Correlating personality with player behaviour and playstyle in single player platformer games

Correlating personality

with player behaviour and playstyle

in single player platformer games

by

Dimitrios Dimitriadis

10734546

DkDimitriadis@gmail.com

June 2016, Amsterdam

MASTER INFORMATION STUDIES : GAME STUDIES

FACULTY OF SCIENCE

1

_Supervisor

₂

_Supervisor

Sander Bakkes

_{Frank Nack}

Abstract

This user study aims to investigate and establish possible correlations between player personality and in-game behaviour and tendencies. A total of 20 participants were asked about their game preferences (whether one is a regular gamer or not) while in addition they took a personality test questionnaire based on the Five Factor Model which in return yielded a personal profile. Moreover, specific game actions had been exclusively mapped to their respective personality traits. These two datasets were then examined using correlation analysis methods in order to establish correlations between the two. The game used is a single player platformer game which affected the amount of traits that were measurable, effectively narrowing the scope of this paper to three personality traits: neuroticism, openness and conscientiousness. The results of the present investigation revealed that neuroticism was consistently positively correlated with in-game actions that stemed from frustration and negative emotions. Openness and conscientiousness were also correlated but the correlation values varied depending on the subject group in question.

General Terms: Modeling, Correlation, Experiment, Measurement

Keywords: Player behaviour, Five factor model, Personality traits, Solo games

1 Introduction

When creating games, one of the most important goals of the designer is to ensure that the player is enjoying himself by finding the ideal balance between boredom and frustration [26]. By ensuring that the game is fun and captivating, developers can hope for higher player retention and game replayability, characteristics that are important for both entertainment and serious games. However, there are still debates in the field regarding the types of players, their behaviour patterns and the elements that people perceive as enjoyable. Researchers have already tried to taxonomize players into categories. One of the earliest and most known examples is seen in Figure 1, where Bartle [14] placed players into four categories according to their in-game behaviours.

Figure 1: The Bartle player types

Bartle has noted that his model mainly applies to MUD (Multi User Dungeon) games and that it is by no means a universal model. Even though a universal model would provide an excellent design tool for game developers it does not seem feasible at the time of writing this paper. With the exception of [19] that made a valuable contribution towards a unified model for player behaviour and personality, most studies that tackle such questions are focused on specific genres or even specific games [13][20] [21][22].

Moreover, it is also academically significant to see if there is a relation between the players’ personalities and their reaction to being challenged in a game. From empirical observations, we see that players that encounter challenges, either see them as a positive trait and become more engaged or, on the contrary, become frustrated which leads to player defection. However, the previous statement is likely to vary when the game genre in question changes.

This paper examines how individuals with different types of personalities act within a specific game, what goals they set for themselves and how they behave in challenging levels. The research question deriving from this is:

What are the correlations between a player's personality traits and the in-game behaviour in the scope of solo platformer games that are challenging?

In order to answer the research question the first action is to select a game that is, by its own nature, challenging and find people that are willing to participate in the study. Once the participants are found they can then be interviewed and also asked to take a personality test using the widespread Five Factor Model. The first action is important as it will help to create meaningful subgroups which will be useful during the analysis process. The personality test will create a personal profile for each participant comprised of independent variables. This profile will serve as the pivot point of this research as it will allow for a linear regression analysis to take place.

Following that, specific in game actions are mapped, one to one, to their matching personality traits of the Five Factor Model. This model enables the in-game actions to be mapped and effectively quantifies the specific playstyle of a participant. Finally, linear regression is used to examine the possible correlations between the personality and the playstyle of a player.

The objective is to be able to come to meaningful and supported conclusions about what do players want and expect from a game based on their type of personality. By achieving this, we provide more tools for studies and researches revolving around games. Some examples include: game designers can create games centered around the users that are playing them rather than trying to address larger crowds, academics that are studying the effect of games on people and their behaviour, businesses that want to increase company and employee relations or productivity.

2 Related Research

Understanding gamers is important both for the game designers as well as businesses [15]. Most of the researchers that have attempted to tackle this task have used models in order to taxonomize players into categories. One of the most intuitive models to understand gamer behaviour is to distinguish players between hardcore and casual [27]. Hardcore gamers are people that own a lot of games, know game terminology and conventions, gaming is part of their lifestyle and enjoy challenges. On the other hand, casual players play fewer games, are not very knowledgeable about them, play to pass time while mainly looking to have fun [16]. However, while this subtle approach can be usefull (designers can simply prompt the players to select a game difficulty), gamers are a wide variety of people and trying to fit them into two categories is not sufficient. To overcome this challenge, researchers make use of personality models in order to have a better understanding and a more thorough analysis of their experiment group.

