Linköpings universitet
Linköping University | Department of Computer and Information Science
Master’s thesis, 30 ECTS | Datavetenskap
2020 | LIU-IDA/LITH-EX-A--20/051--SE
Using gamification in a mobile
application to treat children with
obesity
–
Increasing the motivation to select healthy food
Erik Grundberg
Supervisor : Erik Berglund, Aseel Berglund Examiner : Mikael Asplund
Upphovsrätt
Detta dokument hålls tillgängligt på Internet - eller dess framtida ersättare - under 25 år från publicer-ingsdatum under förutsättning att inga extraordinära omständigheter uppstår.
Tillgång till dokumentet innebär tillstånd för var och en att läsa, ladda ner, skriva ut enstaka ko-pior för enskilt bruk och att använda det oförändrat för ickekommersiell forskning och för undervis-ning. Överföring av upphovsrätten vid en senare tidpunkt kan inte upphäva detta tillstånd. All annan användning av dokumentet kräver upphovsmannens medgivande. För att garantera äktheten, säker-heten och tillgängligsäker-heten finns lösningar av teknisk och administrativ art.
Upphovsmannens ideella rätt innefattar rätt att bli nämnd som upphovsman i den omfattning som god sed kräver vid användning av dokumentet på ovan beskrivna sätt samt skydd mot att dokumentet ändras eller presenteras i sådan form eller i sådant sammanhang som är kränkande för upphovsman-nens litterära eller konstnärliga anseende eller egenart.
För ytterligare information om Linköping University Electronic Press se förlagets hemsida http://www.ep.liu.se/.
Copyright
The publishers will keep this document online on the Internet - or its possible replacement - for a period of 25 years starting from the date of publication barring exceptional circumstances.
The online availability of the document implies permanent permission for anyone to read, to down-load, or to print out single copies for his/hers own use and to use it unchanged for non-commercial research and educational purpose. Subsequent transfers of copyright cannot revoke this permission. All other uses of the document are conditional upon the consent of the copyright owner. The publisher has taken technical and administrative measures to assure authenticity, security and accessibility.
According to intellectual property law the author has the right to be mentioned when his/her work is accessed as described above and to be protected against infringement.
For additional information about the Linköping University Electronic Press and its procedures for publication and for assurance of document integrity, please refer to its www home page: http://www.ep.liu.se/.
Abstract
In a time where child obesity is reaching an alarming level it is important to provide tools for weight loss that are accessible to many children. Furthermore, it is important for these tools to be easy and fun to use to keep the users engaged in the tool over a long period of time, since losing weight is nothing that can be done in a day. By designing and implementing a mobile application for logging food intake and the incorporation of gamification elements, this study seeks to address the problem of users losing interest in the tool and rebounding to unhealthy eating. Gamification is a concept of making appli-cations that traditionally has nothing to do with games into a more game-like experience to enhance user engagement. After the development of the application was complete, the participants of the study used the application as a part of their daily routine to log all the food they consumed. When interviewing the participants after testing the application the overall response for the gamification elements was positive. However, the general design and implementation of the application seemed to affect the user engagement more than the additional gamification elements.
Acknowledgments
I would like to thank my supervisors Erik and Aseel for their involvement in the project and their help with the design of the application. In addition to them, I would like to thank all the people I have been working closely with during this project and helping me keep my spirits up throughout this process.
Contents
Abstract iii
Acknowledgments iv
Contents v
List of Figures vii
List of Tables viii
1 Introduction 1 1.1 Motivation . . . 1 1.2 Aim . . . 2 1.3 Research question . . . 2 2 Theory 3 2.1 Overweight treatment . . . 3 2.2 Gamification . . . 4 2.3 User engagement . . . 7 2.4 Persuasive technology . . . 8 2.5 Related work . . . 9 3 Method 11 3.1 Pre-study . . . 11 3.2 Implementation . . . 11 3.3 Evaluation . . . 13 4 Results 15 4.1 Pre-study . . . 15 4.2 Implementation . . . 16 4.3 Evaluation . . . 18 5 Discussion 24 5.1 Results . . . 24 5.2 Method . . . 25
5.3 The work in a wider context . . . 27
6 Conclusion 28 6.1 Future work . . . 28
Appendices 30
B Interview questions 32
List of Figures
2.1 The relationship between motivational affordances and psychological and
be-havioural outcomes. . . 6
2.2 The Model of Engagement proposed by O’Brien and Toms. . . 7
3.1 The design of the lo-fi prototype. . . 12
4.1 The view when inputting foods. . . 16
4.2 The main view of the application. . . 17
4.3 Example of three weekly goals, two achieved and one not yet achieved. . . 18
4.4 Example of a badge and the progress towards the next level. . . 18
4.5 The three different moods of an avatar. . . 18
4.6 Results for the three statements in the Focused Attention category. . . 19
4.7 Results for the three statements in the Perceived Usability category. . . 19
4.8 Results for the three statements in the Aesthetic Appeal category. . . 20
List of Tables
2.1 An example of how four different recipes affect energy density of the same meal. . 3
3.1 The user engagement scale scoring. . . 13
4.1 Existing features in other similar apps. . . 15
4.2 The score for each of the categories in the UES-SF survey. . . 21
4.3 A summary of notable comments about features in the app. . . 22
1
Introduction
The increasing amount of overweight and obese children and adolescents is becoming a se-rious health problem around the world, especially since this increases the risk of premature death and disability in adulthood [55]. Apart from these long term risks, children who suf-fer from overweight or obesity may also experience breathing difficulties, hypertension (high blood pressure), etc. as well as psychological effects. According to the WHO, there are sev-eral things an individual can do to reduce or prevent overweight, including eating less fat and sugar, increase fruit and vegetable intake, and regularly engage in physical activity [55]. However, performing these activities to prevent overweight is not something that is done in a day or a week. Instead, they will need to span over several months or even years. As a child, it might be difficult to see the benefits of eating healthier and exercising, especially when many children prefer fat and sugary food over vegetables [7]. Even knowing what to do about it might be difficult for them. That is why parents have a big impact on changing the child’s eating behavior [10, 54], however, mobile applications may also help children to improve their lifestyle [41].
1.1
Motivation
As more and more young children get acquainted with mobile technology at an early age [45] the use of mobile devices as a dietary tool could help in preventing obesity in later stages, since early childhood obesity is connected to obesity later in life [53, 30]. Bryne et al. have conducted a study that suggests that the use of mobile applications may be one way of help-ing children to change their eathelp-ing habits [6]. By develophelp-ing an app for children to track their daily intake of food and the amount of physical activity, these children could receive help in eating healthier and increasing exercise.
