen.

Figure 3.21: The presenters screen filled with submitted emotions.

3.6.1 TechnologiesThe front end web application was implemented using ReactJS2, which is aJavaScript library for building user interfaces. The visualizations were builtusing D33, which is a JavaScript library for manipulating documents based ondata. Mentimeter’s design system Ragnar was used to implement the design.The back end was built using NodeJS 4, which is an open source JavaScriptserver environment. Since the application needed to handle real-time requests,

2https://reactjs.org/3https://d3js.org/4https://nodejs.org/en/

25

Page 31: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

a web socket solution was built with SocketIO5. The final solution was deployedon Heroku6, which offers a platform where you can deploy and host your appli-cations.

3.6.2 LimitationsBecause of the time limitation, the visualizations on the presenter’s screen lackedsome detail. The idea was to use the same visualization as in the users’ reflectionview. The complexity of implementing these visualizations became too great,and the final implemented result was a simple circle that had different colorsdepending on their basic emotion, and different opacity and size depending onthe intensity of specific emotion, see Figure 3.22.

The rest of the UI came to look like the final proposed design.

Figure 3.22: The final implemented visualization on the presenters screen.

5https://socket.io/6https://heroku.com/

26

Page 32: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Chapter 4

Methodology

The methodology used in this thesis can be divided into three parts: (1) theliterature review which included the topics covered in the theoretical framework,(2) the design process that explained how the proof of concept was designed anddeveloped prior to the final study, and (3) the final study which explained howthe study was performed. The design process was fully described in Designingthe Concept, section 3.

4.1 Literature ReviewThe introductory work consisted of an extensive literature study that becamethe theoretical framework of the research. This literature study was conductedto get more knowledge in the field, especially about what emotional self-awarenessis, and how one can become emotionally self-aware. The research also exam-ined what kind of emotions researchers agree to exist, i.e., the basic emotions,and what specific emotions they all have. This knowledge was used to createthe emotional vocabulary in the prototypes. The literature study also includedanonymity and interaction design. This research was relevant to design foranonymity and reflection.

All the articles, books, reports, websites and previous investigations wasfound through reliable sources such as Google Scholar, the Royal Institute ofTechnology’s library, or databases with articles that Umeå University studentshave access to.

4.2 Final StudyThe final study aimed to evaluate the participants’ emotional vocabulary range.Furthermore, the test aimed to measure the participants’ perceived anonymity.

4.2.1 ParticipantsSince the test environment was meetings at work, the test participants wereselected and recruited by acquainted persons of the author. These people thenreached out to their colleagues who all chose to participate voluntarily. Noincentive was given to participate in the test. To get a greater sample size,

27

Page 33: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

employees at Mentimeter with little or no prior knowledge about the thesisproject also got to participate in the test. The results from Mentimeter’s em-ployees were weighted against the other test groups to see if there were anygreat differentiation between the groups. More about this in the result chapter5.

A total of 48 participants performed the test, 22 women, 25 men, and onewho preferred not to specify gender. The age span was predominantly in thespan of 25-34 with 41 participants. The rest were evenly distributed between18-24 and 35-44 years old. The distribution for how long the participants hadworked at their current company was balanced between 0-2 years. Only fiveparticipants had stayed with the same company for longer than two years.

4.2.2 MaterialsThe materials used by the test participants were a smartphone, a pen and apiece of paper. The smartphone was used to access the web-based ARS proto-type. The researcher used a big screen (TV or projector) to present and showthe prototype view where the emotions would pop up after the participantssubmitted their reflection. Finally, a web-based survey was used to collect theinformation from the test.

4.2.3 DesignThe study was performed in a group setting. Five of the groups contained fourto six people, and the study performed at Mentimeter contained 23 people.

In the study on emotion vocabulary range, the dependent variable was theemotion vocabulary range while the independent variable was the tool used toreflect. The tool was either a pen and paper or the prototype. Other variableswere age, sex, and amount of time worked at the company. A within-subjects de-sign was applied to evaluate the dependent variable [34]. This meant that everyparticipant tested both versions, i.e., first pen and paper, then the ARS proto-type. The test design and method was decided after personal communicationwith Paul Ekman Group [35].

