"Cibo" - A serious game raising awareness for the effect that different food has on the environment

Fakulteten för teknik och samhälle

Datavetenskap

Examensarbete

15 högskolepoäng, grundnivå

“Cibo” - A serious game raising awareness

for the effect that different food has on the

environment.

Tim Oscar Dahl Karipidis

Examen: Kandidatexamen 180 hp

Handledare: Jeanette Eriksson

Huvudområde: Datavetenskap

Examinator: Steve Dahlskog

Program: Spelutveckling

Abstract

Climate change might be our time's defining issue and our way of producing and consuming food is a large contributor of global climate change. This thesis investigate the possibility of using serious gaming with the intent of raising awareness for the difference in emissions of distinct food types. With the development of a serious game, this paper aims to raise awareness for these issues among university students. The game was evaluated with nine university students and the results suggests that the use of serious gaming with the intent of raising awareness for the difference in emissions of distinct food types show potential, however it was unclear whether or not the students gained a wider understanding of the issue or only acquired some factual knowledge.

Keywords- _{​environment; sustainability; emissions; entertainment; serious games;} game mechanics; text-based learning; global warming;

1. Introduction

The warming of the earth due to global greenhouse-gas emission is today widely accepted in the scientific community. The 2015 Paris agreement, signed by 195 UNFCCC members,1

aims at keeping the rise of the global temperature well below 2°C, aiming for 1.5°C by the turn of the century (UNFCCC. Conference of the Parties (COP), 2015). However this goal can not be achieved by a decarbonisation of the world’s energy system alone. Food production is the largest cause of global environmental change with it being responsible for up to 30% of global greenhouse-gas emissions. The current ways of producing food and the dietary habits of our time, in combination with a projected population growth to about 10 billion people by 2050 will pose a serious threat to the stability of our planet. A shift to a more sustainable diet is therefore vital for reaching the limits set by the Paris agreement (Willett et al., 2019).

Through their research, Willet et al (2019) predicts a gloomy future for us and our planet if we fail to break these dietary habits. But there are reasons to remain hopeful, not all food types are equal, some are better for the environment than others. Legumes produces almost 250 times less greenhouse gases per gram of protein when compared to ruminant meat. Legumes also outperform both pork and poultry in this aspect, producing almost 40 times less greenhouse gases per gram of protein (Tilman and Clark, 2014). Hence a sustainable reference diet would need to radically cut the consumption of produce such as ruminant meat, pork and poultry. Willet et al (2019) presents such a diet, the diet presented will require a substantial increase in the global consumption of foods such as nuts, fruits, vegetables, and legumes. A significant reduction in global consumption of food types such as red meat is also required. Unfortunately the sustainable reference diet is far from the norm in today's society.

Meat consumption has been on a steady rise globally throughout the 20th and 21th century (Godfray et al. 2018). Using the reference diet as a baseline, we consume more than three times the sustainable amount of red meat globally. Europe and North America is by far the largest consumers of red meat per capita, consuming over four respectively six times the sustainable amount of red meat. Sub-Saharan Africa and South Asia are the two only regions in the world currently at or below the baseline for sustainable consumption of red meat (Willett et al., 2019).

Hence there is an urgent need to raise awareness among the general population for the environmental impact our diet has, especially in the western world where red meat is being gravely over consumed (Willett et al., 2019). Willett et al. (2019) discuss different approaches of how to influence people's dietary choices using the Nuffield ladder of policy intervention (Council, 2007) with regards to the problem. The Nuffield ladder (See figure 1.1) is a model containing six different steps of intervention used to influence and change behaviours. The steps at the top of the ladder have a higher effect than the steps at the bottom of the ladder (Council, 2007). At the bottom of the ladder we find the first step, which is fittingly named, “Do nothing”. While this step might seem out of place, since doing nothing is arguably not a form of intervention, Council (2007) argues that to do nothing is in fact an active choice with real consequences which is why it is the first step of the ladder. The second step of the ladder is about providing information, to the public, related to the issue one is trying to solve. Information can be conveyed through many different mediums, in recent times even through digital games (Susi, Johannesson and Backlund, 2007). Going upwards through the ladder the degree of intervention gets higher, the third step is about enabling choice, this could for example mean serving vegetarian alternatives in restaurants and schools. The fourth step is about changing the default policy, for example schools could start serving vegetarian meals as the standard option and meat on occasions or as the alternative. Guiding choices through incentives is the fifth step, this could mean giving tax-breaks if you choose to abstain from meat consumption or by lowering the taxes on vegetarian produce. Guiding choices through disincentives is step number six, this could for example be done through higher taxes on environmentally damaging produce such as ruminant meat. The seventh step is about restricting choices, for example restaurants might not be allowed to serve more than 50 grams of ruminant meat to each customer. The last step of the ladder is the complete elimination of choice, for example by making it illegal to sell and consume ruminant meat.

Figure 1.1 shows Nuffields ladder of intervention, this paper and the artefact produced will focus on the second step of the ladder, providing information.

The focus of this paper will be on the second step of the Nuffield ladder, providing information on the issue of unsustainable food. The second step of the ladder has the lowest degree of intervention, apart from “Do nothing”, but it is arguably also the cheapest step in terms of resources, such as time and capital, committed. The low resource commitment is the reason as to why this step was chosen. It is arguably also the only step, apart from do nothing, attainable for a non-food producer and or non-legislative power. The medium chosen to deliver the information will be a digital game. Digital games have historically been seen to serve one single purpose, that of entertainment. But throughout the 21th century that view of digital games has been challenged by a new type of digital games with more than just entertaining properties. Such games, with a purpose other than just entertainment, has many names and one of them is serious games. Serious games are today being used within a broad range of areas allowing for learning experiences that would otherwise be hard to achieve due to factors such as safety-concerns, time-constraints, cost, etc (Susi, Johannesson and Backlund, 2007). Using serious games in-order to raise awareness for complex sustainability issues has been done before with promising results. Hildmann and Hirsch (2008) presented a resource management game with the purpose of raising awareness for environmental issues, showing some promising results. And both Janakiraman, Watson and Watson (2018) and

Liarakou _{​et al., ​(2012) looked at the prospect of using serious games as tools towards} educating on the matters of sustainable development and environmental sustainability. Both papers conclude that serious games has the potential of being used efficiently towards the changing of attitudes, and behaviours in favour of environmental sustainability.

