=Paper=
{{Paper
|id=Vol-2359/paper19
|storemode=property
|title=A statistical analysis of Steam user profiles towards personalized gamification
|pdfUrl=https://ceur-ws.org/Vol-2359/paper19.pdf
|volume=Vol-2359
|authors=Xiaozhou Li,Chien Lu,Jaakko Peltonen,Zheying Zhang
|dblpUrl=https://dblp.org/rec/conf/gamifin/LiLPZ19
}}
==A statistical analysis of Steam user profiles towards personalized gamification==
https://ceur-ws.org/Vol-2359/paper19.pdf
A statistical analysis of Steam user profiles towards
personalized gamification
Xiaozhou Li, Chien Lu, Jaakko Peltonen, and Zheying Zhang
Tampere University
Kalevantie 4, 33100, Tampere, Finland
{xiaozhou.li, chien.lu, jaakko.peltonen, zheying.zhang}@tuni.fi
Abstract. Gamification is widely used as motivational design towards enhanc-
ing the engagement and performance of its users. Many commonly adopted game
design elements have been verified to be effective in various domains. However,
the designs of such elements in the majority of the target systems are similar. Due
to inevitable differences between users, gamification systems can perform more
effectively when users are provided with differently and personally designed fea-
tures according to their preferences. Many studies have suggested such require-
ments towards personalizing gamified systems based on the users’ preferences,
with categorizing gamification users and identifying their preferences as the ini-
tial step. This study proposes a preliminary analysis of the factors that catego-
rize user preference in a game community, based on the user profiles data of the
Steam platform. It shall not only facilitate understanding of players’ preferences
in a game community but also lay the groundwork for the potential personalized
gamification design.
Keywords: Gamification · Exploratory Factor Analysis · Steam · User Profile ·
Preference · Personalized Gamification.
1 Introduction
Gamification, commonly defined as the use of game design elements for non-game
contexts [12], has been widely adopted as motivational design to support users moti-
vation enhancement and performance improvement. Many game design elements, e.g.,
badges/achievements, points, leaderboard, progress, story, etc., have been adopted in
various service domains and proven effective in many studies [14]. However, the ma-
jority of the gamification systems provide very limited alteration towards different users
but adopt the one-size-for-all design approach instead [32]. Such rigid gameful designs
are to a certain extent ineffective in persuading the users into positive behaviors. Many
studies have shown that different users are likely to be motivated by different game el-
ements and persuasive strategies [31, 32, 40]. Therefore, it is critical to understand dif-
ferent users’ preferences when providing them the personalized gameful experiences.
The studies on the users’ types and preferences regarding gamification systems are
based on the similar studies on game players. A seminal study on the player types for
multi-user dungeon (MUD) games is Bartle’s player typology [2]. Meanwhile, a num-
ber of studies also contribute to extending the user typology framework by focusing
GamiFIN Conference 2019, Levi, Finland, April 8-10, 2019 217
on psychographic and behavioral aspects [15]. Even though the direct connection is
not addressed, such studies on player typology do facilitate the understanding of users
preference of play style and their motivations of playing [15]. On the other hand, a
gamification-specific user typology framework is developed by Marczewski [26], who
proposes six gamification user types based on intrinsic or extrinsic motivational affor-
dances [36] and their different degrees for the users. Furthermore, based on this particu-
lar framework, a 24-item survey response scale is presented to score users’ preferences
regarding the six different types of motivation toward a gameful system, which can
therefore identify a users type and describe his/her preferences [42].
Despite the uniform well-defined player types and gamification user types, such a
‘clear-cut’ categorization approach can be questioned as a player may not belong to
a certain type strictly [15, 21]. In addition, limitations of using survey data towards
such categorization have also been recognized [42]. In this study, we focus on users
of the Steam platform and their community-related behaviors presented on their profile
pages. The users’ Steam profiles provide various information, including the games they
have, the game achievements, item trading, friends, groups, reviews, screenshots, profile
customization options, and so on. The objective nature and large volume of such data
shall has the potential to yield enhanced characterizations of users and their diferences.
Herein, based on factor analysis of large user profile data, we identify the factors that
characterize the differences between Steam users. Instead of a strict categorization of
players, the study aims to answer what are the factors that distinguish Steam users from
one another and determine their preferences, as well as how such distinguishing factors
can be applied to facilitate personalized gamification design.
The paper is organized as follows. Section 2 introduces previous studies on game
players and gamification user typologies and on analysis of the Steam platform and user
data. Section 3 introduces our data collection and analysis methods, Sections 4 and 5
present results and discussion Section 6 concludes.