For the evaluation on perceived anonymity, the dependent variable was PAwhile the independent variables were age, sex, and amount of time worked atthe company. The control variable was the design of the prototype. The PA wasmeasured by a context-independent instrument [28]. The instrument consistedof five questions with responses ranging from strongly agree to strongly disagreeon a 7-point Likert scale. The two first questions were constructed in a waythat if the participants answered with a high number (strongly agree), it wouldmean that they perceived their anonymity as high. The three final questionswere reverse coded, meaning that a low number (strongly disagree), would meanthat they perceived their anonymity as high.

4.2.4 ProcedureThe procedure of the test can be divided into two parts. The first part was thereflection, and the second part was answering a survey.

The reflection started with asking the participants to write down their an-swer, in one word, to the question ”How do you feel?”. Afterward, they were

28

Page 34: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

asked the same question, but this time through the ARS prototype. The par-ticipants used their smartphone to access, and reflect with the help the of theprototype. The participants answered by clicking on the emotion they felt, andafter submission, the visualization was shown on the presenter’s screen.

As for the survey, the participants started by filling in their informationabout age, gender, and how long they worked at their current company. Thenthey moved on to answer what emotion they wrote down on the paper andwhat they answered on the prototype. The participants also filled in what theythought they were sharing to the presenter’s screen, and if they clicked thebutton ”Submit visualization” or ”Finish without submitting”. The final partof the survey was to go through the questions on perceived anonymity.

Both during and after the study, some participants wanted to air theirthoughts about the experience and give comments. These statements, quotes,and comments were written down and saved as additional feedback.

4.3 EvaluationThis section describes the methodology of evaluating and analyzing the results.

4.3.1 Evaluating the ESASucceeding in answering with a more specific emotion was defined by first statinga mood, feeling, physical sensation, or basic emotion that could be classifiedwithin a particular basic emotion, and later stating what research suggests ismore specific within that same basic emotion. See table 5.1 for which answerswere classified as more specific after.

The results that were classified as ”already specific” were the ones that im-mediately answered with a specific emotion. See table 5.3 for which answerswere classified as already specific.

Failing in answering with a more specific emotion was defined by answersthat were hard to classify, or by contradictory answers (e.g., blissful followed byfrustration), see section 5.1.1 for the definition of contradictory answers. Table5.2 displays the answers that were classified as failures.

4.3.2 Evaluating the PAThe PA was first evaluated using a Levene’s test on equal variances with asignificance level of α = 0.05. Every question’s answers from the PA instrumentwere the response, and the factor was whether or not the response came froma Mentimeter employee, i.e., small group size or large group size. This test wasdone in order to decide which ANOVA test could be performed [36].

Then a Welch’s ANOVA t-test [37], with a significance level of α = 0.05, wasindividually performed on every question’s answers from the PA instrument.The key factor was the same as in the Levene’s test. The null hypotheses werethat the means would be equal in both groups. This would indicate whether ornot being a Mentimeter employee, or being in a large group size mattered.

Finally another Welch’s ANOVA t-test, with a significance level of α = 0.05,was individually performed on every question’s answers from the PA instru-ment. This time, the key factor was is the participants thought that they only

29

Page 35: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

shared the visualization, or something else. However, the main groups were stilldivided. The null hypotheses were that the means would be equal in both sit-uations, meaning that the PA anonymity would not change regardless of whatthe participants thought that they shared.

The Welsh’s ANOVA t-test was performed since the data was normally dis-tributed and violated the assumption of homogeneity of variance [37].

30

Page 36: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Chapter 5

Result

The results were divided into two sections, ESA and PA. The results for ESAused the combined data from both Mentimeter employees and the other com-panies involved in the study. This decision was based on the assumption that itshould not matter if you are in a large sized group, or work at Mentimeter whenyou reflect on your emotions. As for the PA, the results were divided becauseof the vast difference in group size, and that Mentimeter employees may havea better understanding of both the actual and technical anonymity that theproduct offers. Figure 5.1 shows how the groups were divided.