Using serious games with the intent to raise awareness for environmental sustainability shows potential (Janakiraman, Watson and Watson, 2018). However which method to use to deliver the learning objectives of the game as efficiently as possible is not all that obvious. Molnar and Kostkova, (2013) tested the learning properties of two different versions of the same game. One version focusing on conveying information through game mechanics and one version conveying information through a more traditional medium, through the use of text. Out of the four learning objectives, only one suggested that the use of game mechanics worked better than text as a mean of delivering the learning objectives. The other three learning objectives showed no statistically significant difference between the two different versions of the game. The authors further discussed the results and concluded that additional research had to be done to draw any conclusions.

While there are plenty of studies which have looked at the prospect of using serious games with the intent of educating, changing behaviours and raising awareness for sustainability and environmental issues (Ouariachi, Elving, and Pierie 2018; Morganti et al., 2017; Madani, Pierce, and Mirchi 2017). There seems to be a gap in the research for serious games which specifically targets raising awareness within the area of sustainable diets and food sustainability. The main purpose of this study is therefore to develop and evaluate a serious game with the intent of raising awareness of the impact that the food we consume have on the earth. A secondary purpose will be to test and evaluate whether there is any difference in how much information is obtained through a game delivering its learning goals using text and a game delivering its learning goals through game mechanics. Two versions of the game will hence be developed, one focusing on delivering information through game mechanics (Version “A”) and one focusing on delivering information through text (version “B”).

The research questions of this paper will hence be the following:

RQ1 “How did the players awareness, for the differences in greenhouse gas emissions that different food types has, change after playing the game? ”

RQ2 “What was the difference in awareness change between the testers who played version ‘A’ and the testers who played version ‘B?”

The contribution of this paper, and the artefact produced alongside it, will be an evaluation of the use of serious games as a tool for raising awareness of sustainability issues, mainly how food consumption affect the environment. A secondary contribution will be the evaluation of which deliverance method is better to provide awareness in a specific matter, delivering information through game mechanics or delivering information through text.

This paper will follow the design research methodology established by Peffers et al. (2008). The design research methodology is a nominell process consisting of six steps with some reiteration between some of the steps. The design research method fits this study well since it involves the design, development and evaluation of a new artefact (the two versions of the game). Since the artefact will be a serious game, an established design framework has been chosen and will be followed while designing the game. The chosen framework is “The ‘I’s Have It” framework established by Annetta (2010), it will be further discussed in section s 1.1 and 2.1.

The rest of the paper is organized as follows. The next section will present the method which is used. After which the Artefact and its design will be explained and presented in depth. The fourth section of this paper will contain the results from the actual study. The fifth section will be comprised of a discussion part. The sixth section will contain the conclusion as well as suggest future work.

1.1 Related Research

Previous research (Janakiraman, Watson and Watson, 2018) highlights the importance of raising awareness for environmental sustainability and suggest that the usage of serious games as a mean to raise awareness for environmental sustainability show potential. Other research (Ouariachi, Elving, and Pierie, 2018) suggest that many communication efforts, regarding sustainability and the environment, fail because of the way that they convey information. Hence a new format for conveying information about sustainability is needed and serious gaming could be that format. Both Janakiraman, Watson and Watson (2018) and Ouariachi, Elving, and Pierie (2018) points out that more research within this field of study is needed. Janakiraman, Watson and Watson (2018) specifically points out the need of more longitudinal studies within the area to test if the knowledge is retained over time and if it produces any lasting behavioral changes. However, this is outside of the scope of this paper since raising awareness and behavioral changes is not the same thing.

Regarding the way that information is conveyed within serious games, Molnar and Kostkova, (2013) suggests that game mechanics could be a slightly better way of conveying information, in some instances, when compared to plain text within serious gaming. The authors concluded that additional research had to be done to draw any conclusions, the authors also suggested trying to replicate the study using different game genres. While this study is similar to the work conducted by Molnar and Kostkova, (2013) there are some key differences. Molnar and Kostkova had a test group consisting of children while this study will focus on university students. Another key difference is the fact that this study will focus on raising awareness for sustainability and environmental issues, while the Molnar and Kostova study tried to convey factual information about bacteria. And as a second objective it will look at the differences between conveying information through text element within the game and conveying information with game mechanics.

2. Method

The design research methodology by (Peffers et al. 2008) is an iterative process which consists of six steps. The design research method fits this study well since it involves the design, development and evaluation of a new artefact (the game itself).

The design research methodology consists of the following six steps (see figure 2.1). Step 1 - _{​Problem identification and motivation}

In the first step the research problem will be defined and a possible solution should be justified.

Step 2 - _{​Define the objectives for a solution}

In the second step the objectives of the solution should be deduced. The objectives should be rational and based on the problem specification.The objectives can be quantitative or qualitative.

Step 3 - _{​Design and development}

In this step an artefact should be created. An artefact could be any designed object in which a research contribution is embedded in the design. The design of the artefact is also included in this step, this means that the artefacts functionality and architecture should be defined.