According to [23], personality “refers to individual differences in characteristic patterns of thinking, feeling and behaving”. The study of personalities can be separated into two fields. One is to examine the differences in specific types of behaviour such as approachability or confidence. The other is to understand how all these traits are combined into one for every individual. Creating a personality profile is a more complete and diverse method to evaluate a target audience. This means collecting information about people's’ behaviour, talents, sensations, intuitions, preferences, thinking patterns, social traits and other personal indicators. From a personality profile it is easier to make assumptions and predictions about how individuals might react to events they encounter as well as what elements they will enjoy more. This method has been used in various fields, for instance in medicine ,business and marketing.

One paper that has already tried to tackle the challenge of correlating preferences and perceived difficulty of games to the player's personality traits based on the Five Factor Model is [20]. This model will also be used in the experiment. A more thorough description of it follows in a later part of this paper. In the experiment conducted by [20], a total of four game genres were taken into account and two games per genre were used. “Conscientiousness was negatively correlated with perceived ease of first-person shooter games” and extraversion and agreeableness were positively correlated to the liking of dancing games. While the number of correlations that were discovered is relatively low, it still shows that people with similar personality traits are likely to share the same preferences and perspectives in games. Additionally, the results should serve as a warning that perhaps there are game genres that, by nature or by the limitations of the research, will not provide positive results.

One of the most thorough studies concerning personality and playstyle was done by Bateman and Boon [16]. It is one of the few recent studies that do not utilize the Five Factor Model. The authors chose to use another well know personality model, the Myers-Briggs Type Indicator (MBTI) as well as the hardcore-casual gamer model which was mentioned previously. Based on these models and using a questionnaire to note the player's’ gaming habits, Bateman and Boon were able to taxonomize the players into four basic playstyle categories: Conqueror (winning and challenge driven), Manager (process oriented with strategy or tactical components), Wanderer (search for enjoyment through fun and novelty), and Participant (Story and social driven). One of the most common criticisms of the Bartle type models is that no one is just one type of player. The demographic game design model shown in Figure 2, provides descriptions for six possible modes of play and effectively fills in the gap between the rest of the player types.

[21] attempted to utilize the Five Factor Model to investigate how vulnerable players are to violent video games. By using an elaborate spherical model to predict the effects of these type of games, it was found amongst other things that when the FFM traits were examined individually, none had a great impact on the investigated subject. It was when the traits were combined that the first correlations started to appear. Therefore, it is possible for a correlation to exist between playstyle and a set of psychological traits instead of a single one.

Griebels [22] focused his research on how the player's avatar and playstyle is connected to their personality. Using the popular game The Sims 2 and the Five Factor Model (by the means of a questionnaire) Griebels discovered that neuroticism, openness and conscientiousness are the traits with the most significant correlation values between personality and in-game behaviour. Results showed that parts of the subject’s personality were cast in-game through their avatars. However, it is noteworthy that these findings are specific for a simulation game (The Sims).

Based on the aforementioned, a viable approach to taxonomize players is to make use a personality model and attempt to built a gamer classification model onto it. Personality models have been in use for a long time and have concrete and trustworthy foundations. The most used personality model in recent studies and the one this study chooses to use is the five factor model. Moreover, most of the studies that used a specific game genre (e.g Bartle [14] focusing with multiplayer dungeon games and Griebels [22] focusing on simulation games) note that their results

should not be considered universal and clearly state the game genre that their study was based. Genre and game style limitations apply to this paper as well which will use only a platformer game. Because of this and for reasons that will be expanded later on, the scope of this paper focuses on three of the five personality traits of the Five Factor Model.

Figure 2: Bateman and Boon player taxonomy; A demographic

game design model filling in the gaps between the Bartle player types.

3 Methodology

3.1 Participant selection

The participants of this study were selected from the social and professional group of the author. They were either classmates, work colleagues or personal friends. These three groups ensured a disciplinary variation between the participants as well as a variety in gaming experiences including hardcore gamers, casual gamers and non-gamers. However, this also meant that age profile of the subjects was mainly people in the twenties or early thirties as no older people were surveyed. It should be also noted that all of the participants have completed a college degree or an even higher form of education , were single and had no children.

3.2 Experimental Procedure

To begin with, all volunteers had to take an online personality test, the Five Factor Model Test [19]. This test returns a percentage for each of the five factors based on each person's answers to a set of 50 questions. The participants were asked to rate their own personality using the scale shown in Figure 3. Additionally, subjects were asked about their gaming history and their difficulty preference (easy,normal or hard).