Figure 5.1: The amount of participants in their respective groups.

5.1 Emotional Self-AwarenessOut of the 48 responses, only one data point was removed since this participantfailed to fill out the form correctly. This participant filled in ”Submit” as theanswer to what emotion was submitted.

From the valid 47 responses, the results showed that 24 participants suc-ceeded in answering with a more specific emotion through the prototype, com-pared to what they wrote in the paper beforehand. Ten participants alreadyanswered a specific emotion before using the prototype. Thirteen participantsfailed in specifying a more specific emotion through the prototype.

31

Page 37: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Succeeding in Being More SpecificBefore After Before AfterBubbly Excitement Happy ExcitementCalm Peace Happy Sensory pleasureCalm Sensory pleasure Happy AmusementContent Peace Happy AmusementContent Peace Happy PeaceContent Excitement OK / Tired PeaceCurious Excitement Relaxed PeaceCurious Peace Stressed NervousnessCurious Pride Stressed HelplessnessGreat Peace Stressed FrustrationHappy Excitement Stressed FrustrationHappy Peace Stressed Anger

Table 5.1: Answers that succeeded in being more specific by using the prototype.

Already Specific AnswersBefore AfterAnnoyed AngerExcitement ExcitementExcitement ExcitementExcitement WonderExcitement ExcitementExcitement ExcitementExcitement 80%, stressed 20% RejoicingWorried DiscouragementWorried NervousnessApathetic Annoyance

Table 5.2: Answers that were specific on the first go.

Failing in Being More SpecificBefore After Before AfterAlright Hopelessness Hungry AmusementAngry Enjoyment Pondering WonderWorried Peace Slow DisappointmentConfused Annoyance Giddy-brained NervousnessContent Nervousness Stressed PeaceFocused Enjoyment Tired / Slow AnxietyHappy Anxiety

Table 5.3: Answers that failed in being more specific by using the prototype.

5.1.1 Contradicting AnswersSeven participants gave contradicting answers. Answers that were counted ascontradicting switched between basic emotion categories before and after.

32

Page 38: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Contradicting AnswersBefore AfterStressed PeaceContent NervousnessWorried PeaceHappy AnxietyApathetic AnnoyanceAngry EnjoymentAlright Hopelessness

Table 5.4: Answers that were contradicting before and after using the prototype.

5.1.2 Recurring AnswersA couple of answers written before using the prototype were recurring (see Table5.5). Participants who wrote down the same emotions in the first stage, laterdiverged in the second, e.g. two participants answered stressed beforehand,and then diverged by answering frustration and hopelessness. Some recurringanswers were also contradicting; these were removed from the recurring answerstable.

Recurring Answers# Before After # Before After1 Happy Excitement 15 Excitement Excitement2 Happy Excitement 16 Excitement Excitement3 Happy Amusement 17 Excitement Excitement4 Happy Amusement 18 Excitement Excitement5 Happy Peace 19 Excitement Wonder6 Happy Peace 20 Calm Peace7 Happy Sensory pleasure 21 Calm Sensory pleasure8 Stressed Frustration 22 Curious Excitement9 Stressed Frustration 23 Curious Peace10 Stressed Nervousness 24 Curious Pride11 Stressed Helplessness 25 Content Peace12 Stressed Anger 26 Content Peace13 Worried Discouragement 27 Content Excitement14 Worried Nervousness

Table 5.5: Recurring answers before using the prototype and their answers after.

33

Page 39: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

5.2 Perceived Anonymity

Perceived AnonymityMean Values of Questions (Q)

Group Size Q1 Q2 Q3 Q4 Q5Large 5.87 5.57 3.35 2.97 2.87Small 4.52 4.96 4.04 4.12 3.68

Table 5.6: The mean values of the different groups from the PA questions (Q).

The Levene’s test on equal variances got a probability of p > 0.05, meaningthat the null hypothesis could not be rejected. The variances were not equal.

The results from the Welch’s ANOVA t-test, that investigated if the PAdifferentiated between the group sizes, showed that only question one (I amconfident that others do not know who I am), was significant with a probabilityof p = 0.0054. The rest of the questions showed no significance between groups.However, question four; (others are likely to know who I am), got p = 0.0680,which is not significant, but bordering to. Question four was reverse coded fromquestion one.