Step 4 - _{​Demonstration}

In this step the artefact is to be demonstrated by solving some instances of the problem. This could for example involve a case study, simulation, proof, experimentation or other appropriate activity.

Step 5 - _{​Evaluation}

The evaluation step involves comparison between the objectives of the solution with the observed results from the demonstration. It is possible to iterate back to step 3 from this step or to continue to the final step, communication.

Step 6 - _{​Communication}

The final step involves communicating the problem and its importance, the artifact and its utility and novelty, the rigor of its design and its effectiveness to relevant audiences.

Figure 2.1 gives an overview of the steps and shows the steps from which iteration is possible.

2.1 Usage of the method

Problem identification and motivation

The world is warming due to global greenhouse gas emissions and with food production being responsible for up to 30% of the emissions, a shift in dietary trends is vital. Willett et al. (2019) looks at the Nuffield ladder of policy intervention with regards to the problem. The second step of the Nuffield ladder is about providing information, and while being the second least efficient level of intervention on the ladder it is the only step (apart from the first step, “Do nothing”) feasible without extensive resources (Council, 2007). Providing information can be done with the use of a wide range of different mediums, including digital games such games are called serious games (Susi, Johannesson and Backlund, 2007). Delivering information through serious games can be done in many different ways, two of them being through plain text and or through game mechanics. Evaluation of which method of information deliverance is better has been done before, a study conducted by Molnar and Kostkova, (2013) suggested a slightly better result for deliverance through game mechanics, but the results were mostly inconclusive and the authors suggested that further research on the topic was necessary (Molnar and Kostkova, 2013). As regards to the usage of serious games with the objective to raise awareness for environmental sustainability. Janakiraman, Watson and Watson (2018) and Liarakou _{​et al.​(2012) looked at the possibility of using} serious games with the intent of raising awareness for sustainable development and environmental sustainability. Both papers concluded that this was a valid use of serious games that showed potential.

Define the objectives for a solution

Hence the solution will be two versions of a serious game focused on raising awareness for a more sustainable diet. The game will have one single learning objective, which will be the following: “Different types of food releases different amount of greenhouse gas emissions”. One version of the game (Version “A”) will be delivering the learning objective through game mechanics and the other (Version “B”) will deliver the learning objective through text. The main objective of the solution is to increase awareness of the impact different food types has on the environment. A second objective of the solution is to evaluate if there is any difference in awareness change between the two versions of the game. The first objective of the solution was chosen because of the urgent need to provide the public with information regarding the food we consume and the effect it has on our environment. The first objective is closely linked to the first research question of this paper (RQ1). The second objective was chosen because of the need for further research about the difference of information acquisition between delivering information through game mechanics and text.

Design and development

The two versions of the game will be designed following the “The ‘I’s’ Have It” framework established by Annetta (2010). According to the framework a serious game designed with the intent to educate should contain the following six elements:

Identity Description_​:

The player should feel a sense of identity within the game in-order to facilitate the immersion of the game and motivate the player. Players get more motivated and engaged with the content if they are allowed some form of choice of how the character that they are going to control will look (Annetta. 2010).

Implementation:

The game will utilize a customization system which will allow the player to buy, choose and combine a wide variety of skins and hats. The purpose of this content is to facilitate a sense of identity within the game. Giving the player the choice of customizing their “avatar” has been done before showing promising results with regards to player immersion and engagement (Annetta and Holmes, 2008).

Immersion Description_​:

The players should be immersed in the game in-order to facilitate a state of flow for the player. Good immersion depends on many different variables, Annetta (2010) suggests using Csikszentmihalyi’s (1990, pp. 48–67) eight flow characteristics as a guideline when designing serious games. Some of which are having an identity, getting clear feedback, good usability, and feeling that the goal of the game is attainable.

Implementation_​:

The game will as previously stated utilize a customization system to facilitate a sense of identity. The game will focus on easy controls to move the player around (good usability), ingame actions will always result in some kind of visual feedback (clear feedback), this will be true for both versions of the game although version “A” will have more visual feedback than version “B”.

Interactivity Description:

Social engagement is an important part of learning, Annetta (2010) therefore suggests that serious game should contain some socially engaging elements.

The game will have an online highscore board displaying the names, scores and characters of the players with the top ten highest scores.

Increasing complexity Description:

The difficulty of the game should ideally scale with the growing competence and skill of the player. If the game does not become more challenging the player will become disengaged with the game.

Implementation_​:

As the game progresses the difficulty of the game will increase to make the game more challenging. Another layer if increased complexity will be the high score board which will be tougher to reach as the already set high score increases. The customization system, which features buyable skins and hats, could arguably also be seen as a layer of increased complexity. The argument behind this is the fact that there will be some skins and hats which are more expensive than others, getting these skins and hats will require more commitment from the player than getting the less expensive ones.

Informed teaching Description:

Informed teaching is about utilizing the computer to record the players actions and based on those actions give the player some form of feedback.

Implementation:

Data such as score will be collected and later displayed as feedback to the player using the high score screen.

Instructional Description_​:

There must be some embedded learning elements within the game. When the players are immersed within the game, the learning content can be “stealthy”. Meaning that it does not have to be explicitly stated what the players are supposed to learn.

Implementation:

The goal of the game is to survive as long as possible and this can only be achieved by consuming food. There will be five types of food present in the game, ruminant meat, legumes, poultry, tropical fruit and wheat. All types of food awards the player with energy (allowing the player to survive a bit longer) but also releases some greenhouse gases. The embedded learning element is the fact that some food types releases more greenhouse gases than others and choosing the right food types is vital for survival.