A common technique in various weight loss interventions is to set goals. These goals have been shown to increase the effectiveness of the weight loss if they are made concrete rather than vague such as “do your best” [51]. The concept of goals is also commonly present in gamification, the implementation of game mechanics in non-gaming applications. These goals can then be combined with the concept of badges, a way to give the user a reward when completing goals or performing certain actions. Another game mechanic that has previously been used to enhance the user’s eating is avatars [28]. This game mechanic is based on a
1.2. Aim
virtual representation of the user that responds, in a good or a bad way, to what the user does.
Furthermore, gamification have been shown to have a positive effect on user engagement in multiple areas. However, most of the literature reviewed by Darejeh and Salim [8] targets educational software, and the software that targets health have a specific condition in focus, such as cancer or diabetes.
1.2
Aim
This project aims to investigate how an application designed to encourage a healthy lifestyle could benefit from gamification in order to help children improve their eating habits and increasing physical activity over a long time and thus acquire a more favorable body compo-sition.
1.3
Research question
When the study is performed, one main question is taken into consideration.
How does gamification by means of avatars, badges, and goals affect user engagement in an application that encourages healthy eating?
This question intends to answer how a mobile application can be designed and imple-ment gamification eleimple-ments in order too keep the user engaged in the application over a long period of time. The concept of engagement includes, but is not limited to, spending time in the application and enjoying it. To answer this question a mobile application containing all these gamification elements is developed and tested. The design and implementation of gamification elements is a key part in distinguishing this specific application from other sim-ilar products. Thus, these elements have a prominent part in the design. The participants of the study are asked to use the application as part of their daily routine. They are then interviewed and asked to answered a survey about their engagement in the application. The questions in the interview are focused around the affect of the gamification elements to be able to answer the research question in an exhaustive way.
2
Theory
In this chapter theory related to the subject of the thesis is presented. This is to give the reader a basic understanding of the topics included in the study.
2.1
Overweight treatment
The energy density (kcal/g) of foods have been found to affect satiety (the feeling of beeing well fed) and energy intake, which ultimately affects the body weight [24, 46]. To lower the energy density of food Rolls et al. [47] suggest that one might add water-rich fruits or vegetables as well as reducing the fat contents of the meal. In the study Rolls et. al provide an example of this method, and the suggested recipes can be seen in table 2.1. This is further supported by Rolls and Bell [46] in another study where the meals of the participants were analysed and found to contains roughly the same amount of weight everyday rather than the same amount of calories.
In a study targeting children aged 8 to 12 Epstein et al. [15] analyse the effect of two different diets, one increasing the amount of healthy eating, and one reducing high energy-dense foods. Both test groups showed a decrease in zBMI, a BMI adjusted using age and sex [25], at both 12- and 24-month follow-ups. However, the reduction was greater in the healthy eating test group with a reduction of 0.36 zBMI units compared to a reduction of 0.13 zBMI
Table 2.1: An example of how three different recipes affect energy density of the same meal [47].
2.2. Gamification
units in the other group at the 24-month follow-up. The parents of the children in the increase healthy eating group also showed a greater reduction in concern about their child’s weight.
2.1.1
Traffic light labeling
To make the selection of healthier food easier the Traffic Light Labeling (TLL) method may be used. The method consists of categorising all food into three different levels [4]:
• Green - GO!: These items are healthy and should be consumed every day without any specific limitations. Almost all vegetables and fruit are in this category. Other common foods in this category are whole-wheat bread, oatmeal, and low-fat milk.
• Yellow - SLOW!: Items in this category should be the second choice when choosing what to eat. This category includes white grains (e.g. pasta or rice), ham, and 100% fruit juice. • Red - WOAH!: Foods that lack nutrients or are energy-dense belong to this category. These foods should be eaten sparingly. French fries, chocolates, sugary cereals, and soft drinks are all examples of foods in this category.
Ease of identifying healthy food
In 2009 Borgmeier and Westenhoefer [3] conducted a study among German adults on how the different labeling formats affected the ability to pick the healthiest choice. The participants were shown one of the following five food labeling techniques: (1) a single healthy choice marking, (2) a multiple TLL where fat, saturated fat, sugar, and sodium were all marked individually with a color, (3) a monochrome Guideline Daily Amount (GDA) label, (4) a col-ored GDA label, and (5) unlabeled food. When asked to compare foods from different food categories pair-wise there was a significant difference between different labeling techniques in how well the subjects were able to determine which one was healthier. Out of 28 com-parisons, the group that had been exposed to foods without labeling made the least correct decisions (20.2 ˘ 3.2). The group using the TLL technique scored the highest with a score of 24.8 ˘ 2.4. Thus showing that the TLL might be a suitable method for easier identification of healthy food.
2.2
Gamification
Gamification is defined as “the use of game design elements in non-game contexts.” by Deter-ding et al. [13] in 2011 and “as a process of enhancing a service with affordances for gameful experiences in order to support user’s [sic] overall value creation” [23]. In later years gami-fication has become a well-established technique in Human-Computer Interaction [42]. This trend has also made gamification a more frequent subject for research [19].
The previously mentioned definitions rely on the fact that gamification is based on the usage of game elements. However, there is no strictly defined set of elements common in games that would automatically generate a gameful experience [23]. Even if there is no such set defined there are still some game elements that occur often in gamification. In a literature review on gamification made by Hamari et al. [19] in 2014 points, leaderboards and badges were found to be the most common variants by far.
2.2.1
Game Design Elements
Deterding et al. [13] defines five different abstraction levels of game design elements with Game design methods, such as playtesting and playcentric design, as the highest level. On the other end of the spectrum Game interface design patterns is found that consists of badges, leaderboards, levels etc. This level of abstraction is also mentioned as motivational affordances
2.2. Gamification
by Hamari et al. [19]. A more thoroughly description of the most common motivational affordances follow below.
Points
Points could be considered as a simple way of indicating progress in games. This element may appear in multiple ways in games such as: goals scored in sports games, experience points in role playing games, or distance traveled in endless runner games (Temple Run, Minion Rush etc.) [49].
Leaderboards
In order to compare the results of one player with other players games might utilize leader-boards where the results of different users are shown [49]. The leaderboard might be global for all players or divided into different categories, such as age or nationality.
Badges
Badges, or achievements [49], is a way to reward users for performing a specific task or feat. This game element has been implemented by both Microsoft and Sony for Xbox and Playsta-tion respectively. The concept of using badges was analyzed by Denny [12] in 2013 to assess whether badges would have an impact on how students contributed to an online learning tool. The result was a positive effect on student contribution without any significant loss in the quality of the answers. The students also expressed enjoyment of being able to earn badges and most students had a preference of having them over not having them.
The use of badges is also one of the primary gaming elements non-gaming services may use to be gamified according to Hamari [20]. Hamari also suggest that the implementation of badges increases user activity.