5.2.1 Interaction Design and PAThese results are from the survey where the participants answered questionsabout the prototype and their thoughts; What did you think that you wouldsubmit to the main screen? and Which button did you click in the end?.

Figure 5.2: The amount of participants in their respective groups that thoughtthey would only share the visualization.

Forty-three percent of the Mentimeter employees, and 68 percent of the othergroups thought they would share only the visualization, see Figure 5.2.

34

Page 40: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

The biggest misconception was that either only the specific or basic emotionlabel would be submitted to the main screen. The biggest misconception fromMentimeter was that the specific emotion label would be submitted (five partic-ipants), and from the other companies, the basic emotion label was the biggestmisconception (four participants).

Furthermore, the last Welch’s ANOVA t-test, with the key factor beingwhether the participants thought they only shared the visualization, showedno significant difference in the large group. However, from the smaller groupsizes, question 3; I am easily identified as an individual by others, and question4; others are likely to know who I am, showed a significant difference in PAwhether or not the participant thought they shared the visualization, see Figure5.3. The figure displays that the participants who believed that they only sharedthe visualization were more likely to feel anonymous.

Figure 5.3: PA answers on question 3 (left) and 4 (right) from the small groupsizes, depending on what they thought they shared.

Regarding the last button, almost everyone chose to submit their answer tothe presenter’s screen, see Figure 5.4. Only one participant in each group choseto click Finish without submitting.

35

Page 41: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Figure 5.4: The amount of participants that clicked Submit visualization orFinish without submitting.

5.3 Feedback from ParticipantsThe feedback during, and after the study concerned many different aspects.Four quotes collected during the study were:

– “Will we talk about this later?”. This was said while the participant wasnavigating the prototype. This participant was actively hiding his screenfrom the other participants in the room.

– “Can I submit more than one?”. This was said while the participant wasnavigating the prototype.

– “I am learning a lot of new words, this is perfect”. This also was said whilethe participant was navigating the prototype.

– “Who is angry?”. This was said as a red emotion visualization was appear-ing on the presenter’s screen. The tone was curious rather than harsh.

After the study, some participants aired their thoughts. One participantstated that he only clicked submit visualization because it was a test. He thencontinued by stating that if it was not a test, he would not have clicked thatbutton since he thought that the name of the emotion would appear on thescreen, which felt scary. When he realized that only the visualization appearedon the screen, he said that he changed his mind directly and that he felt enoughanonymity to always submit his answer.

Some groups discussed how they could use this tool in their daily work, andhow great it would be. One group thought this tool would be perfect to useevery morning to have a check-in. They said it would be useful because youcould share whatever you were feeling, see the different colors, and understandthat you are not alone, even if everyone is saying that they are fine. To take itto the extreme they gave the example; if everyone shared a ”red” emotion, thenyou could take immediate action, “Okay, let’s have a break before we continueworking”.

36

Page 42: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Another group wanted to use the tool to measure how employees feel in asprint. They argued that you could track the activities in a sprint and correlatethem with the emotions to identify trends in vulnerable, or uplifting activities.

A small group of four people stated after the study that it was easy to seewho submitted what because as someone submits, they look up to see theiranswer, and then you can easily see what that person feels.

37

Page 43: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Chapter 6

Discussion

The discussion includes three sections, ESA, PA, and Limitations. Each sectiondiscusses their respective underlying research theory and results relative to theobjective of the research: to evaluate the participants’ emotional vocabularyrange and their perceived anonymity while reflecting on their emotions andsharing them through an ARS. The aim is also discussed: how can interactiondesign, emotion visualizations, and added design friction raise emotional self-awareness by increasing the emotional vocabulary range and what is needed inorder to make people feel confident enough to share their current emotion in areflection session in a meeting context.

6.1 Emotional Self-AwarenessSince the evaluation of the emotions was classified manually, no statistical testswere performed on this data. It might be misleading to state that somethingwas significant when the outcome could have been different if another personclassified the emotions. Hence the results were used as mere guidance in howthe prototype performed.