Figure 3: Scale used in personality questionnaire

The next step was to choose a game to act as the medium that would allow to observe and document the gaming behaviour of the subjects. Preferably, the game has to have simple mechanics and clear instructions in order to achieve a low entry level for the players. A 2D platformer named SuperMeatBoy was selected, since it complies with these requirements. Additionally, there is a large amount of people that are already familiar with this game genre due to the fact that it includes a lot of famous franchises with mainstream characters (e.g Mario and Sonic). A platformer is a type of game that involves controlling an avatar and overpassing various challenges of each level which amongst other things include jumps, obstacles, enemies, puzzles and so on. At first these games were mainly played in arcade machines but over time and with the technological advancement they found a new home in computers, handheld mobile devices and almost every other medium. Below, in Figure 4, is the first level of SuperMeatBoy where the player can get accustomed with the controls.

Figure 4: The first level of SuperMeatBoy showing the player

avatar in red (bottom left) and the level end as the pink avatar (center of the screen)

The experiment subjects played levels of the first world of the game (20 available levels plus the final level which is the boss), without any interference or assistance from outside factors. A large percentage of gamers were able to complete all levels whereas non gamers managed to complete approximately half of them. Each level is won by reaching its respective end. Additionally, some levels include optional elements such as: portals (which initiate a mini game that if won awards the player with a new character) or bandages (which can be spent to acquire new characters). It is notable that most of the subjects were surrounded by friends, coworkers or acquaintances although the number of spectators never was very high. Their play session was recorded using a screen recording software named Bandicam [28] and this video was later used to analyze each subject’s playstyle. This method was preferred from observing and taking notes while the game session was occurring because it was observed during the experiment that the latter approach caused many players to stress every time they died or whenever an observation was written down. It is expected that not all players will reach the final level, but all sessions will have at least 15 minutes of active play time.

3.3 Data analysis procedure

All of the defined in-game actions were attributed to strictly one factor of the big five model. While watching the recording of the play session, each time an action or its opposite occurred it was

noted under its respective factor. Finally, a score was attributed for each factor of the Five Factor Model ranging from -3 to 3. This scale was selected because it was narrow enough to represent the summary of the actions that were observed and because it has 0 as its center point (all playstyles were initiated as “neutral”). The previously described procedure yielded a set of values for each factor of the Five Factor Model as a result. This dataset along with the dataset that originated from the questionnaire were analysed using linear regression analysis in order to investigate for possible correlations between the two. The former acted as the dependent variable while the latter was used as the independent variable. In regards to the group of participants, first it was analysed as a whole and then separately, for gamers and non gamers, which added up to three groups in total.

The first output of this analysis was a Pearson correlation coefficient to help determine the degree of linear dependence of the observed playstyle and the personality test scores. Additionally, scatter plots were created for the three factors that are the focus of this thesis: openness, conscientiousness and neuroticism. These scatter plots helped to visualize the nature of the relationship and the type of correlation between the two variables by including a line of best fit as well as an identity line (with a slope of 1) to serve as an observational point.

4 FFM in relation to in game behaviour

By the end of the 20th century, most evidence was suggesting that almost all of the current personality models could be reduced to include only 5 factors. A prevalent model, named Five-Factor Personality Model, included the following personality traits: openness, conscientiousness, extraversion, agreeableness, and neuroticism [2]. Since then, the Five Factor Model has been found to be able to generalize across cultures [3] [4] and it does not seem to deteriorate with the passage of time [5]. Table 1 summarizes the model used in this study; a mapping of personality traits to their respective in game actions.

FFM trait Expected playstyle

Openness • Experiment with pathing instead of memorizing tactics

• Explore the level instead of focusing on completing it Conscientiousness • Pause on challenging tasks and

devise a plan of actions

• Focus on completing all active tasks encountered in a level Extraversion • Not applicable

Agreeableness • Not applicable

Neuroticism • High levels of frustration and stress caused by in game failures • Go on a state of “tilt” and select

sub optimal strategies and paths

Table 1: Game actions mapped to their respective personality

traits

Openness to experience: This variable expresses how receptive a

person is to new experiences and stimuli. People with high scores are unconventional and intellectual (not necessarily smart), with

imagination and appreciation for the arts and philosophy [8] [10]. They will not hesitate to hear and accept new ideas and will prefer to have a wide range of options given to them. Low scorers are not very open to new ideas and prefer situations that they have dealt with before rather than facing something new. People that score high on openness are almost always open to accept new ideas and are not likely to stick to a single strategy or path. In the game, once a method that has a high success rate is discovered by the players (for instance, jump a wall and immediately wall jump across) it is expectable of them to keep using that method. However, since there are multiple paths or ways to reach the end of a level or an item, subjects that score high on this category are expected to continue experimenting with their movements in order to explore alternative pathing that could be either more enjoyable, challenging or efficient.