Streaks
Streaks gives the user a bonus for consecutive usage of the application. The user might be awarded streak points for logging in to the application or sending messages to your friends daily. Streaks might also give bonus points or multipliers to encourage users to return to the application [48]. If the user fails to maintain the streak the bonuses accumulated are reset to zero.
Goals
Deterding et al. [13] presents clear goals as an example of Game design principles and heuristics. This is at level three of the abstraction levels meaning that it could be based on elements from lower levels, such as points. An example of a goal could be to collect a specified amount of points.
The use of goal-setting has also been studied in the context of motivating physical activity by Munson and Consolvo [29]. In their study the users reported a benefit from being able to set their own goals.
2.2.2
Ownership and Possession
Yu-kai Chou [35] presents possession as one of the eight core drives in the Octalysis frame-work for gamification. This is the drive that keep users motivated because they feel like they own something. Chou also claims that when a player owns something she wants to make it better. This also applies to a person that spends time customizing an avatar. An example of a game with high drive of Ownership and Possession is Farmville where the user owns a fully
2.2. Gamification
Figure 2.1: The relationship between motivational affordances and psychological and be-havioural outcomes [19].
customizable farm, and on the other side of the spectrum is Candy Crush which is based on individual sessions where only a small amount of data is carried over between sessions.
2.2.3
Outcomes of gamification
When studying empirical works on gamification Hamari et al [19]. conceptualizes gamifica-tion into three main parts: the implemented motivagamifica-tional affordances mengamifica-tioned previously, the resulting psychological outcomes, and the further behavioural outcomes. The relation-ship between these can be seen in figure 2.1.
Psychological outcomes
Studies analyzing the psychological outcomes of gamification mostly focus on the attitude, enjoyment and motivational aspects. To evaluate these aspect either interviews or question-naires may be used. As of 2014 there was only one study, performed by Hamari and Koivisto [21], that used validated psychometric measurements, such as recognition, to evaluate the results [19].
An example of where motivational affordances affects the psychological outcomes can be seen in a study by Fitz-Walter et al. [16]. In the study, achievements were added to an application to promote exploration of a university campus and over 95% of the participants agreed to that the achievement system was fun to use and that the achievement system added value to their orientation experience.
Behavioral outcomes
Behavioral outcomes are the most studied outcomes of gamification and was studied in 21 out of 24 studies analyzed by Hamari et al. [19]. To evaluate these outcomes experiments or statistical analyses were performed in already existing services or implementations that were designed by the researchers. Additionally, survey methods may be used to analyze use intentions.
In the study by Fitz-Walter et al. [16] mentioned previously, the increased motivation made the users go out of their way to explore new places on campus. This can be seen as an example of a positive behavioural outcome from gamification.
2.2.4
Gamification as a way of increasing health
To encourage a healthy lifestyle several mobile applications have been made with different approaches. MyFitnessPal1utilizes both streaks and leaderboards to promote healthy eating, while FitBit2uses challenges to make their users more active.
1https://www.myfitnesspal.com 2https://www.fitbit.com
2.3. User engagement
Figure 2.2: The Model of Engagement proposed by O’Brien and Toms [34].
2.3
User engagement
In a time where the user has access to thousands of applications within the click of a button, it is very important to prevent the user from simply abandoning your app in favour of another. This is where the user engagement comes into play, because a successful application is not just usable, it engages the user. Lehmann et al. [26] defines user engagement as “the quality of the user experience that emphasises the positive aspects of the interaction, and in particular the phenomena associated with being captivated by a web application, and so being motivated to use it” . In addition to the previous quote, this section aims to provide an in depth explanation of what defines user engagement and how to measure it in the context of a mobile application.
2.3.1
The Process of Engagement
After conducting a combined literature review and exploratory study on user engagement, O’Brien and Toms [34] presented a model to describe the process of engagement. The pre-sented model begins with the point of engagement, then comes the period of engagement, and the engagement ends with the disengagement step. The disengagement is then followed by the reengagement when the user reaches the point of engagement again, and completes the authors Model of Engagement. The full Model of Engagement can be seen in Figure 2.2.
• Point of engagement - The initial state of engagement that starts when the user is pre-sented with the system. This might be triggered because the system has an aesthetic that appeal to the user, or the user has an interest or motivation to use the system. When participating video gamers were interviewed on this topic both reviews and previews were mentioned as key factors for the point of engagement.
• Period of engagement - After the point of engagement the user enters the period of engagement. In this stage the attention of the users and their ability to focus on the task given played a key part. The video gamers expressed challenge as an important part of maintaining the engagement, one interviewee said “If I’m not challenged, why bother?”. An engaging aspect of the challenge mentioned by other gamers was a diffi-culty that adjusted to the players skill level, maintaining a good challenge even though the player gets better.
• Disengagement - This occurs when the user makes the decision to stop using the prod-uct. This could be both internal decisions, originating from e.g. eye strain or simply wanting to do something else, or external factors, such as an alarm ringing or some-one interrupting the session by talking to the user. The difficulty of video games were
2.4. Persuasive technology
mentioned in this part of the process too, if a task was too difficult it simply was not enjoyable. One participant expressed it in the following way, “...some of those scenarios are so difficult, you’ve tried ten times and you’re "I give up. I’ll just put it away for a couple of days"”.
In addition to this study, Jacques et al. [43] states that a system needs to retain its attractiveness during the course of interaction or the user might become disengaged. The system must also help the user achieve their goals, otherwise the user might lose their interest and end their involvement.
• Reengagement - Just because a user has disengaged from a task or system does not mean that they have ended their engagement in it. This is where the reengagement step comes into the process, to bring the user back to the point of engagement. Among the video gamers in the study this took place when returning to a paused game from taking a short break to eat or drink etc. If a user decides to purchase and play the sequel of a game this can also be seen as an act of reengagement, but in a more long term perspective. Overall the users returning to a system did so because they had a positive experience of the past use of the system. Having fun and being rewarded were among those motivators that affected this experience the most.
2.3.2
Measuring user engagement
There are several ways of measuring user engagement using certain metrics. Common met-rics used as proxies for this are session duration, page views, and number of unique users [26, 39]. All of these metrics are based on user interaction, but rather than focusing on the perceived experience of the user it only captures the statistics.
User Engagement Scale
O’Brien and Toms do [33], on the other hand, suggest a self-reporting tool to measure engage-ment, the User Engagement Scale (UES). The tool consists of a survey where the user gets to answer 31 questions about the specific product. These questions are categorized into six dif-ferent categories: Focused Attention, Perceived Usability, Aesthetics, Endurability, Novelty, and Felt Involvement.
However, after being evaluated in other studies [32, 56], this scale was later revised into the 30 question UES Long Form LF) and the twelve question UES Short Form (UES-SF). The categories were also revised into Aesthetic Appeal, Focused Attention, Perceived Usability, and Reward [31].