The findings showed that over 50 percent of the participants could be morespecific using the prototype, suggesting that their vocabulary range may be in-creased. It is not guaranteed that the participant will remember the name ofthe emotion, or the definition. What it does indicate, is that the participantswere enabled to identify a more specific emotion with the prototype. However,since the classification of participants being successful in specifying their emo-tions were manually decided, there is an error margin that should be taken intoconsideration. The results were decided from the author’s learned experienceand research about emotions, which means that it may be different if anotherperson decoded the responses.

Furthermore, one factor that helped the participants in being more specificmay have been the fact that the UI design provided a discoverable vocabularythat enabled learning. One of the participants clearly stated that he learned alot of new words. This is helpful when trying to identify a specific emotion thatmay not be stored in the individual’s personal vocabulary from before.

Ten out of 47 participants (21 percent) were already responding with a spe-cific emotion on the first go. This seems to affirm what was stated in the

38

Page 44: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

introduction, that only 36 percent of individuals can identify their emotions asthey happen. This also highlights the importance of a tool like the one createdin this research.

Regarding the participants who failed in identifying a more specific emotionthrough the prototype, one aspect is of interest. Seven participants respondedwith contradicting answers before and after. I would like to argue that this maynot be a failure in itself, but instead emphasizes the fact that multiple emotionsmay occur at the same time. The current design solution only offered theparticipants to pick one emotion which may conflict with what the participantsare actually experiencing.

Another interesting fact was the recurring answers. Participants who forexample, answered stressed or happy before, had divergent answers after. Thisbrings attention to the fact that people perceive the same feeling differently,even with contradictory basic emotions, e.g. stress to anger, and stress to hope-lessness (sadness).

6.2 Perceived AnonymityAs stated in the method, if the two first questions got high numbers, it wouldimply high perceived anonymity. Conversely, if the three last questions gothigh numbers, it would imply low perceived anonymity. The desired responsesshould, therefore, be high numbers for the first two questions, and low for thethree last questions. This also means that if a participant is consistent, theirresponse numbers should follow this high/low pattern. However, some responsesfrom the survey did not make sense, e.g., high answers on all of the questions.This may imply that it was hard to understand and answer the questions. Thevariance test could not affirm equal variances between the groups, which meansthat each group were not coherent in their anonymity. In spite of that, theanonymity was perceived high in general, rather than low.

Mentimeter perceived their anonymity significantly higher on one question.Furthermore, that question had a reverse coded question which was borderingsignificant. Even though not all answers were significantly different, it denotesthat either a bigger group size and/or working at Mentimeter is a factor thatcan influence the perceived anonymity. The large group size may as well makethe users feel like one in the crowd, which makes it harder to figure out whosubmitted what. It must be pointed out that it is impossible to determine whichfactor may be the greatest; group size, or working with the product every day.However, the size of the audience seems to be influential, and the responsesshould possibly not show in real time as the users submit their answer. Men-timeter currently offers the ability to hide/show answers, and this could havebeen implemented in favor of the perceived anonymity. What speaks againstthat is the fact that the average Mentimeter audience size is 20 people, whichis closer to the large group than the small group sizes studied in this research.A suggestion would be to let the presenters decide how they want to show theresults: in real time, or all at once.

The color of the visualizations should be discussed as well. Feedback, ob-servations, and quotes from the participants suggested that the color mattered.One participant was hiding his screen, one participant curiously asked who wasangry as a red visualization appeared, and one group discussed the fact that a

39

Page 45: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

small group could easily figure out who submitted what by merely looking atthe color and the people in the room. These are all concerns that influence theperceived anonymity. The fear of one’s basic emotion being exposed could in-crease if an audience draws conclusions about the colors and comments on them.There was a participant who expressed his fear of talking about the emotionsafterward, which exhibit this fear. The fear of being exposed, or expressing emo-tions publicly, is coherent with previous research and the interviews conductedin the design process. One could ask, should the colors be removed? I would liketo argue that the colors may provide valuable feedback for the audience. Twogroups support that. Colors are providing the ability to track the emotionaljourney a group experience during a sprint. Moreover, if there are a lot of angeremotions, some action could be taken from a company perspective, e.g., takinga break. My assumption is conclusively that if a group continuously use thereflection tool, they will learn to feel more secure in sharing.