• Subjects are expected to experiment with their pathing to try various different ways to reach a point instead of memorizing a working tactic that works most of the time.

Additionally, since high scorers of this category welcome new stimuli, we expect them to actively explore levels for hidden elements and not try to navigate straight to the finish line. It should be noted that this should not be interpreted as wanting to collect and achieve everything but rather as wanting to explore and learn more about their surroundings [13].

• Subjects are expected to deviate from the original goal of finishing the level in order to explore each or some of the levels but don't fixate on them. (Exploring is not the same as acquiring).

Conscientiousness: Conscientious people tend to plan their

actions in advance and are not likely to be spontaneous. According to [8] conscientiousness is “manifested into three relating facets: achievement orientation (hardworking and persistence), dependability (responsible and careful) and orderliness (planful and organised)”. More specifically, people that are high scorers will always prefer to have a plan and will not choose to immediately jump into action [7]. These are also people that depend their success on various variables of a scenario to remain constant.

Based on the abovementioned facts, it is fairly obvious that people who score high on this factor will not rush into the level. On the contrary, they are expected to take a pause and consider their actions upon encountering a part of the level that they find challenging, in order to construct a plan of actions and try execute them instead of rushing in and hoping that their reflexes, and maybe some luck, will help them advance forward.

• Conscientious people are expected to stop upon encountering a part of the game that looks challenging and consider their next moves.

Moreover, throughout a level it is possible for multiple tasks to be initiated at the same time, most likely though they have to be completed separately ( for instance upon warping on a portal the player is transported out of the level). Conscientious individuals are characterised by a crave for achievement, persistence and focus on tasks. They do not wish to leave objectives unfinished or behind [13]. It is likely that they will try to finish any task they have started; whether that is finishing a level, getting to a portal or collecting a bandage.

• Conscientious people are expected focus on completing the tasks they have at hand, whether they are trying to finish the level of get a bandage/portal.

Extraversion: Extraversion is related to positive emotions and is

a trait of people who like to be in social situations and are often characterised as “the life of the party”. It is a prominent factor of personality models, even before the emergence of the Five Factor Model [8]. Extraverted people are impulsive and active people, talk assertively and engage with others around them [9]. Additionally, they are energetic people who enjoy being social, are generally less dysphoric and not so worried about themselves [8]. It is noteworthy that scoring low at this category should not be associated with the opposite of the characteristics that were mentioned before [7].

In solo games, like the one used for this experiment, it will be difficult to assign expected actions that correspond to this factor, since there are no game controlled characters (NPCs) or other players that the subject can interact with. Additionally the scope of the research was decided to be about enjoying the game and not about competing with each other meaning that elements like a high score tables are also not included. The absence of social elements also solidifies the role of neuroticism as the pivotal factor of this experiment which is in line with the challenging nature of the selected game. Therefore, the lack of social and competitive elements in the game and the experiment leaves no possible actions that could be correlated with this trait.

• Due to game and experiment constraints there are not in game actions that can be correlated with the personality score of extraverted people.

Agreeableness: Individuals that score high on this aspect are

expected to get along with others and generally be friendly, approachable people that are easy to collaborate with [8] [11]. They also like to be helpful and even compromise for the sake of the social harmony [8]. A high score is also related to empathy and being eager to assist as well as receive assistance.

Similarly to extraversion, it is hard to define the actions that agreeable people would likely perform in the game, mainly because social aspects and interactions are not a part of this game due to its solo nature. There is no part in the game where a player works together or interacts with game elements that have a personality. Also the lack of spectators or other social game elements will not provide the ground for the subjects to fully indulge their agreeableness and how they interact socially. It should be noted that some subjects did play surrounded by friends or colleagues, however, this was not true for all participants. Therefore it would not be feasible to incorporate these interactions onto this model because the experiment variables would not be equal for all the subjects.

• Due to social surrounding differences and in game constraints there were no actions that could be correlated with the the personality score of agreeable people.

Neuroticism: This trait is generally related towards negative

emotions such as stress or fear and defines the emotional stability of a person. According to [5] neuroticism is “the most pervasive trait across personality factors and it is prominent in nearly every measure of personality”. People with high levels of neuroticism are more prone to negative moods, can experience anxiety, insecurity, depression and can be easily irritated [8]. Low scorers

on the other hand have a better time when dealing with tense situations.