2.4
Persuasive technology
Persuasive technology is technology used to promote desirable changes in behaviour, feel-ings or thoughts about a subject through human-computer dialogue and influence [2]. These technologies are often used to persuade users to purchase a specific product, but might be used to improve the user’s lifestyle, or make the user more environmentally friendly.
In 2002 Fogg [17] suggested that the persuasive technologies could be categorized into five types of social cues: physical, psychological, language, social dynamics, and social roles. As a part of the social roles Fogg mentions pets as an example use of persuasive technology. This adaption can be seen in Tamagotchis that were first introduced in the late 1990’s and later followed by Nintendo’s Pocket Pikachu. The big difference between these were that Nintendo’s alternative was equipped with a pedometer. Thus, in order to make this creature thrive the user would have to be physically active, e.g. by walking, running, jumping etc., to activate the pedometer. This small device containing a digitized pet can be seen as a simple device made to persuade the user to change its behaviour.
2.5. Related work
2.4.1
Virtual pets
The idea of virtual pets have since the introduction of Tamagotchis been seen in multiple different forms including Nintendogs, and Kinectimals where the user would interact with the pets using a camera and specific gestures.
The use of virtual pets has also been implemented in applications that haven’t been purely hedonic. In a study performed by Pollak et al. [40] they analyze how the incorporation of a virtual pet in a mobile application would affect the users’ habits of eating breakfast. The study showed that in the group without the pet the participants only ate breakfast 20% of the times, while the group using the pet ate breakfast 52% of the times. When interviewed after the project about what they liked about the application one users said the following about the pet “It interacts with me. It does what I do.”. This can be seen as an example of the Similarity design principle suggested by Oinas-Kukkonen and Harjumaa [36]. The principle states that the user is more easily persuaded by systems that remind them of themselves, e.g. by imitating the user’s behavior.
2.4.2
Rewards
Another design principle suggested by Oinas-Kukkonen and Harjumaa [36] is the Rewards principle. The principle suggests that “Systems that reward target behaviors may have great persuasive powers.”. This technique is used in multiple different applications in the form of visual rewards [29, 28], called intrinsic rewards. If rewards instead are given outside the virtual domain, e.g. in the form of vouchers or discounts, these are described as extrinsic rewards. The use of rewards is one of the most common strategies used in health applications to promote a healthy lifestyle [38].
When Luhanga et al. [28] studied the effect of intrinsic and extrinsic they found that the extrinsic rewards were the biggest motivator to use the application in the beginning. How-ever, at the end of the study all the main motivators were intrinsic.
2.4.3
Reminders
In the Fogg Behavior Model (FBM) Fogg [18] describes behavior as a product of three things: motivation, ability, and triggers. One type of trigger Fogg introduces is the Signal, these are triggers that doesn’t increase the motivation of the user, nor increases the user’s ability to perform a task. In the scope of persuasive technology this signal can be seen as a Reminder, which is one of the design principles introduced by Oinas-Kukkonen and Harjumaa [36]. This principle states that “If a system reminds users of their target behavior, the users will more likely achieve their goals.”.
The use of reminders was studied by Bentley and Tollmar [1] in an application devel-oped to log well-being behaviour. During a one-month pilot study without reminders the frequency of logging was 12%. This number then increased to 63% in the first month of the full study using reminders, the logging then remained at 55-65% for the following month. When conducting interviews at the end of the study several subjects stated the importance of the notifications.
2.5
Related work
This section presents previous research related to this study. Since the study involves both healthy eating and gamification, research related to both these topics will be covered.
2.5.1
Impact of knowing the health rating
Between 2009 and 2012 Thorndike et al. [52] conducted a study on how the Traffic Light Labelling of products affected sales. The data was collected from a large hospital cafeteria
2.5. Related work
with a mean of over 6500 daily transactions. The study started with a baseline period of three months to collect initial data. After this baseline period, the items sold in the cafeteria were labeled according to the TLL technique and the shelves rearranged to make items with a green label more visible. All these labels and rearrangements were made permanently. When the results were analysed they showed that the proportion of red items sold decreased from 20% to 16%, a relative change of -20%, the same number for green items was +12%. This suggests that the TLL technique has a sustained effect on items sold.
Similar studies have also been performed by Olstad et al. [37] and Ellison et al. [14] with comparable results, further supporting the claim that the TLL impacts peoples choice of food.
2.5.2
Gamification as a way of increasing user engagement
In 2016 Stanculescu et al. [50] conducted a study in an enterprise environment in cooperation with IBM, and one of their research questions was “Which game mechanic is more effective for driving user engagement?”. The authors found that the implementation of leaderboards or badges, and both combined, had a great impact on how much the user used the system, thus being a more engaging configuration. When analysing the aesthetic aspect, proposed by O’Brien and Toms [33], of the system, the version using both badges and leaderboards re-ceived a significantly higher rating than the other configurations. This surprised the authors since all the different versions were designed to have the same look and feel.
The previously mentioned UES have also been used to evaluate the effect of gamification combined with mid-air haptic feedback in a digital signage campaign. In this study Limerick et al. [27] found that signage containing more gamification and intractability had higher values for all the four categories in the User Engagement Scale and the user also spent more time watching those posters.
3
Method
In this chapter the method of how the study was conducted is described. Since the project involves a pre-study, an implementation and an evaluation part all of these will be covered in detail.
3.1
Pre-study
Before the implementation a discussion with the stakeholder took place. During this phase different approaches and goals of the project were considered. Already existing applications with similar functionality were also studied, both with and without the stakeholders pres-ence.
When an approach and goals were agreed upon, a lo-fi prototype of the application’s design was made. In the prototype the user would input a food by searching for its name and then selecting it from a list showing the food along with a categorisation according to the Traffic Light Labelling system, and the number of calories per 100 grams of that food. After inputting all the eaten food an overview would be shown to the user with all the meals and the amount of food for each category summed up. The design of the prototype can be seen in figure 3.1. This prototype was then showed to a qualified dietitian and further discussed to comply with the requirements from the intended end-users, the dietitian’s patients.
3.2
Implementation
Since the app was made from scratch several decisions had to be made. These decisions included what platforms, frameworks, and back-end technologies etc. to be used.
3.2.1
Choosing gaming elements
When determining which gamification elements to incorporate in the application Chapter 2 was evaluated. The elements found to be most suitable for this project were:
• Goals • Badges
3.2. Implementation
Figure 3.1: The design of the lo-fi prototype.
• An avatar
Other gamification elements were also mentioned in Chapter 2, however, those were not prioritized to limit the scope of the project to a reasonable level.