One interesting fact to connect back to the interviews was the decision toremove the labels of the emotions. This decision was made on the presump-tion that it will make the user feel more confident in their anonymity. Theparticipants in small group sizes who believed they only submitted the visual-ization confirmed this presumption. On the other hand, this significance wasnot observed among the Mentimeter employees who believed they only sharedthe visualization.

There was a vast difference in percentage regarding what the different groupsthought they submitted to the main screen. Forty-three percent of Mentimeteremployees, and 68 percent of the other groups thought they submitted only thevisualization. This dissimilarity may have risen from the fact that it is uncon-ventional in the Mentimeter product to submit something different than whatwas selected. The participants in the other groups presumably have less expe-rience using Mentimeter and learned the design as they were using it, unlikethe Mentimeter employees who already have their preconceptions of how theproduct works. This suggests that the design was communicating what wouldbe submitted. However, the button stated ”Submit visualization”, which com-municated what will be submitted, but conventional submit patterns suggeststhat the user will submit what was selected, i.e., the text. The unconventionaldesign used in the prototype may be the reason that the overall results werequite low.

Finally, everyone except one participant in every group clicked ”Submit visu-alization”. One participant in the test clearly stated that he only clicked ”Submitvisualization” for the fact it was a test. Thus, the suggestion is that the highfrequency of submissions is that the test scenario biased the participants. Thefrequency may be higher in a real scenario.

6.3 LimitationsAt the beginning of the research and design process, there was a vast initiativeto include design friction to make the users detach themselves from autonomousinteractions to promote mindful reflection. However, it would have been verydifficult to measure whether or not the users actually became more mindful sincethe result of mindful interactions would be that the users actually reflected.The on-boarding was an extra step, and a very atypical design for Mentimeter,

40

Page 46: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

therefore it may have disrupted the autonomous behavior, but, there was no wayto actually measure whether or not the users actually reflected, or just pickedan emotion. Therefore, the on-boarding mostly served as an introduction towhy ESA is important.

41

Page 47: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Chapter 7

Conclusions

The aim was to gain understanding about how interaction design, emotion vi-sualizations, and added design friction can raise emotional self-awareness byincreasing the emotional vocabulary range. Furthermore, the aim was to findout what was needed in order to make people feel confident enough to sharetheir current emotion in a reflection session in a meeting context. The objectiveof this research was to evaluate people’s emotional vocabulary range and per-ceived anonymity as they were reflecting on their emotions and sharing themthrough an ARS.

In summary, this paper argued that emotional self-awareness could be learnedand achieved by naming emotions. Therefore, an emotion vocabulary was de-signed in the form of an ARS. The users and prototype were studied in a groupmeeting environment where the users should identify their current emotion. Ifthey wanted to, the users could choose to share their emotion, as an abstractvisualization, to the group. This study of the research suggested that the pro-totype was a useful tool for the users in both helping them to learn emotiondefinitions and further specify their emotion. Additionally, the study affirmedthat multiple emotions could occur at the same time. This proposes a changein the design. The tool should preferably let the user choose multiple emotions.

Regarding anonymity, previous research suggested that perceived anonymitymay change behavior. This was not observed. However, group size and divergentthoughts of the design proved to influence the perceived anonymity. A largergroup size seemed to improve the perceived anonymity. With reference to theresults of the study, sharing abstract visualizations could make people feel moreanonymous than sharing the label of the emotion. Finally, as the color of thevisualizations were exposing the basic emotions, it showed to be easier to figureout who was feeling what, since there were fewer people to analyze. However,colors yet provide valuable feedback for the group and the presenter, while stillkeeping the most private emotions anonymous.

The results of this study can be seen as a mere indication that the designof the tool was useful. But the big question remains: did the tool increaseemotional self-awareness?