Demonstration

The artefact will be demonstrated with the use of A/B testing. A/B testing is a method used to compare different versions of an artefact with regards to some given metrics. This method fits the purpose of this paper well since the secondary goal is to see if there is any difference between delivering information through the use of game mechanics and text (Desurvire and El-Nars, 2013). The two versions of the artefact will be tested with the help of university students as the main test group. The specific group of testers was chosen because of the design framework which was designed to target users from grade 5 through graduate school (Annetta, 2010). The goal is to have at least 3-5 testers testing each distinct version of the game, for a total of 6-10 testers. Two pilot studies will also be conducted, one for each version of the game, this is done to ensure that the questions are understandable.

Pre- and post-questionnaire will allow for an easy comparison between the two test-groups and will be used to gather data from the testers since it is a widely practised and tested method (Font, Hedvall, and Svensson 2017)(Guillén-Nieto and Aleson-Carbonell 2012)( Molnar and Kostkova, 2013).

Before the testing can begin, the users will fill out a questionnaire (Appendix A) to test their prior knowledge about the emission rates of different food types. Half of the testers will get to play version “A” of the game and the other half will get to play version “B”. After the testing has been carried out, the users will fill out another questionnaire (Appendix B) to see if the users have acquired any new knowledge.

The tests will be conducted over the internet, with the help of communication software such as “Skype” and “Discord”, with one tester at a time. Two pilot studies will be conducted, one for each version of the game, with the purpose of receiving feedback on the test itself. The results from the two pilot tests will be excluded from the final results. The testers who participate in the pilot studies will be made aware that they are allowed to give feedback and opinions at any time.

The questionnaires will be comprised of three main questions. The purpose of the first question is to find out some basic knowledge about the testers, such as do they believe in global warming, do they follow any specific diet. The second question is comprised out of ten sub-questions, each one addressing a specific food type. The testers are asked to place each food type within one of six emission intervals with regards to their co2 emissions per kcal. If the testers answer correctly their answer is marked as correct and they are awarded one point, if the testers chose a neighbouring interval, their answer is marked as partial and they are awarded half of a point. Otherwise their answer is marked as incorrect and they are awarded zero points. The third question is similar to the second question but only addresses five food

types. Another difference is that the third question asks the testers to place the food types in intervals with regards to their co2 emissions per protein content.

After the second questionnaire has been completed, interviews will be held with the testers, the purpose of the interviews is to get more in-depth insight about what knowledge the testers acquired by playing the respective versions of the game. The questions asked during the interviews can be found in Appendix C.

Explanation of how points are awarded for the second and third question

Correct _{​- The testers are awarded one point if they manage to place the food type within the} correct interval and the answer is marked as “correct”.

Partial _{​- If the testers manage to place the food type in a neighbouring interval of the correct} interval they are awarded half of a point and the answer is marked as “partial”.

Incorrect _{​- If the testers does not manage to place the food type with the correct interval or a} neighbouring interval they are awarded zero points and the answer is marked as “incorrect”. Evaluation

During the evaluation phase the data collected during the demonstration phase will be compared between the two test groups. The data from the pre-questionnaire (Appendix A) will be compared with the data from the post-questionnaire (Appendix B), this will be done for both groups. This is done with regards to the primary goal of evaluating the use of serious games in order to raising awareness for the impact different food types has on the environment. After the pre- and post-questionnaires have been compared, the results will be compared between group A and group B. This is done with regards to the secondary goal of evaluating if there is any difference in how much is learnt by delivering information through game mechanics and by delivering information through text. There will also be a comparison between the food types which appear within the game and the food types which does not appear within the game.

Communication

During the communication phase the results from the demonstration and evaluation phase will be presented and a discussion with regards to the research questions. This is achieved through this paper.

2.2 Pilot-Tests

Two pilot tests were conducted in order to ensure that all of the questions were clear and to avoid misunderstandings during the real test sessions. Both testers were made aware that they

would get to participate in the pilot study and that they were allowed to give feedback and opinions at any time during the test session.

2.2.1 First Pilot-Test

The first pilot-test was conducted over Discord in swedish. The tester was made aware that this was only a pilot study and that she could give feedback and opinions at any time. The tester filled out the pre-questionnaire and then played the “game mechanics” version of the game. The tester played the game for a total of 14 minutes before filling out the post-questionnaire. Immediately after the interview was conducted.

During and after the test session the tester raised the following concerns with the testsession: ● To little information was given about how to play the game prior to the test.

● It was hard to understand what all of the food-types were in swedish, mainly “Ruminant meat” and “Legumes”

● The difference between “Temperate fruits” and tropical fruit needs to be clarified. The feedback was addressed immediately in the following ways:

● The pre-test script was revised to include a more detailed explanation about how to play the game. This include an explanation about which keys are used to move the player character and an explanation about the energy and emission part of the game. ● More information was added in both of the questionnaires about which food types are

included in each category. For example “Ruminant meat” was changed to “Ruminant Meat (Beef, Goat, Lamb etc)” , “Legumes” to “Legumes (soybean), “Temperate Fruit” to “Temperate Fruit (Apple, Pear, Grape) and “Tropical Fruit” to “Tropical Fruit(Avocado, Kiwi, Lemon)”.

2.2.2 Second Pilot-Test

The Second pilot-test was conducted over Skype in swedish. The tester was made aware that this was only a pilot study and that he could give his feedback and opinions at any time. The tester filled out the pre-questionnaire and then played the ”text“version of the game. The tester played the game for a total of 11 minutes before filling out the post-questionnaire. Immediately after the interview was conducted.

During and after the test session the tester raised the following concerns with the testsession: ● “Greenhouse gases in co2 eq” was hard to understand and should probably be

explained.