3.2.2
Framework
To facilitate development for both iOS and Android, the React Native1framework was chosen for development. This decision was based on the ability to write code that could be generated into one app for each platform, instead of developing one app in Java, for Android, and one in Objective-C, for iOS. The React Native framwork was combined with Expo2to easily test
the application on a physical device instead of an emulator. This approach also simplified the process of testing the app on multiple phones, since only the Expo app and a link to the project is needed.
3.2.3
Back-end
When selecting a technology to handle the back-end logic, a custom implementation was deemed too time consuming and out of scope for this project. Instead Google’s Firebase3,
and in particular Cloud Firestore, was used for the back-end. These services handled user authentication and the storing of user data.
The use of Firestore’s field:value structure in combination with React Natives usage of JavaScript provided a simple conversion between database data and in-app use of the data. It also made it simple to save data to the database. Since the Firebase’s NoSQL-storage is schema-less an iterative development process is more straightforward since new data, con-taining different fields, can be inserted without having to remodel the database. If a
tradi-1https://facebook.github.io/react-native/ 2https://expo.io/
3.3. Evaluation
Table 3.1: The user engagement scale scoring. Strongly Disagree Neither agree Agree Strongly
disagree nor disagree agree
1 2 3 4 5
tional SQL database would have been used every table would have been needed to be redone every time an update was made that altered the structure of the table.
The data containing all of a user’s logged food was stored online to enable users to log into their account across multiple devices and still see the latest version of the data. In a bigger scope this could also have made it possible for dietitians to see how their patients are eating.
3.2.4
Front-end
The avatars were rendered using Scalable Vector Graphics (SVG). This format is composed of shapes rather than pixels, which makes it possible to scale images without any loss in quality. It also improves the ability to animate images seamlessly since the difference between the rendered frames are calculated instead of saved as separate images. The use of SVGs also simplifies the ability to alter single parameters of the pictures, such as the color of the avatar.
3.3
Evaluation
In order to evaluate the application, multiple persons tested it and then answered a survey complemented with an interview.
3.3.1
Distribution of the app
The app was distributed by sending an Android Package (APK) file directly to the users in the study. This was the only version distributed due to the simplicity of sharing and installing the app. This method was selected over publication on Google Play and App Store to only let certain selected people get access to the application.
3.3.2
Participants
Seven participants took part in the study, where three of them were female and four of them were male. The participants were recruited through friends and family and had an age distri-bution of 1 participant under 18 years, 4 between 18 and 28, and 2 participants over 28 years of age. The median age of the participants was 22 years. When doing the experiment the participants were asked to use the app to log all their food and physical activity for as long as they wanted, limited to one week at most. Two of the participants used the app for the full week while the rest settled for three days or less.
3.3.3
User Engagement Scale
The user engagement was measured using the User Engagement Scale, presented in section 2.3, where the user is given a survey containing multiple statements and then rate each one on a scale from 1 (Strongly disagree) to 5 (Strongly agree) [31]. The full scale can be seen in Table 3.3.3.
The short version of the UES was used since this was developed to encourage a higher completion rate of the entire questionnaire. However, since the target group for this study consisted of both children and adults the questionnaire was translated to Swedish to ensure that all participants understood the questions in the same way. The original, non-translated,
3.3. Evaluation
version of the survey can be seen in Appendix A. The questions sent to the users were the following:
FA-S.1 Jag förlorade mig själv i upplevelsen. FA-S.2 Tiden flög förbi när jag använde FoodApp. FA-S.3 Jag blev uppslukad av upplevelsen.
PU-S.1* Jag kände mig frustrerad när jag använde FoodApp. PU-S.2* Jag kände att FoodApp var förvirrande att använda. PU-S.3* Det var påfrestande att använda FoodApp.
AE-S.1 FoodApp var attraktiv.
AE-S.2 FoodApp var estetiskt tilltalande. AE-S.3 FoodApp tilltalade mina sinnen. RW-S.1 Det var väl värt att använda FoodApp RW-S.2 Min upplevelse var givande.
RW-S.3 Jag kände mig intresserad av den här upplevelsen.
To calculate the overall engagement score, all the individual items are added together and then divided by the number of statements. When doing this it is important to notice that items marked with * are reverse coded. Thus, a score of 1 should be counted as a 5 and vice versa for these items. Individual scores for the different categories can also be calculated by adding the items for each sub-scale and divide by three. For example Rewards is calculated by adding RW-S.1, RW-S.2, and RW-S.3 and then divide the total by three.
3.3.4
Interviews
All the participants were invited to interviews to validate the results and capture thoughts the participants were not able to express in the survey. The interviews were performed in a semi-structured way, an interview where the respondents are asked the same initial questions that may progress into different follow-up questions. All the predefined questions asked can be seen in Appendix B.
To be able to elicit information about the interviewees opinions on the gamification el-ements there was one predefined question about each of the implemented elel-ements. These questions were used as a starting point for a discussion where the participant would highlight their own reflections of the elements. This method was chosen over a traditionally structured interview with short direct answers because then the participants would not have been given the same freedom to express their own thoughts about the gamification elements. However, the predefined questions were simple yes or no questions to make the interviewee think about the element without having the pressure to come up with a long answer. The follow-up ques-tions would then have an open-ended character to pinpoint why the participant answered the way he did.
In addition to the questions about the users perception of the implemented gamification elements there were questions regarding their overall experience with the application and how the application compares to other similar ones. This was to distinguish between how the gamification affected and how the general design of the application affected the user. These questions were also used to investigate if the user would mention the gamification elements without being specifically asked about them, and thus noticing if the interviewee had any reflection about them without being reminded of them.
4
Results
The results of the different phases of the study are presented in this chapter.
4.1
Pre-study
Before starting the implementation, other applications were studied to get an idea of what aspects of an app that were important to incorporate. Based on the apps mentioned in 2.2.4 together with FatSecret1and Lifesum2, food logging and activity logging was deemed as key features to include in the app. The different features for each of the studied apps can be seen in Table 4.1.
One prominent feature in the previously mentioned apps was counting calories, this fea-ture was, however, excluded from this project. The reason for excluding calories was the input received from the dietitian during the pre-study. According to her calories are not a suitable method of increasing a child’s health. Instead of using calories as a measurement for what foods are healthy, the Traffic light labeling mentioned in 2.1.1 was used. Every food available from the Swedish National Food Agency’s food database3 was given a category: Green, Yellow, or Red in accordance with the Swedish National Food Agency’s recommen-dations [9].
Another feature used by the studied apps that was excluded from this project was the weight goals. This was because children that are still growing will lower their zBMI by in-creasing their height instead of lowering their weight.
1https://www.fatsecret.se 2https://lifesum.com/
3http://www7.slv.se/SokNaringsinnehall
Table 4.1: Existing features in other similar apps.