Translating these traits into player behaviour, it is logical to assume that these players will get easily frustrated with game elements that they find difficult to overcome. Consistent failures on the same part of a level could occur from various scenarios. The players might fail to recognise what actions they need to perform in order advance through a challenge or they might be unable to translate a plan of actions that they devised into the game due to poor mechanics. Moreover, the insecurities and negativity of neurotic players will kick in when they die or when they fail at a challenge multiple times, causing stress and discomfort.

• High scorers on neuroticism are expected to get frustrated with various elements of the game that are hard to overcome and stress over failures due to them taking them game too seriously.

Finally, it has been found that “Under conditions of no slack capacity or short-time horizons (which produce stress) the decision process will resemble crisis decision-making-resulting in significant implementation errors” [12].A similar effect has been observed during this experiment where subjects are trying the same tactic repetitively even though it is clear that it does not help the player to advance forward (for instance, trying to jump over a wall that is higher than the highest jump of the avatar). This is similar to a state known as tilt in competitive online gaming and poker. Tilting is term used to describe an emotional state in games which causes players to use less optimal strategies that they would normally would due to confusion or frustration [25].

• High scorers on neuroticism are expected to try the less optimal paths or strategies due to them being in a state of tilt even when it is visible that it they are not working.

5 Results

The play sessions took place either at the home of the subject or at the authors work space when the subject was a colleague. The completion of all experiment procedures provided two datasets and a set of personal information for analysis. The first dataset was the result of the personality questionnaire and consisted of the results of each participants per personality factor, in percentages. The second dataset was the observed playstyle from the screen recordings of the play sessions. The observations range between 3 and -3 and are directly mapped to the first dataset. The personal information consisted of name, age, video game preferences such as experience with platformer types of games and preferred difficulty.

5.1 Experiment group profile

A total of 20 people took part in the experiment comprising of 8 females and 12 males. Their age spanned from 21 until 33 years old and are presented in Figure 5. Additionally, half of the participants were or have been, at a point in their lives, active video gamers, whereas, the other half of the participants never had a good relationship with video games. Surprisingly, given the popularity of platform video games, only one non gamer has ever played one, whereas all the questioned gamers have had experience with this genre. In regards to the participants preferred difficulty, seven (35%) participants generally prefer to play in easy mode, twelve (60%) in normal and only 1 (5 %) person has

answered that he prefers hard difficulty modes.

Figure 5: Age of participants

All of the gamers have played platformer games, however, none of them had already been acquainted or played SuperMeatBoy, meaning that none of the subjects had any advantages compared to the rest of the group. A handful of gamers though, had already heard about the game from friends or read about it on the internet. Regarding the average playtime per week of gamers, two of them play one hour per week, five play around five to ten hours per week and the rest three gamers do not play anymore, but used to do so in the past. All non gamers did not have a regular gaming schedule. Instead they responded that they play rarely, usually when prompted by a friend.

The main part of the analysis consists of correlating the test results from the questionnaire with the observed playstyle of the participants. This analysis will use on three of the five factors of the Five Factor Personality Model, namely openness, conscientiousness and neuroticism. As mentioned previously, the very nature of platformer games does not allow for them to be used as a reliable medium to evaluate the remaining two factors, agreeableness and extraversion since the social elements needed for these traits were missing from the game and its environment, virtually and physically. In regards to the analysed groups, the sample group is analysed firstly as a whole and afterwards, the same analysis is performed on the gamers and the non-gamers groups.

5.2 Data analysis

At the beginning of the analysis the first logical step was to calculate the pearson correlation coefficient for all applicable possible scenarios of the experiment. This would allow to get a first glimpse of the stronger and the weaker correlations. The results are presented in the following Table 2.

Whole group Gamers Non Gamers Openness r = −0.0229 r = 0.0807 r = −0.1965

Con/ness r = 0.0195 r = −0.3511 r = 0.5373

Extraversion X X X

Agreeableness X X X

Neuroticism r = 0.5126 r = 0.3088 r = 0.1851 Table 2: Pearson correlation values for the whole group (n=20),

the gamers group(n=10) and the non gamers group(n=10)

Notable values that stand out from this table are the two highest positive correlations; neuroticism for the whole group and conscientiousness for the non gamers group. Neuroticism is consistently positively correlated amongst all groups in question. On the other hand openness and conscientiousness have their correlation values fluctuate between positive and negative signs as the focus swifts between the groups.