● The game would probably be more immersive/better with sound. The feedback was addressed in the following ways:

●

● The sound feedback was taken into consideration, but no changes were made to the game.

2.3 Method Discussion

The design research methodology by Peffers et al (2008), offers a clear iterative process which enables the design, development and evaluation of an artefact. In order to answer RQ1 and RQ2 there is an absolute need for the development and evaluation of an artefact, this makes the design research methodology into a suitable method for this study.

2.3.1 Ethical Considerations

Each tester will be made aware that their answers to the interview questions and pre- and post-questionnaires will be used and published in this paper. However no personal information about the testers will be made public and throughout this paper the testers will remain anonymous. The testers will have to chose for themself whether or not to opt-in for data collection while playing the game. The data collected, in and of itself, is only related to ingame activities and do not contain any personal or sensitive information (except for the ingame name chosen by the testers).

3. The Artefact

The artefacts that has been produced during this study is two versions of a serious game designed in accordance with “The ‘I’s’ Have It Framework” established by Annetta (2010). The framework argues that serious games used in education should contain a specific set of elements, all of which are presented in greater detail in section 2.1.

Both versions of the game were developed for PC using the Unity game engine. The choice of which engine to use for the development of the game fell upon Unity because of previous experience and ease of use.

Version “A” of the game focuses on conveying information using game-mechanics. It has more in-game visual feedback than version “B” of the game. Version “A” also utilizes an interactive loading screen, where the player can deduce information about the attributes of the different foods in the game by interacting with them. Version “B” of the game has less

in-game visual feedback and only utilizes a text based loading screen. Other than these factors the two versions of the game are practically identical.

Figure 3.1 shows the interactive loading screen from version “A” of the game.

3.1 Design and Framework

The game was designed using the ‘I’s’ Have It Framework established by Annetta (2010). The framework is explained in greater detail under section 2.1.

3.1.1 Character Design

The playable character in the game is an elephant, this was an active design choice made with the utmost consideration. The motive behind using an elephant is the fact that elephants are commonly known for their good memory, hence the old saying “an elephant never forgets”. The fact that elephants are not carnivorous animals in real life made the choice even more compelling since the goal of the game can only be attained by consuming as little meat as possible, which fits well with the real behaviors of elephants.

3.2 Game-Mechanics Main game loop

The game lets the player control an elephant searching for food in a river. There are five different types of food in the river, ruminant meat, legumes, poultry, temperate fruit and wheat. The player consumes energy while moving and regain energy when eating the food that floats in the river. Each food type awards the player with a specific amount of energy but also releases a specific amount of greenhouse-gases. The player is also awarded with points each time it consumes any type of food.

Once the game has ended the player gets a final score based on the points collected during the game session. The player is awarded in-game currency, which it can spend in an in-game shop, based on the score that it gets. The awarded currency ranges from ten to one hundred. The longer that the player manages to stay alive the harder the game gets. The ratio of ruminant meat floating in the river increases overtime forcing the player to either be very conservative with the energy it got or release more emissions in-order to regain its energy. The purpose of making the game increasingly harder is to avoid having the player become disengaged and bored with the game (Annetta. 2010).

The player is awarded a specific amount of money at the end of game, based on the score attained, which can be spent buying new hats and skins for the elephant. The players with the best scores are shown in a scoreboard allowing for competition among the playerbase. The scoreboard will give the players a sense of community, allowing the players a form of social interaction within the game. To reach the top of the scoreboard will be increasingly harder as the player-base grows, adding another layer of increased complexity to the game.

Annetta (2010) argues that more complex games requires a higher degree of interactivity and since this game is not very complex the limited social aspect of the game should still be enough to make the player feel connected to one another.

Figure 3.3 shows an example the high-score screen.

End Conditions

There are two possible ways for the game to end. Either the player runs out of energy or the player releases too much emissions.

The colour of the environment gives the player a clue as to how high the emission levels are. This is done by lerping the colours of the vegetation between bright green to a orange/brown using the emission levels as a gradient. The green color represent an alive and healthy environment and the orange/brown environment represents a dry/hot environment in which the vegetation cannot thrive.

The amount of emission that the player is allowed to release is based upon the 5 (Gt CO2 -eq/yr) limit presented by Willett et al. (2019). The 5 GT is divided by ten billion (the same amount of population used in the calculations by Willett et al. (2019)) and then once again divided by 365 to represent what one individual is allowed to release during one day. Which comes down to about 1.3 kg of CO2.

The players energy-level is displayed as a health-bar ranging from bright yellow if the energy is full and black if the player has run out of energy. If the energy drops to zero the game will end.

Figure 3.4 shows the environment at its starting condition (left) and its end condition (right).

3.2.1 Customization System

The player has the option to customize the elephant by buying new skins and hats in an in-game shop. The game currently features fourteen different skins and twelve different hats which can be combined with each other. Making it possible for the player to attain 168 unique looking elephants. The currency that the player can use to buy new skins and hats has to be earned through playing the game, giving further incentive to play the game. A game can yield between ten and one hundred ingame currency. The cost of the skins and hats range between seventy-five and fifteen-hundred ingame currency.

The purpose of this content is to give the player a sense of identity making the player more involved and motivated to play the game and also complying with the first element mentioned (Identity) in the ‘I’s’ Have It Framework (Annetta. 2010).

Figure 3.5 displays the “Zebra-Phant” skin which can be purchased within the game.