2 Related Work
2.1 Player Types and Gamification User Types
The aim of segmentation in marketing is to identify different customer groups so that
they are served with products and services that match their unique needs. Studies on
player types also serve this purpose. The majority of the prevailingly cited studies focus
on the player segmentation in terms of the behavioral and psychographic attributes in-
stead of geographic or demographic ones [15]; our focus is similar, since our Steam pro-
files did not contain demographic/geographic attributes and we focused on the available
profile information reflecting player behavior. When available, our modeling principle
could accommodate demographic/geographic attributes as covariates.
Bartle’s seminal player typology — Achiever, Explorer, Socializer and Killer — is
based on the things people enjoy about MUD in either an action or interaction dimen-
sion towards either players or the game world [2]. It is also criticized for being dichoto-
mous and too simplifying, as well as focusing on only one game genre instead of a broad
range [3, 15, 42]. Extending Bartle’s typology model, many studies have proposed sim-
ilar typology models for online game players with specialized focuses [43, 45]. Many
GamiFIN Conference 2019, Levi, Finland, April 8-10, 2019 218
other studies present different ways of categorizing players based on their various moti-
vation and behaviors when not fixating on online games [21, 39]. Such player typology
models provide ways to detect the difference in players and their preference regarding
motivations and behaviors in general. On the other hand, many studies also focus more
specifically on players’ preferences regarding game design elements [11, 19].
The studies on gamification user types also adapt the results from the player typol-
ogy studies. Such studies are mostly supported by the research on behavior motivations
and personalities [29,36]. Regarding the user typology in the gamification domain, Mar-
czewskis gamification user type model is the most cited work [26]. Motivated by the in-
trinsic and extrinsic motivational factors of the users, which is defined by the Self Deter-
mination Theory (SDT) [35], Marczewski categorizes the users of gamification services
into six types, including socializers, achievers, philanthropists, free spirits, players, and
disrupters. Other studies also attempt to provide adapted typology frameworks regard-
ing specific domains [1,44]. Meanwhile, adapting Marczewski’s gamification user types
model, Tondello et al. present and validate a standard scale to determine users’ prefer-
ence towards gamification systems regarding different motivation types [42]. Based on
that, their subsequent works contribute to suggesting gameful design elements regard-
ing user preferences, personalizing persuasive strategies, and creating a recommender
system model for personalized gamification [32, 40, 41]. However, mentioned as their
limitation, the data are self-reporting and subject heavily to participants’ personal un-
derstanding of survey statements and preferences towards diverse game elements. Thus,
relevant objective data with a larger sample volume can address such limitation and can
also yield alternative results.
2.2 The Steam Platform and Users
Steam, a popular digital game distribution platforms, has drawn attention from the
academia. Becker et al. analyze the role of games and groups in the Steam community
and present the evolution of its network over time [5]. O’Neill et al. also investigate the
Steam community but focus on the gamers’ behaviors, in terms of their social connectiv-
ity, playtime, game ownership, genre affinity, and monetary expenditure [30], whereas
Blackburn et al. focus more specifically on the cheating behavior [7]. Many other stud-
ies also investigate the various perspectives of players’ behaviors on the Steam platform.
For example, Sifa et al. investigate the players’ engagement and cross-game behavior by
analyzing their different playtime frequency distributions [37,38]. Baumann et al. focus
on “hardcore” gamers’ behavioral categories based on their Steam profiles [4]. Lim and
Harrell examine players’ social identity and the relation between their profile maintain-
ing behaviors and their social network size [22]. Meanwhile, other scholars also study
the other perspectives of Steam, such as, recommender systems for its content [6], early
access mechanism [24], game updating strategies [23], game reviews [25], and so on.
However, research on characterizing players based on their Steam profile data towards
analyzing players’ preference to different game design elements is still limited.
GamiFIN Conference 2019, Levi, Finland, April 8-10, 2019 219
3 Method
3.1 Data Collection
A web crawler based on the Beautiful Soup Python module was created to collect data
from public user profiles. The data collection proceeded in a “snowball” manner. The
crawler started from one user’s Steam profile URL which was selected at random from
the top 10 Steam user leaderboard, and crawled the list of the user’s friends profile URL.
Iteratively, the list of users was grown via crawling the friends of each of the existing
users on the list and appending the results to the end of the list. Although guarantee-
ing an unbiased sample from such a huge base is difficult and our gathered dataset is
necessarily small, it can still achieve a good representativity. Duplicated profile URLs,
as well as private ones from which no valid data can be obtained, were eliminated.