In regards to the whole experiment group, openness and conscientiousness are not significantly correlated with the observed playstyle Figure 6 & 7. On the contrary, it can be seen on Figure 8 that neuroticism is highly correlated (r = 0.51). It is also worth mentioning that openness to experience had the highest test results, in regards to personality test scores, with a mean of 80.3 percent when compared to conscientiousness and neuroticism who have means of 63.9 and 41.44 respectively. This is visible below in Figures 6, 7 and 8 where most points in the first graph are close to the right side whereas in the following two graphs the points are more placed closer to the center.

Figure 6: Openness graph; high personality test scores but no

correlation.

Figure 8: Neuroticism graph; positive correlation

Following the non-gamers group, openness is negatively correlated (r= -0.2) with observed playstyle Figure 9. Conscientiousness and neuroticism are positively correlated (r = 0.53 and r = 0.18 respectively) as shown on Figure 10 & 11. Similarly to the previous scatter plots of the whole group, openness personality test values are again the highest followed by the other two traits.

Reflecting the change of the Pearson correlation coefficient, the slopes of the trend lines of openness and conscientiousness are actually steeper in comparison to the whole group plots. Openness is fairly more negative and conscientiousness has switched signs and is a little above 45 degrees of slope. On the contrary, the trend line of neuroticism has come closer to the X-axis. Regarding neuroticism, excluding two points, all non-gamers had an observed playstyle values of 2 or 3 which are forcing the trend line to be above the X-axis.

Figure 9: Openness graph for Non-Gamers; negative correlation

and high personality test scores

Figure 10: Conscientiousness graph for Non-Gamers; high

positive correlation

Figure 11: Neuroticism graph for Non-Gamers; positive

correlation and high occurrence of neurotic actions.

Finally, in the gamers group, openness was not significantly correlated (r = 0.08) Figure 12, conscientiousness is negatively correlated (r= -0.35) as shown in Figure 13 and neuroticism, similarly to the first two graphs, is again positively correlated (r= 0.31) in Figure 14.

Personality test scores for openness and neuroticism are high and low respectively. The scores of conscientiousness are mostly located in the middle, between 55 and 75 percent. It is also noteworthy that the trend line of conscientiousness has almost been inversed when compared to the one of the non gamers group. Neuroticism is still positively correlated but games had much less neurotic actions in their playstyle in comparison to non gamers.

Figure 12: Openness graph for Gamers. No observed

correlation.

Figure 13: Conscientiousness graph for Gamers; Scores

between 55 and 75. Correlation trend line is almost inversed.

Figure 14: Positive correlation but low occurrence of neurotic

actions

6 Discussion

The focus of this paper was to discover correlations among personality traits and peoples in-game behaviour. As mentioned before, two personality traits of the Five Factor Model, extraversion and agreeableness, were not measurable due to the solo nature of the game used for the experiment. Nevertheless, the analysed data indeed suggests that the examined personality traits of openness, conscientiousness and neuroticism can be correlated towards in-game behaviour.

Some of these relations surfaced in certain groups of the participants, however the trait of neuroticism was found to have consistent results for the experiment group as a whole. The game that was used for this experiment is known for being challenging. Perhaps with an alternative and more forgiving game the trait of neuroticism could surface different results. In this experiment, neuroticism was positively correlated with game actions that have been marked as neurotic such as: repeatedly try tactics that were not effective or stress more about their game performance. Finally, it is worth mentioning that non gamers had a significantly higher neurotic observed playstyle compared to the group of gamers (1.7 mean compared to -1.4 mean respectively).

Conscientiousness was the trait that demonstrated the highest fluctuation between the groups of gamers and non-gamers. More specifically, gamers correlated negatively while non-gamers correlated positively. At this point it should be noted that, as expected, gamers had increased mechanical and motor skills compared to non gamers and were not challenged as much as the latter group. This could likely be there reason for the above results. Gamers were able to be more relaxed and enjoy the game without focusing too much on all the possible tasks. Non gamers were focused on their performance even though they were instructed that performance is not measured in this study, and were more reluctant to skip levels.

Openness was the trait with the least amount of significant correlations amongst the examined traits. For the total amount of participants and for the group of gamers the correlation coefficient value was approximately zero. The non gamer group of this study though, yielded a negative correlation value. Due to low amount of information for this personality trait not much else could be concluded about it.

In conclusion, this study provided insight into a field that has not been widely researched, a person’s playstyle mapped to their respective personality traits. The latter, served rightfully as an independent variable, since it has solid foundations, theories and a variety of usable models. Even though this study should be considered as an exploratory pilot study there are still solid conclusions to be made. Neuroticism could be considered as the pivotal factor since the game that was selected was aimed to surface neurotic tendencies from the players. It was the only trait that had consistent results amongst the whole scope of the data

analysis. This suggests that similar studies can focus on simple games with clear focuses and goals if they wish to center their research around specific aspects of the personality.