3.2.2 Learning Elements

The information conveyed to the player in the game is that different food types has different co2 emission rates per kcal (energy) that they provide. Ruminant meat and poultry releases more greenhouse gas emission per kcal than legumes, temperate fruit and wheat does (Tilman and Clark, 2014). The information is conveyed in a stealthy way in version “A”, meaning that it is never explicitly stated how much co2 each food type emit. Instead the player must make the connection by them self while playing the game. To guide the player into making these connections there are enhanced visual effects, when the player consumes any food-type a red skull and smoke appear. The scale of the skull is proportionate with the amount of co2 that the food-type releases, this mechanic is only implemented in version “A” of the game. In version “B” of the game the player can explore the properties of each food-type in the loading screen by hovering the mouse pointer above the food type which will display an information box with the properties of the food type .

Another mechanic that has been implemented in order to guide the player into making the desired connection, the vegetation in the background changes colour, from a bright green to a light brown/orange tone as the emission levels rises. Once the emission levels are too high the players loses, further helping the player to make the connection between the different emission levels of each food type. The emission levels are also displayed in the player

interface using a emission-bar that fills up each time the player consumes any food. This is true for both versions of the game.

3.2.3 Data Collection

If the player gives their consent to data collection, data will automatically be collected from each game and stored in a google sheet. The data is collected to enable the high score screen and statistics screen to function correctly. The data is collected with the intent to give the player enhanced feedback on their performance as well as to foster a sense of community (the high score screen).

4. Results

In this section the results from the study will be presented. 4.1 Data-Presentation

In this section data from the actual study will be presented. A total of nine people participated in the study out of which one were female and eight were male. The answers from one of the participants will be disregarded as a result of him not understanding the questions, this brings down the total presented answers to eight. The age of the participants ranged from 20 to 36 years with a median of 22 years. All participants answered that they believe in global warming and when asked if they followed any specific diet all except one answered that they eat everything. The one that answered differently is vegan.

4.1.1 Pre- and Post-Questionnaire

Question 1: _{​How much of the global greenhouse gas emissions do you think the global} agricultural and food production is responsible for?

There were no significant difference between the answers given in the pre-questionnaire and the answers given in the post questionnaire for question one. Only one tester (mechanics version) changed their answer from “31-40%” to “>50%”, the rest remained the same. This was true for both the mechanics version and the text version. (The correct answer to this question was “21-30 %” (Willett et al., 2019) )

Question 2: How much greenhouse gases (in co2 eq) does the following food types emit per

kcal?

All but one tester improved their score between the pre- and post-questionnaire, the one that did not improve their score, answered exactly the same in both questionnaires hence scoring

the same both times. The testers increased their score from an average of 5,5 points to 6,68 points, an increase of 1,18 points or +21,1% (see table 4.1).

The testers who played the text version gained a slightly worse score in both questionnaires than those who played the mechanics version but improved their score between the pre- and post-questionnaire slightly more than those who played the mechanics version. The testers who played the text version scored on average 4,83 points in the pre-questionnaire and 6,16 points in the post questionnaire, giving an average increase of 1,33 points or +27,5%. The testers who played the mechanics version scored 5,9 points in the pre-questionnaire and 7 points in the post-questionnaire giving an increase of 1,1 points or +18,6% (see table 4.1). “Tester 2” saw the greatest improvement, scoring 5 in the pre-questionnaire and 8,5 in the post-questionnaire, giving an improvement of 3,5 points or +70% (see table 4.1).

“Tester 6” had the best score of 8,5 points in both the pre- and post-questionnaire, but given the increase of 0 points they had the worst improvement of all testers (see table 4.1).

As a whole the testers improved their score for seven out of the ten food types (tropical fruit, maize, vegetables, ruminant meat, legumes, temperate fruit and wheat). The testers gained a worse score for two food types (pork and sugar) and the score for the last food type (poultry) remained the same (see figure 4.1 - 4.2).

Table 4.1 Summary of score for all testers from question 2

Tester Pre-Score Post-Score Change

1 5,5 6,5 +1 (+18%) 2 5 8,5 +3,5 (+70%) 3 5 5,5 +0,5 (+10%) 4 4 5,5 +1,5 (+37%) 5 6,5 7,5 +1 (+15%) 6 8,5 8,5 +-0 (+-0%) 7 4,5 5 +0,5 (+11%) 8 5 6,5 +1,5

(+30%)

Average 5,5 6,68 +1,18

(+21,1%)

Figure 4.1 - Answers from all testers given for question 2 in the pre-questionnaire. Fully correct answers are marked as green, partially correct answers are marked as yellow and incorrect answers are marked as red.

Figure 4.2 - Answers from all testers given for question 2 in the post-questionnaire. Fully correct answers are marked as green, partially correct answers are marked as yellow and incorrect answers are marked as red.

Excluding the food types which does not appear in the game

When excluding the food types which does not appear in the game and only looking at those who are present in the game (ruminant meat, legumes, poultry, temperate fruit and wheat). Six of the testsers improved their score between the pre- and post-questionnaires and two testers scored the same amount of points both times. The testers increased their score from an average of 3,06 to 4,18 an increase of 1,12 points or 36,6%. The testers who played the text version of the game gained a slightly worse score in both questionnaires than those who played the mechanics version, but improved their score between the pre- and post-questionnaire slightly more than those who played the mechanics version. The testers who played the text version scored on average 2,16 in the pre-questionnaire and 3,83 in the post-questionnaire, giving an average increase of 1,67 points or 77,3%. The testers who played the mechanics version scored on average 3,6 points in the pre-questionnaire and 4,4 points in the post-questionnaire giving an increase of 0,8 points or +22,2%.

As a whole the testers improved their score for four out of the five food types, the score for the last food type (poultry) remained the same (see figure 4.1 - 4.2).