42

Page 48: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

7.1 Future WorkThe method to evaluate the participants’ vocabulary before and immediatelyafter using the prototype was chosen because of the time limitation. It would beinteresting to track users’ vocabulary for a longer time to see if their independentvocabulary would increase. This could provide a better indication of ESA.

Talk about it is one of the most significant values Mentimeter brings. Oneway to look at the research done it that it is half circle. The focus was on givingthe people in the audience a tool to truly reflect and be honest about it. Theresearch did not focus on the other half circle, i.e., what would happen afterthe audience submitted their answer. The hypothesis is that talking about itcould promote the domino effect, or that sharing would bring the group closertogether. To go full circle, additional research is needed. Questions to ask wouldbe: how can Mentimeter let the presenter use the tool to promote discussionafterward? And what is necessary for the audience to dare to be vocal?

Talk about it also shift the attention to the presenter or the one leading thereflection. A trustful person is vital for the people in the audience to feel confi-dent enough to share. This is nothing that can be controlled from Mentimeter’sperspective. However, if the tool offers to guide the presenter to deal with theresults, and how to interpret them, maybe both the presenter and audiencewould feel more confident to talk about it afterward.

43

Page 49: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Bibliography

[1] Pogosyan, M.: The benefits of emotional awareness, in-sights into our emotions from paul ekman (2018) https://www.psychologytoday.com/us/blog/between-cultures/201801/the-benefits-emotional-awareness, accessed 2019-01-31.

[2] Goleman, D.: How emotionally intelligent areyou? (2015) http://www.danielgoleman.info/daniel-goleman-how-emotionally-intelligent-are-you/, accessed2019-01-23.

[3] Lambie, J.: Emotion experience, rational action, and self-knowledge. Emo-tion Review 1(3) (2009) 272–280

[4] Pervez, M.A.: Impact of emotions on employee’s job performance: An ev-idence from organizations of pakistan. OIDA International Journal of Sus-tainable Development 1(5) (2010) https://ssrn.com/abstract=1668170,accessed 2019-01-24.

[5] Goleman, D.: Emotional intelligence http://www.danielgoleman.info/topics/emotional-intelligence/, accessed 2019-01-24.

[6] Gerbrandt, S.: Self-awareness: Its relationship to personal effectivenessand success in the workplace. ProQuest Dissertations Publishing (2005)

[7] Goleman, D.: Emotional intelligence. Bantam Books Inc (1995)

[8] Tartakovsky, M.: When you’re scared of feelingyour feelings (2018) https://psychcentral.com/blog/when-youre-scared-of-feeling-your-feelings/, accessed 2019-02-27.

[9] Katharina Kircanski, Matthew D. Lieberman, M.G.C.: Feelings into words.Sage Journals 23(10) (2011) 1086–1091

[10] McLaren, K.: Your emotional vocabulary list https://www.karlamclaren.com/emotional-vocabulary-page/, accessed 2019-05-13.

[11] Margaret E Morris, Qusai Kathawala, T.K.L.E.E.G.F.G.M.L.W.D.: Mo-bile therapy: case study evaluations of a cell phone application for emo-tional self-awareness. Journal of Medical Internet Research 12(2) (2010)

[12] Sebastiano Bagnara, S.P.: Design for reflection. Work 41(1) (2012) 1108–1113

44

Page 50: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

[13] Robin H Kay, A.L.: Examining the benefits and challenges of using audi-ence response systems: A review of the literature. Computers & Education53(3) (2009) 819–827

[14] Travis Bradberry, J.G.: Emotional intelligence 2.0. TalentSmart Inc (2009)

[15] Mentimeter: Create fun and interactive presentations https://www.mentimeter.com/why, accessed 2019-01-18.

[16] Cullen, E.: Great leadership - how to holdmeetings and workshops on sensitive topics (2018)https://www.mentimeter.com/blog/great-leadership/how-to-hold-meetings-and-workshops-on-sensitive-topics, ac-cessed 2019-01-21.