To reduce crawling time while achieving reasonable coverage, only profile URLs were
crawled, and from the initial data pool of 2561387 unique user profile URLs, we col-
lected the profile information on a random subset of the URLs which includes 60267
users. The crawled features include Levels, Showcases, Badges, Number of Games,
Screenshots, Workshop Items, Videos, Reviews, Guides, Artworks, Groups, Friends,
Items Owned, Trades Made, Market Transactions, Achievements, Perfect Games, Game
Completion Rate, and four binary profile customization related variables: Avatar, Sta-
tus, Background, and Favorite Badge customization (customized or not). To summarize
the binary variables per user, we define an aggregate value called Profile Customization
whose value is the percent of ‘customized’ values: for example, if a particular user cus-
tomized three of the four items mentioned above, his/her Profile Customization score
will be assigned as 0.75. In addition, each user’s active time span was also collected
based on the time when the user last logged off and the time when the user created the
account, using the SteamAPI. To take the user activity into account, we further com-
puted the duration the profile had existed using the above-mentioned information and
utilized it to normalize the profile variables, by simply dividing each variable by the
profile duration.
3.2 Exploratory Factor Analysis
To uncover the underlying structures of the Steam user profiles, an exploratory factor
analysis (EFA, [13]) is conducted. It enables us to reduce the complexity of the data,
explain the observations with a smaller set of latent factors and discover the relations
between variables. Unlike clustering which discovers groups of players, EFA discovers
underlying axes characterizing players and their differences. In game culture studies,
EFA has been widely used especially in studies related to user/player types and user
motivations (e.g. [42, 43]). Extracted EFA factors can also be a basis for analysis such
as clustering (player segmentation) or prediction in follow-up work; we focus on dis-
covering underlying axes of variation in Steam user profiles through EFA and their
applications in gamification.
One common issue in EFA is how to decide the number of factors. In this paper,
the parallel analysis (PA) introduced by Horn [18] is adopted to solve the problem. It
has been widely used and has given good results in recent research works (e.g. [33,
GamiFIN Conference 2019, Levi, Finland, April 8-10, 2019 220
34]). Several comparative studies (e.g. [8, 46]) have shown that it is an effective way to
determine the number of factors.
Table 1. Result of Parallel Analysis
Factor Observed Eigenvalue Simulated Eigenvalue
1 3.104 1.031
2 2.744 1.025
3 1.650 1.021
4 1.382 1.018
5 1.167 1.015
6 1.130 1.011
7 1.073 1.008
8 1.027 1.006
9 0.916 1.003
In PA, the Monte Carlo simulation technique is employed to simulate random sam-
ples consisting of uncorrelated variables that parallel the number of samples and vari-
ables in the observed data. From each such simulation, eigenvalues of the correlation
matrix of the simulated data are extracted, and the eigenvalues are, as suggested in
the original paper [18], averaged across several simulations. The eigenvalues extracted
from the correlation matrix of the observed data, ordered by magnitude, are then com-
pared to the average simulated eigenvalues, also ordered by magnitude. The decision
criteria is that the factors with observed eigenvalues higher than the corresponding sim-
ulated eigenvalues are considered significant. Hereby, we conduct the parallel analysis
task with 5000 simulations to determine the number of factors.
To simplify interpretation of the factor analysis result, the varimax rotation tech-
nique [20] which maximizes the variance of the each factor loading is employed. Re-
sults with an alternative rotation approach promax [17] were similar.
4 Result
4.1 Factor Analysis
The result of the parallel analysis is shown in Table 1. Based on the mentioned criteria,
the turning point can be found easily by examining the differences between observed
eigenvalues and simulated eigenvalues. Since the simulated eigenvalue becomes greater
than the observed eigenvalue in the 9th factor (1.003 and 0.916 respectively), the first
8 factors are retained. The corresponding factor loadings can be found in Table 2. A
cross-loading of the variable Profile.Customization was found on Factor 1 and 7, we
further computed the Cronbach’s alpha [9] on those two factors to evaluate their internal
consistency and the values are found acceptable (0.87 and 0.71 respectively).
4.2 Factors Interpretation
Based on the result of EFA, we interpret each of the eight factors and summarize each
of the unique preference attributes of Steam users.