6.1 Limitations

There is a handful of limitations regarding this study. Initially, while the sample size of 20 people was adequate to discover some significant correlations, it is generally accepted that larger sample sizes can yield better results, provide greater statistical power and enable researchers to discover smaller correlations. Additionally, the sample is, in its entirety consisted of young adults of 20 to approximately 30 years old. Their demographic background is also quite similar since all of the participants have acquired a university or a higher level education degree, however the type of each person's field of study did vary among them.

Furthermore, there is a risk of the participants gameplay being altered by the fact that they are aware of the fact that they are taking part in a research and their play session will be recorded and later analysed. It should be mentioned that none of the subjects were aware of what personality traits the study aimed to correlate and what in-game actions were taken into account. Nevertheless, there were cases where the Hawthorne effect [24] could be observed. For instance, subjects admitted that they felt slightly apprehensive about skipping a level of the game, mainly because they felt that it will affect their study score, despite the fact that they were informed that success or failure of a level is not in this research's scope.

Additionally, there was a large gap in terms of mechanical and motor skills when comparing the best and the worst performing subjects. While the in-game performance did not matter, the aforementioned gap effectively minimizes the effect of the game challenges for the skilled players and on the other hand maximizes it for the less skilled players.

Finally, another limitation of this research which has already been mentioned before in this paper is the very nature of platformer games. More specifically, the lack of social in-game elements prevents this study from exploring all five factors and instead focuses on openness, conscientiousness and neuroticism.

6.2 Future research

Seeing the limitations of this study, future researchers interested in the relation between personality and in-game behavior should make an effort to include more games of the same genre in order to find consistencies among them and be able to derive a more general overview for that genre in question. Games, with their plethora of different elements, have become a rich medium used for entertainment, education, marketing and many other applications. Even if the mechanics are the same for games of the same genre, a change in other elements like aesthetics or storyline can have a significant impact on players and affect their choices

and playstyle. Alternatively, and for the same reasons, it can also be beneficial to allow the players to select the type of game that they want to play to increase the chance of them enjoying the play session.

Moreover, during the course of this study, another limitation that was noticed was that the used method for observing play style was too upfront and probably created a bias, influencing the actions of the participants. In order to get more true/precise results, the subjects should not have been aware that their play session is going to be analysed. This would minimize the stress induced from the process and the added feeling of being evaluated and would hopefully result in the subjects playing naturally and enjoying the game instead of focusing on succeeding.

It was mentioned previously in this paper, that researchers who examined the effects of violent video games [21] had trouble figuring out correlations in their analysis until they created a model that combined the subjects personality traits into a model that unifies them instead of examining each one individually. The previous method can be beneficial in the scope of this research because it is likely that such an approach could reveal additional and hopefully stronger correlations.

7 Aknowledgements

I would like to thank the following people for the guidance, support, advice, feedback and realization of this study:

• Sander Bakkes • Frank Nack

• Bente Rosan de Vries • Danai Kostoula • Shoshannah Tekofsky • Pieter Spronck • Isabelle Lamers

8 References

[1] Costa, P.T.,Jr. & McCrae, R.R. (1992). Revised NEO Personality Inventory (NEO-PI-R) and NEO Five-Factor Inventory (NEO-FFI) manual. Odessa, FL: Psychological Assessment Resources.

[2] Goldberg, Lewis R. "An alternative" description of personality": the big-five factor structure." Journal of personality and social psychology 59.6 (1990): 1216. [3] McCrae, Robert R., and Paul T. Costa Jr. "Personality trait structure as a human universal." American psychologist 52.5 (1997): 509.

[4] Pulver, Aleksander, et al. "A Big Five personality inventory in two non‐Indo‐European languages." European Journal of Personality 9.2 (1995): 109-124.

[5] Costa, Paul T., and Robert R. McCrae. "Personality in adulthood: a six-year longitudinal study of self-reports and spouse ratings on the NEO Personality Inventory." Journal of personality and social psychology 54.5 (1988): 853.

[6] Costa, Paul T., and Robert R. McCrae. "Four ways five factors are basic." Personality and individual differences 13.6 (1992): 653-665.

[7] Costa, Paul T., and Robert R. McCrae. "Neo PI-R professional manual." (1992): 653-65.