[17] : Voters data (2019)

[18] Ingvar, N.: The value of mentimeter. Personal Communication (2019)

[19] Ekman, P.: What scientists who study emotion agree about. Perspectiveson Psychological Science 11(1) (2016) 31–34

[20] Goleman, D.: An ei-based theory of performance. In Cary Cherniss, D.G.,ed.: The Emotionally Intelligent Workplace: how to select for, measure, andimprove emotional intelligence in individuals, groups, and organizations.Jossey-Bass, San Francisco, CA, USA (2001) 27–47

[21] Miller, M.: Getting unstuck: The power of namingemotions (2016) https://www.6seconds.org/2018/01/21/getting-unstuck-power-naming-emotions/, accessed 2019-02-11.

[22] I.S.P. Nation: Vocabulary size, growth and use. In: The Bilingual Lexicon.John Benjamins Publishing Company, Amsterdam/Philadelphia (1993)115–134

[23] Siang, T.: What is interaction design? (2019) https://www.interaction-design.org/literature/article/what-is-interaction-design, accessed 2019-05-02.

[24] Smith, G.C.: Languages of interaction design. In: Designing Interactions.MIT Press, Cambridge, Massachusetts, USA (2007) xvii–xix

[25] usability.gov: Interaction design basics https://www.usability.gov/what-and-why/interaction-design.html, accessed 2019-05-02.

[26] Anna L. Cox, Sandy J.J. Gould, M.E.C.I.I.I.R.: Design frictions for mindfulinteractions: The case for microboundaries. CHI Conference ExtendedAbstracts (2016) 1389–1397

[27] Shneiderman, B.: The eyes have it: A task by data type taxonomy forinformation visualizations. IEEE Symposium on Visual Languages, Pro-ceedings (03 2000)

[28] Dwight M. Hite, Troy Voelker, A.R.: Measuring perceived anonymity: Thedevelopment of a context independent instrument. Journal of Methods andMeasurement in the Social Sciences 5(1) (2014)

45

Page 51: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

[29] (Firm), I.: The Field Guide to Human-Centered Design. IDEO.org /Design Kit (2015)

[30] Knapp, J.: Sprint. Transworld Publishers Ltd (2016)

[31] Ramsey M. Raafat, Nick Chater, C.F.: Herding in humans. Trends inCognitive Sciences 13(12) (2009) 504

[32] Nilsen, J.: Usability 101: Introduction to us-ability (2012) https://www.nngroup.com/articles/usability-101-introduction-to-usability/, accessed 2019-04-20.

[33] Nilsen, J.: Thinking aloud: The #1 usability tool (2012) https://www.nngroup.com/articles/thinking-aloud-the-1-usability-tool/, ac-cessed 2019-04-20.

[34] Budiu, R.: Between-subjects vs. within-subjects study design (2018)https://www.nngroup.com/articles/between-within-subjects/, ac-cessed 2019-05-23.

[35] Group, P.E.: Emotional self-awareness test design. Personal Communica-tion (2019)

[36] n/a: Levene test for equality of variances (n/a) https://www.itl.nist.gov/div898/handbook/eda/section3/eda35a.htm, accessed 2019-06-04.

[37] Stephanie: Welch’s anova: Definition, assumptions (2016) https://www.statisticshowto.datasciencecentral.com/welchs-anova/, ac-cessed 2019-06-04.

46

Page 52: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Appendices

47

Page 53: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

A Interview Questions

Figure 1: Interview questions.

48

Page 54: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

B Design Iteration 1

Figure 2: Visualization top or bottom.

Figure 3: Navigation: Multiple pages.

49

Page 55: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Figure 4: Navigation: Expand/Collapse.

Figure 5: Navigation: Expand/Collapse with visualization layered.

50

Page 56: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

C Design Iteration 2

Figure 6: Navigation: Expand/Collapse with added filter and final response asmodal

Figure 7: Navigation: Scroll with final response as state change

51

Page 57: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

D Design Iteration 3

Figure 8: Navigation: Toggle with filter

52

Page 58: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Figure 9: Navigation: Toggle with filter and clearing button to go back

53

Page 59: HOW DO YOU FEEL?1325718/FULLTEXT01.pdf · to Ekman, to have better awareness, because “Usually we are not aware that we are emotional until afterwards, when we say something like,

Figure 10: On-boarding version

54