Calories Food logging Activity logging Weight goal Social feed Lifesum Yes Yes Yes Yes No
MyFitnessPal Yes Yes Yes Yes Yes
FatSecret Yes Yes Yes Yes Yes
4.2. Implementation
Figure 4.1: The view when inputting foods.
4.2
Implementation
The core of this application is an application where the user logs all the food eaten and all the activity performed. Every food item is then classified with a green, yellow or red dot, according to the traffic light system mentioned in 2.1.1. When inputting the food, the user is shown a list of suggestions containing all foods including the sought after term and the classification of all the suggested items. An example of the process can be seen in Figure 4.1.
The user can also specify the amount eaten and see the nutritional values for every prod-uct. In addition to the food, the user input different activities. However, these activities are not classified and does not contain any additional information, e.g. calories burned.
When all the foods and activities for a day is inputted an overview of the day is shown. This view, the main screen, is also the first view the user sees when entering the application. The main screen of the application can be seen in Figure 4.2.
4.2.1
Gamified application
In addition to the core application, multiple gamification elements mentioned in Chapter 2 were implemented to enhance the users experience and increase user engagement.
Goals
The user is able to set their own goals related to both eating and activity. The different types of goals that can be set are:
• Minimum number of green category food - Daily and weekly • Maximum number of red category food - Daily and weekly • Minimum amount of activity - Daily and weekly
• Minimum number of days containing only green category food - Weekly • Minimum number of days without red category food - Weekly
4.2. Implementation
Figure 4.2: The main view of the application.
The user can set any number for the goals and adjust them freely. An example of set goals can be seen in Figure 4.3.
Badges
Badges were implemented in the application for the user to be rewarded for repeatedly per-forming certain actions, such as exercising. In Figure 4.4 the badge for activity minutes is shown together with the progress towards the next badge in that category.
Avatar
To give the user something more than just herself to care for, an avatar was implemented in the application. The food the user inputs into the application affects the overall happiness of the avatar, when the user eats healthy, exercise and fulfill all set goals the avatar becomes happy. The three different moods of an avatar can be seen in Figure 4.5. The avatar is also
4.3. Evaluation
Figure 4.3: Example of three weekly goals, two achieved and one not yet achieved.
Figure 4.4: Example of a badge and the progress towards the next level.
Figure 4.5: The three different moods of an avatar.
customisable to the extent that the user can choose one out of five possible appearances. This corresponds to the Ownership and Possession drive in the Octalysis mentioned in 2.2.2.
4.3
Evaluation
The evaluation for the study was split into two pieces, a survey to get the results for the user engagement scale, and interviews to capture more qualitative responses.
4.3.1
User engagement scale
The seven participants that tried the app were asked to answer the user engagement survey mentioned i Section 3.3 and all of them responded. The results from the survey can be seen in
4.3. Evaluation
figure 4.6 - 4.9 where the results for each question is shown. The x-axis represents the user’s opinion, ranging from 1 (Strongly disagree) to 5 (Strongly Agree). The y-axis represents the number of users giving a specific answer. When reading the results it is important to remem-ber that the questions in figure 4.7 are reverse coded, meaning that a disagreeing answer is favorable. 1 2 3 4 5 0 1 2 3
Jag förlorade mig själv i upplevelsen
1 2 3 4 5
0 2 4
Tiden flög förbi när jag använde FoodApp
1 2 3 4 5
0 2 4
Jag blev uppslukad av upplevelsen
Figure 4.6: Results for the three statements in the Focused Attention category.
1 2 3 4 5
0 2 4
Jag kände mig frustrerad när jag använde FoodApp
1 2 3 4 5
0 2 4
Det var påfrestande att använda FoodApp
1 2 3 4 5
0 2 4
Jag kände att FoodApp var förvirrande att använda
4.3. Evaluation
1 2 3 4 5
0 1 2
FoodApp var estetiskt tilltalande
1 2 3 4 5
0 1 2
FoodApp var attraktiv
1 2 3 4 5
0 2 4
FoodApp tilltalade mina sinnen
Figure 4.8: Results for the three statements in the Aesthetic Appeal category.
1 2 3 4 5
0 2 4
Det var väl värt att använda FoodApp
1 2 3 4 5
0 2 4
Min upplevelse var givande
1 2 3 4 5
0 2 4
Jag kände mig intresserad av den här upplevelsen
Figure 4.9: Results for the three statements in the Reward category.
When averaging these results, a score for each of the four categories are found. These category scores can be seen in Table 4.2, it is worth noting that the Perceived Usability (PU) score have been reversed to match the format of the other scores.
The results for each category are then averaged and the total engagement score was found to be 3.31.
4.3.2
Interviews
When performing the interview, several aspects of the application not covered in the survey was covered, one of these were the comparison to other similar applications.
4.3. Evaluation
Table 4.2: The score for each of the categories in the UES-SF survey.
Category Score
Focused Attention (FA) 2.38 Perceived Usability (PU) 4.24 Aesthetic Appeal (AE) 3.10
Reward (RW) 3.52
One respondent compared FoodApp to MyFitnessPal and found it easier to input food in MyFitnessPal, it was especially the ability to save complete meals composed of several foods that stood out in the favour of MyFitnessPal. This feature was also something that was mentioned by another respondent as a point of improvement of FoodApp. This respondent had never used MyFitnessPal but instead he had used FatSecret, which also contains this feature.
The use of the Traffic Light Labelling was something the respondents commented on as a positive feature. Among the respondents that had a desire to lower their BMI, the opinion on the categorisation was slightly more positive. The categorisation had a small impact on the behaviour of some of the respondents in the way that after seeing that a specific food had a red category this food was avoided in the future.
Even if the concept of labelling the food was seen as positive, the individual categorisation was not always seen as positive. Two interviewees expressed that they thought the labelling was wrong. In their view the categorisation was too strict and not in line with what they have been taught about healthy food. Suggestions arose on this problem during the interviews, where one was to have an explanation on why an item was labelled the way it was. Another suggestion was to introduce different modes where the user would specify how physically active it was and depending on this the categorisation of the foods would be stricter or more relaxed. Another negative experience one participant got from the labelling was that she felt judged by the red, and seemingly bad, color. In an application utilising calorie count this would not happen in the same way since 1000 kcal is always 1000 kcal, regardless of they are from potatoes and salad or chocolate bars.
An implementation of reminders in the form of notifications was suggested by one par-ticipant since this would have helped him to keep inputting food consistently throughout the day. As the implementation was now he said that he only entered the food into the app once per day which made it difficult to remember exactly what he ate, and in particular the amount of each food.
In table 4.3 a summary of notable comments about features, and lacking features, of the app can be seen.