Excluding the food types which appear in the game

When excluding the food types which appear in the game and only looking at those who are not present in the game (tropical fruit, pork, maize, vegetables and sugar). Three of the testsers improved their score between the pre- and post-questionnaires while three of the testers gained a worse score between the pre- and post-questionnaires. The remaining two testers scored the same amount of points both times. The testers decreased their score from an average of 2,56 to 2,5, an decreased of 0,06 points or 2,3%. The testers who played the text version of the game gained a slightly better score in the pre-questionnaires than those who played the mechanics version, but gained a worse score in the post-questionnaire. The testers who played the text version scored on average 3 in the pre-questionnaire and 2,33 in the post-questionnaire, giving an average decrease of 0,66 points or -22%. The testers who played the mechanics version scored on average 2,33 points in the pre-questionnaire and 2,6 points in the post-questionnaire giving an increase of 0,27 points or +11,5%..

As a whole the testers improved their score for three out of the five food types (tropical fruit, maize and vegetables). The testers gained a worse score for the remaining two food types (pork and sugar) (see figure 4.1 - 4.2)

4.1.2 Interviews

Each playtesting session was followed by a individual interview.

Regarding the first question (_{​What do you think the goal of the game was?​)​, five of the testers} (tester: 1, 2, 3, 4 and 5) answered that the goal of the game was related to educated/affect/teach/create understanding about the relationship between food and carbon emissions. Three of the testers (tester: 6, 7 and 8) answered that the goal was related to show/enlighten people that meat is bad for the environment/ releases more co2 than beans etc (see table 4.2).

Regarding the second question (_{​Do you think the game tried to convey any type of} information?_{​) all of the testers answered that they thought that the game tried to convey some} type of information. Two of the testers (tester: 1 and 2) thought that the information that the game tried to convey was related to what we eat has an impact on the world and that we need to reflect about the choices we make when it comes to food consumption. Six of the testers (tester: 3, 4, 5, 6, 7 and 8) answered that they thought that the game tried to convey that meat is bad. Three of the testers (tester 4, 6 and 8), also thought that the game tried to convey that meat has higher co2 emissions than plant based food (see table 4.3).

Regarding the third question (_{​Do you believe that you gained increased knowledge about the}

properties of different food types? _{​)​, three out of the testers answered that they thought that} they had gained a better understanding regarding the different properties of different food types. Four of the testers answered that they had not gained an increased understanding and one tester did not know whether or not they had gained an increased understanding or not (see table 4.4).

Table 4.2 Answers given by all testers to interview question 1

Question:

_​“​

Tester

Answer

1

2

3

4

5

specifically carbon dioxide emissions._​”

6

7

8

Table 4.3 Answers given by all testers to interview question 2

Question:

_{​ “​}

Tester

Answer

1

2

3

4

5

6

7

affects the environment. I don’t know if that counts as information, information sounds more specific. It did not convey any information if you mean exact numbers._​”

8

Table 4.4 Answers given by all testers to interview question 3

Question: “

_​

Tester

Answer

1

2

3

maybe?._​”

4

5

6

7

I do not know how much carbon dioxide meat emits, I just know that it emits a lot. Maybe it would have been better with numbers, kind of like the emission meter been in numbers instead? That it increases by a number every time you pick something up._​”

8

It feels like it should be a lot from a case-by-case basis, I understand that beans require a lot of resources to produce. Different cows are treated differently and should also release different amounts of carbon dioxide. Then beans and fruits are also sprayed (with pesticides), which may not be good for the environment. I feel that I haven't got a bigger picture of it all. Then I also think that I have a pretty wide knowledge base, so the game is maybe a little too basic or low-level for someone with my education. It probably had to go deeper into the subject to give me something._​”

5. Discussion

In this section the results from the study will be discussed and evaluated with regards to the previously stated research questions. _{​A discussion of the study's limitations and weaknesses} will also be presented further down in this section.

The data generated from the pre- and post-questionnaire suggests that the testers gained some increased awareness regarding how much emission different food types releases. Out of ten tested food types, seven yielded a positive change in score between the pre- and post-questionnaires out of which four appeared in the game while three did not. One food type yielded the same score and two food types showed a decrease in score between the pre- and post-questionnaires, both of which did not appear within the game. There were some minor differences in the results between the two versions of the game, overall the testers who played the text version saw a larger increase in score (1,33 points) than the testers who played the mechanics version (1,1 points). These results were in line with the findings of Molnar and Kostkova, (2013) who also saw very little difference between conveying information through text and game mechanics in four out of five learning objectives evaluated.

One interesting notion of the results from the pre- and post-questionnaires is the fact that almost all of the increase in score (1,12 points out of 1,13 points) came from food types which were present in the game. The reason as to why this is the case is because both of the food types, which yielded a score decrease between the questionnaires, belonged to the category “food types which does not appear within the game”. This caused a substantial decreased in the average score for the aforementioned category. The two food types in question were pork and sugar, one possible explanation as to why pork yielded a negative result could be the fact that it is similar to ruminant meat in terms of appearance and taste. This could have prompted the testers to associate it with a similar emission rate as that of ruminant meat, such a emission rate is significantly higher than the actual emission rate of

pork. Looking at the test-results from the questionnaires this seems to be the case. The reason as to why sugar yielded a decrease in score between the questionnaires, is not as clear. It could be due to a combination of things, one of which could be the fact that it, as a food type, contains a lot of kcal, making the co2 emissions per kcal very low. This combined with a lack of knowledge of the actual co2 emissions might have forced the testers into making uneducated guesses regarding to the co2 emissions per kcal. Enforcing this claim is the fact that no food type similar to sugar appeared within the game, leaving the testers to figure out the answer on their own. When looking at the answers given in the post-questionnaire, sugar shows the largest answer spread of all the tested food types (answers fell within four different intervals), this could offer some support to the claim that the testers had to guess while answering. Regardless of what the reasons, as to why the two food types yielded a decrease in score, might have been, the fact remains that almost all of the increase in score (1,12 points out of 1,13 points) came from food types which are present in the game. This could suggest that the testers gained some increased knowledge about the food types which are present in the game, but could not apply the knowledge on a more general basis. This notion raises the question whether or not an overall increase in awareness was achieved or if the testers only gained some factual knowledge about the different foods which are present in the game. The data from the pre- and post-questionnaires cannot on its own answer the question of whether or not the game raised the environmental awareness of its testers. Therefore we must also consider the results from the interviews which yielded some very interesting answers. The answers given to question one ( _{​What do you think the goal of the game was?) and two}