[8] Judge, Timothy A., et al. "The big five personality traits, general mental ability, and career success across the life span." Personnel psychology 52.3 (1999): 621-652.

[9] Watson, David, and Lee Anna Clark. "Extraversion and its positive emotional core." (1997).

[10] en.wikipedia.org/wiki/Big_Five_personality_traits [11] Barrick, Murray R., and Michael K. Mount. "The Big Five personality dimensions and job performance: A meta-analysis." (1991).

[12] Hrebiniak, Lawrence G., and William F. Joyce. Implementing strategy. New York, NY: Macmillan, 1984. [13] van Lankveld, Giel. Quantifying Individual Player Differences. PhD thesis, Tilburg University, The Netherlands, 2013.

[14] Bartle, Richard. "Hearts, clubs, diamonds, spades: Players who suit MUDs." Journal of MUD research 1.1 (1996): 19.

[15] Zammitto, Veronica Lorena. Gamers' personality and their gaming preferences. Diss. Communication, Art & Technology: School of Interactive Arts and Technology, 2010. [16] C. Bateman and R. Boon, 21st Century Game Design, Hingham, MA: Charles River Media, 2005.

[17] A. Rollings and E. Adams, Andrew Rollings and Ernest Adams on Game Design, Indianapolis, Ind.: New Riders Games, 2003

[18] T. Hartmann and C. Klimmt, “The Influence of Personality Factors on Computer Game Choice,” Playing Video Games: Motives, Responses, and Consequences, Mahwah, N.J.: Lawrence Erlbaum Associates, Inc., 2006, pp. 115-131. [19] Goldberg, Lewis R. "The development of markers for the Big-Five factor structure." Psychological assessment 4.1 (1992): 26.

[20] Wang, Yu-Chi, Michael Bradlee Sutherland, and Kent L. Norman. "Relating Five Factor Personality Traits to Video Game Preference.

[21] Markey, Patrick M., and Charlotte N. Markey. "Vulnerability to violent video games: a review and integration of personality research." Review of General Psychology 14.2 (2010): 82.

[22] T. Griebel, “Self-Portrayal in a Simulated Life: Projecting Personality and Values in The Sims 2,” International Journal of Computer Game Research, vol. 6, December 2006.

[23] Kazdin, Alan E. "ENCYCLOPEDIA 0P PSYCHOLOGY." (2000).

[24] McCambridge, Jim, John Witton, and Diana R. Elbourne. "Systematic review of the Hawthorne effect: new concepts are needed to study research participation effects." Journal of clinical epidemiology 67.3 (2014): 267-277. [25] Griffiths, Mark, et al. "Online poker gambling in university students: Further findings from an online survey." International Journal of Mental Health and Addiction 8.1 (2010): 82-89.

[26] Csikszentmihalyi, Mihaly, and Isabella Selega Csikszentmihalyi, eds. Optimal experience: Psychological studies of flow in consciousness. Cambridge university press, 1992.

[27] Fritsch, Tobias, Benjamin Voigt, and Jochen Schiller. "Distribution of online hardcore player behavior:(how hardcore are you?)." Proceedings of 5th ACM SIGCOMM workshop on Network and system support for games. ACM, 2006.

Appendix

Personality test questions.

These questions are derived from the essential characteristics of each factor of the Five-Factor personality model and are used in multiple online versions of the personality test . This specific question set is provided for free by Truity Psychometrics LLC.

I often feel blue.

I feel comfortable around people. I believe in the importance of art. I have a good word for everyone. I am often down in the dumps. I make friends easily.

I tend to vote for liberal political candidates. I believe that others have good intentions. I am always prepared.

I dislike myself. I don't talk a lot. I have a vivid imagination. I make people feel at ease. I pay attention to details. I have frequent mood swings.

I am skilled in handling social situations. I carry the conversation to a higher level. I respect others.

I get chores done right away. I panic easily.

I am the life of the party. I enjoy hearing new ideas.

I accept people as they are. I carry out my plans. I rarely get irritated.

I know how to captivate people. I am not interested in abstract ideas. I have a sharp tongue.

I make plans and stick to them. I seldom feel blue.

I have little to say. I do not like art. I cut others to pieces. I waste my time.

I feel comfortable with myself. I keep in the background. I avoid philosophical discussions. I suspect hidden motives in others. I find it difficult to get down to work. I am not easily bothered by things.

I would describe my experiences as somewhat dull. I do not enjoy going to art museums.

I get back at others.

I do just enough work to get by. I am very pleased with myself. I don't like to draw attention to myself.

I tend to vote for conservative political candidates. I insult people.

I don't see things through. I shirk my duties.