4.3.3
The effect of gamification elements
During the interviews the participants were asked questions specifically about how the im-plementation of gamification elements affected them. Those questions were asked to get a deeper understanding about how the gamification had affected the user engagement, and thus being able to find an answer to the research question presented in section 1.3. In table 4.4 the impact of the different gamification elements, according to the outcomes of gamifi-cation presented in section 2.2.3. This models states that motivational affordances leads to psychological outcomes, which then leads to behavioral outcomes.
As seen in the table, only the goals implemented had any effect on the users behaviour and thus this can be seen as the most important gamification element in the application. The avatar had a psychological effect on the user which made that the second most important gamification element in the study. Least important to the users engagement was the badges
4.3. Evaluation
Table 4.3: A summary of notable comments about features in the app.
Positive comments Negative com-ments
Suggested im-provement
Traffic Light Labelling Good thing to include categorisa-tion
Wrong categorisa-tion, too strict cate-gorisation
Introduce different modes adjustable by the user
Custom recipes - It took too much time to input the same recipe multi-ple times Feature to save custom recipes would have enhanced the experience
Reminders - It was difficult for the user to remem-ber to input their food regularly
Add a feature to send notifications when no food had been logged in a day
Table 4.4: The effect each of the implemented gamification elements had on the users. Motivational Psychological Behavioral
Badges Partly No No
Avatar Yes Yes No
Goals Yes Yes Yes
since these only partially increased the users motivation to engage in the app. Below are more thoroughly descriptions on how each element affected the users.
Goals
The ability to set goals was used by most participants and overall appreciated. One goal type that was extra appreciated was the activity goal type, where the user set a goal for how many minutes of activity she would perform daily. This was also the only goal that had a noticeable effect on the users behaviour, for example by walking to a destination instead of taking the car.
Badges
The badges did not have as good effect on the behaviour as the goals. However, the badges were also appreciated in general. One comment that arose about the badges was the scaling between the different levels. The thresholds for each new level was fairly far apart and thus made it difficult to achieve new levels during the testing period. Some levels were even impossible to achieve since they simply required more days to complete than the test lasted.
Avatar
The use of an avatar was also well received by the participants. Several interviewees ex-pressed that they reacted emotionally to the avatar’s mood. The comments by the avatar was seen as a good way to get indications on how to behave to increase its mood and make it happier. They were, however, not always as descriptive as desired and the user could not al-ways figure out how to make it happy. The design of the different avatars were also brought up during the interviews. The application currently have five different avatars, that are vari-ations of two different avatars. A suggestion received during an interview was to make the avatars more diverse, and also to include “cuter” avatars. This suggestion was based of the
4.3. Evaluation
thought that with an avatar that the user could relate more to the user would feel a stronger bond between her and the avatar and thus get more affected by the avatars feelings.
5
Discussion
In this chapter the results and the methods to achieve those are discussed.
5.1
Results
This section covers a discussion on the results of the study, taking some of the related work mentioned in Chapter 2 into consideration. This is to put the study into perspective and analyse why the results turned out as they did.
5.1.1
User Engagement
The overall score from the survey is difficult to evaluate since there is only one number and nothing to compare it to. Nonetheless, the different category result can be compared to each other. When doing this, one can see that the score for Focused Attention (FA) was consider-ably lower than the other scores. O’Brien and Toms defines FA as “feeling absorbed in the interaction and losing track of time” [31], this might seem positive for applications where the user is supposed to spend a lot of time. However, in an application made for quick logging it might be seen as a negative thing to lose yourself in the experience or having time slipping away when you use the app. This might relate to the fairly high score for Perceived Usability (PU) since a major part of the design process for the application focused on making the app as simple as possible to use. That high result for the PU may also have a high result for the user’s overall engagement of the application since the Perceived Usability is found to be the attribute that contributes most to user engagement, according to Holdener et al. [22].
5.1.2
The impact of gamification
The implemented gamification elements seemed to have a positive effect on the user engage-ment based on the interviews. This was as expected since the studies engage-mentioned in Chapter 2 all had a positive result from using gamification as a mean of increasing user engagement.
Avatar
Multiple participants expressed a desire for a wider variety of avatars, and avatars that were “cuter”. By implementing these suggestions, the Aesthetic Appeal (AE), and thus the overall
5.2. Method
user engagement, could increase. A wider selection of well designed avatars could also in-crease the amount of time the user spent in the engagement step described in section 2.3, as an unattractive system might make the user loose interest in it [43].
Goals
The goals did not have a significant impact on the users engagement in the app. They did however, have an impact of the user behaviour outside of the application. Since the goals could be set and then left as they were, they required little interaction from the user. If the application instead would present the user with a daily goal, the user might open the ap-plication more often to see what their current challenge is. The concept of a challenge in itself might also be something that improves the user engagement, since this is one of the motivational affordances mentioned by Deterding et al. [13]
Badges
In the application the user had to navigate to a separate tab in order to see the badges and the current progress towards the next levels. One user expressed that she would open the badges tab from time to time just to see how she progressed and if she had unlocked any new badges, but mostly she did not think about them. To prevent this, the progress of a badge could instead have showed up as a small notice on the screen when the user achieved e.g. 50% of the level. This could have served as a reminder [36], mentioned in Section 2.4.3, for the user to think of the badges more often. By thinking more of the badges, the user might have come back to the application for more days since some of the badges required healthy eating over several days. The introduction of a streak badge could also have increased the users motivation to use the app more consistently since this badge would reset to zero if the user did not use the app one day [48].
5.1.3
Implementation in general
One user mentioned the lack of notifications in the application. If that feature would have been added, the users might have been more likely to return to the application [36, 1]. With an increased amount of participants using the app for a longer period of time this could have provided more correct responses to the survey as well as more specific answers to the interview questions.
The categorisation of the food was also a main point when evaluating the project. All of the participants reacted mostly positive to it but since some participants disagreed with the categorisation, the user engagement could have been lowered. One participant even stopped using the app almost immediately when seeing how hard it would be to avoid the red food. The categorisation did however help some participants in their choice of food after seeing the red or yellow dot next to the food. This, together with the study by Borgmeier and Westen-hoefer [3], can be seen as an indication to keep experimenting with the Traffic Light Labelling to increase the number of healthy food choices.
5.2
Method
Since the project consisted of three distinct phases, all of these will be discussed in this section.
5.2.1
Pre-study
When doing the pre-study a more through cooperation with qualified dietitians could have contributed toward an application that was more accurately developed for the intended end-users, a dietitian’s juvenile patients. This cooperation could also have increased the accuracy of the classifications of the different food items in the app, since the categorisation was now
5.2. Method
made by laypersons. The categorisation of the food was one thing that was mention during the interviews as a key to frustration when using the app. This was also reflected in the UES score since that decreased the perceived usability, even if the frustration was not a part of the interaction with the app, but rather the content of the app. With a more correct classification this could have been avoided and the survey results could have given a score closer to the truth.