(_{​Do you think the game tried to convey any type of information?​) ​during the interviews,}

suggests that the testers understood the purpose of the game, that being “trying to raise awareness for the effect different food types has on the environment”, yielding answers such as:

“_{​Understanding that it is not sustainable to eat as we do and what the climate impact the} food we eat has...” (_{​See table 4.2, tester 2)}

“T_{​o affect us players to think about the relationship between what you eat and their} emissions.” (_{​See table 4.2, tester 4)}

​“​To educate people about agriculture and environment, specifically carbon dioxide

emissions._{​” ​(​See table 4.2, tester 5)}

Such answers give a strong indication that the testers had a clear picture of what information the game tried to convey. However even though the pre- and post-questionnaires showed an increased score with seven out of eight testers, only three out of eight answered that they thought that they had gained increased knowledge about the properties of different food types. Four testers answered that they had not gained increased knowledge and one answered that they did not know whether or not they had gained increased knowledge. Three out of the

four that answered that they had not gained increased knowledge during the final question, motivated their answers with a lack of challenge and feedback from the game, giving statements such as:

“_{​No, I already knew most things. If I had received a fact after I filled out the pre and post} (questionnaire) then maybe I would know if I have gained an increased understanding, but now I do not know how well I performed...”_{​ (see table 4.4, tester 6) .}

“_{​No, I already knew that meat emits more than vegetables and I haven't been told what the} difference is between them. There is a huge skull when you consume meat, but that does not tell how much worse meat is than vegetables. It just means that meat is worse. ” _{​(see table}

4.4, tester 7).

“_{​No I do not think so. I did not change my answers to the questions in the different forms, the} first time I answered that meat releases more and I did so in the second form as well…” _​(see

table 4.4, tester 8).

“...._{​the game is maybe a little too basic or low-level for someone with my education. It} probably had to go deeper into the subject to give me something.” _{​(see table 4.4, tester 8).}

These statements suggests that there might have been issues with the design of the game, not being challenging enough and not giving the players a satisfying amount feedback. Another common theme among the four testers who answered that they had not gained increased knowledge, was that they perceived that they already knew what the game tried to inform them about. One could argue that trying to raise awareness about an issue with those who are already aware of said issue is a futile task. On the other side of the spectrum, the three testers who answered that they had gained increased knowledge gave statements which are in line with the core message of the game. Such as:

“Absolutely, a little higher understanding that meat and animal products emit more greenhouse gases than vegetable foods. I knew it was high, but didn't think it was that high”

(see table 4.4, tester 2)

​“​Yes, I would say that. I would probably say that beans and wheat seem to be the best for the environment. And meat, specifically beef is worst for the environment based on your game. _​”

(see table 4.4, tester 4)

​”​Yes, I think so. If you get the same energy from food that does not emit as much, then you should eat that, if you want to save the world._{​” (see table 4.4, tester 5)}

The results from the questionnaires and interviews suggests that the testers might have gained a greater understanding for the emissions of different food types, however it is hard to

determine to what extent this holds true. It is also unclear whether or not the testers gained a general awareness or only acquired some factual knowledge.

The study suffered from a number of limitations and weaknesses which could have had an impact on the final results. One of the most significant weaknesses of the study is the limited numbers of testers and the uneven distribution of them between the two test groups (Three testers playing the text version and five testers playing the mechanics version). Any follow-up study should consider utilizing a larger group of testers and try to distribute them more evenly. It is also possible that the study targeted the wrong target group, the design might have been more suited for a younger audience. Another possible limitation of the study is the two versions of the game being too similar to one another, with only minor differences setting them apart. It is very possible that this is the reason as to why the results from the pre- and post-questionnaires yielded so similar results between the groups. The last major limitation has to do with how question three in the pre- and post-questionnaire was formulated. The purpose of the question is to find out if the testers know how much co2 each food type emit per gram of protein (co2 divided by gram of protein). However the game never conveys the protein content of any food type, placing huge importance on the testers pre-knowledge of the different food types protein content. Another problem with question three, which made the results from the question useless, was due to human error. While constructing the question wrong intervals where set, which may have had a significant effect on the results. This made the results from question three useless with regards to the purpose of this study, hence they have been excluded.

One suggestion for a possible improvement on the design of the game/study could be to incorporate the pre- and post-questionnaires within the game. This could lead to a more streamlined testing-process, allowing the testers to allocate more time on playing the game instead of on opening/finding the questionnaires. How this might have affected the results are unclear and could be the target for future research.

The study yielded some very promising and interesting results suggesting that serious games could be used to raise awareness for sustainability and environmental issue. There were minor difference in the results between the text version and the mechanics version. These results were in line with the findings and conclusion made by Molnar and Kostkova, (2013), whom also saw very little difference in their study. Overall the results from the questionnaires and the interviews suggest that the testers might have gained a greater understanding for the relationship between food and emissions. The study partially confirmed the research of Janakiraman, Watson and Watson (2018) who came to the conclusion that serious gaming could produce a positive attitude towards environmental sustainability, however further research is required in order to make any real conclusion.