5.2.2
Implementation
During the course of the implementation phase, all the work was done during one single iteration. By performing multiple iterations, with user tests in between each, instead it would have been easier to adjust the project according to input from the users.
The scope of the project also turned out a bit bigger than expected. Since the application had to be created from scratch, many hours had to be put into just learning how to create an application and making it work. If the project instead would have been based of an al-ready existing application, more time could have been put into enhancing the gamification elements of the app. These enhanced elements could have contributed to a higher impact of the gamification.
5.2.3
Evaluation
A major drawback of this evaluation was the relatively low number of participants. Since there were only seven people who tested the app the results of the study cannot be seen as fully reliable. In addition to the low number of overall participants only four of them had ambitions about lowering their BMI. This might have led to a lowered overall engagement score since a significant amount of the participants might have been fully uninterested in the application.
With a bigger number of participants A/B-testing could have been done of the app. In A/B-testing two, or more, versions of the same application are developed with minor differ-ences between them. In this case one version could have been stripped of all the gamification elements. Another approach could have been to make a separate version for every element of gamification. By doing this the specific impact of the gaming elements could have been isolated. As the study was performed now it is impossible to conclude whether the results are because of the gamification or simply of how the application was designed. The study in an enterprise environment at IBM mentioned, in Section 2.5.2, used this approach to study the effects of different game mechanics [50]. In that study the authors found correlations that would have been impossible to find if only one test group would have existed, thus the study on FoodApp could have shown a more comparable result if A/B-testing was utilised.
When seeing that only four of the users used the app for three days or more the idea of measuring user retention arose. User retention is how often and for how long the user returns to an application. This might seem like a good thing to measure since loosing weight is a long term project and cannot be done in a couple of days. This also leads to the idea to perform the same study with more participants actually interested in lowering their BMI. It would then have been interesting to perform a more long term evaluation where the BMI was measured at the start and after three, six, or even 12 months, the results could then have been compared to studies with similar goals like the one by Epstein et al. [15] described in Section 2.1.
User engagement scale survey
When doing the evaluation, the questionnaire was translated to Swedish to ensure that every participant in the study would be able to comprehend the meaning of each question. When doing this, the question arose about how questions FA-S.1 I lost myself in this experience and FA-S.2 I was absorbed in this experience would be translated. The way the translations were
5.3. The work in a wider context
formulated might be seen as the same question simply formulated in two different ways. This might skew the results compared to if the original English survey had been done, and ultimately lower the reliability of the study. When Holdener et al. performed a study using the User Engagement Scale the survey was translated to German. When doing the translation some of the items where removed from the survey because no good translation was found or that the users could not relate to the question [22]. The same thing could have been done in this survey to get the most accurate reflection of the users engagement.
Although the survey was translated to Swedish the responses from the survey did not always match the opinions elicited during the interviews. To prevent this the survey could have been translated using a simpler language. This might have changed the original mean-ing of the questions slightly but the end result might have been closer to the users actual engagement in the application.
5.2.4
Source criticism
The area of healthy eating is fairly widespread with many good sources. Because of this it was quiet simple to find good, peer-reviewed sources. However, most of the sources on the effect of the Traffic Light Labeling concern the effect on adults instead of the intended users of this application, children. This was deemed irrelevant and the choice to use TLL remained. The peer-review sources found on TLL were British or American with suggested labeling contradicting the recommendations given by the Swedish National Food Agency. Because the primary end-users of the application were Swedish it was decided to use the Swedish National Food Agency’s brochures [9] and leaflets instead as these were deemed thrust wor-thy as well.
5.3
The work in a wider context
The effect of gamification is not limited to just mobile applications for healthy eating or mo-bile applications in general. It could be used in a variety of contexts ranging from politics and marketing [44] to increasing student motivation [5, 11]. By studying the effects of gam-ification further and making the psychology behind it more widespread among the public it might be easier for the general public to resist the negative intentions where gamification is used.
If the use of TLL was introduced to more people at a young age it could be possible that more people get a better understanding of what foods that are healthy and not. This might finally lead to a decrease in overweight among the public.
6
Conclusion
The purpose of this project was to investigate how certain gamification elements could af-fect user engagement in an application for healthy eating. To guide the study, the research question that was chosen and presented in Chapter 1 was:
How does gamification by means of avatars, badges, and goals affect user engagement in a healthy eating application?
When interviewing the participants about the application, all of the gamification elements got positive comments in general and the chosen elements seemed to have a positive effect on the user engagement. Even if the badges got positive comments, they might not be the most effective way of increasing user engagement in this particular context. Instead the avatar and the goals might be more suitable for a healthy eating application. However, since the evaluation only consisted of a few people the results can not be seen as fully reliable, but rather as an indication that the effect was positive.
The Traffic Light Labelling for the foods were more commented on during the interviews than the gamification elements, and thus they seemed to have a bigger effect on the user engagement. This indicates that the design, and overall implementation, of the app is more important for user engagement instead of just the incorporation of gamification elements.
6.1
Future work
Since the scope of the project was rather limited there are several directions that this project may be continued in. This section will suggest some of the possibilities, both regarding re-search specific continuation and how to improve the implementation.
6.1.1
Research specific
Since losing weight is a long term process, a continuation of this project could be to focus the research on user retention. This way the usage of the application over time could be studied and evaluated in a more life like scenario. That kind of evaluation could also be combined with physical measurements of the participants to see if the application actually had an effect on the health of the users, since that is the underlying goal of using the application.
6.1. Future work
In addition to the previously mentioned points of improvement for the future, the actual evaluation for the project might be done differently. As mentioned in section 5.2 the survey would provide a more comprehensible result if done in the form of A/B-testing. This way the numerical result could be compared to something and thus actually provide some value when trying to answer the research question. This approach would require a higher amount of participants, one way to achieve this could be to make the application publicly available, for example by releasing it in both Google Play and Apples App Store.
6.1.2
Implementation specific
The most commented point of improvement for this project was the labelling of the food. Most of these comments regarded the strictness of the categorisation, saying that too much of the food was considered as red food. This could be further developed by introducing a way for the user to adjust how the labelling worked, for example by utilising different algorithms based on the users activity level, or to let the user specify how to categorise the food by herself.
In addition to the labelling, the avatar might be improved greatly, especially the interac-tion with the avatar. If the users acinterac-tions were more closely connected to the avatar and the avatar would react to more things that the user did, this might increase the user engagement. Suggestions that were mentioned when talking to the interviewees about this included cos-metic changes to the avatars, for example hats or skins. This could be done by introducing a virtual shop where the user could exchange eaten green foods for the new items. The im-plementation of such a shop could improve the user engagement by adding more intrinsic rewards, mentioned in section 2.4.