GamiFIN Conference 2019, Levi, Finland, April 8-10, 2019 221
Table 2. Loadings of the Extracted Factors
Variable Factor 1 2 3 4 5 6 7 8
Level 0.641 -0.005 0.004 -0.002 0.008 -0.013 -0.263 0.002
Showcases 0.026 0.107 0.065 0.828 0.162 0.180 0.028 0.067
Badges 0.954 0.033 0.004 0.010 0.006 0.043 0.016 0.004
Games 0.019 0.511 0.020 0.016 0.108 0.365 0.030 0.088
Screenshots -0.000 0.118 0.332 0.046 0.344 0.039 0.022 0.490
Workshop.Items 0.007 -0.045 0.042 0.127 0.789 -0.027 0.003 -0.082
Videos 0.002 -0.030 -0.066 0.046 -0.074 -0.022 -0.003 0.901
Reviews 0.002 0.232 0.039 0.044 0.769 0.039 0.018 0.113
Guides 0.002 0.024 0.879 -0.031 -0.090 -0.003 -0.001 -0.002
Artwork 0.004 -0.010 0.836 0.101 0.192 0.006 0.018 0.030
Groups 0.078 0.017 0.020 0.031 0.026 0.008 0.951 0.009
Friends 0.947 0.002 0.004 0.043 0.007 0.014 0.202 0.001
Items.Owned 0.004 0.048 0.005 0.049 -0.004 0.733 0.006 -0.022
Trades.Made -0.003 -0.142 -0.002 0.281 -0.063 0.551 0.003 -0.061
Market.Transactions 0.017 0.116 0.001 -0.063 0.044 0.645 -0.007 0.049
Achievements 0.005 0.865 0.014 0.125 0.014 -0.010 -0.001 -0.011
Perfect.Games 0.003 0.847 0.006 0.210 0.105 -0.045 -0.002 -0.017
Game.Completion.Rate 0.008 0.274 0.013 0.852 0.054 -0.004 0.003 0.021
Profile.Customization 0.808 -0.007 -0.008 -0.019 -0.015 -0.016 0.553 -0.007
Factor 1: Elite (Level, Badge, Friends, and Profile Customization) Factor 1 in-
dicates the users’ tendency to become the elite of the Steam community. The elite users
focus on their social comparison advantages over the others by enhancing their quantifi-
able social scores, such as, levels, badges, and friends numbers. According to Steam’s
unique mechanism, the users can upgrade their levels and earn more badges without the
requirements of exerting more effort in actual gameplay. Therefore, the elite users tend
to value their social achievement more than experiences in gameplay. In addition, they
also prefer profile customization in order to present their unique social identity.
Factor 2: Achiever (Games, Achievement, and Perfect Games) Users’ tendency
in Factor 2 indicates their preference towards mastering the games. They focus on com-
pleting games thoroughly and obtaining as many in-game achievements as possible.
They also tend to enlarge their game collection whenever possible. Compared to the
elite users, the achiever users prefer to put their effort in games and less in social.
Factor 3: Provider (Guides and Artworks) Users with high attribute in Factor 3
love to provide facilitation to the others with gameplay guides and self-created unique
game-related arts. Different from elite and achiever users who focus on their social
presence or achievement, the provider users tend to be more altruistic and care about
other users and their game playing.
Factor 4: Completer (Showcases and Game Completion Rate) Similar to the
achiever users, the completer users also focus on gameplay but less on achievements.
They prefer to finish the games that they start but have less intention of pursuing the
full achievement by investing extra amount of hours. Meanwhile, they like to show
their possessions, e.g., showcases, as much as possible, but put less effort on organizing
compared with the elite users.
GamiFIN Conference 2019, Levi, Finland, April 8-10, 2019 222
Factor 5: Improver (Workshop Items and Reviews) Users with high value on
Factor 5 focus on game improvement. They make efforts to add unique experiences to
games via workshop items and reviews. These encourage developers to improve the
games and publish better games in the future. Similar to provider users, they are also
altruistic but focus more on game quality.
Fig. 1. An Example of User Preference Attributes Radar Chart
Factor 6: Trader (Item Owned, Trades Made, and Market Transaction) The
trader users do not pay much attention to either games or social, but to buying and sell-
ing game related virtual items instead. According to Steam’s mechanism, users neither
have to own or play games to obtain items nor have to become friends with others or
join groups to make trades. Thus, trader users tend to make the community a business
playground, buying low and selling high.
Factor 7: Belonger (Groups and Profile Customization) Similar to the elite users,
the belonger users also tend to focus more on social interaction than gameplay, when
the difference is that the belonger users prefer the feeling of relatedness and belonging,
rather than social comparison. Belonging to social groups is always their first priority.
Having a proper customized profile is thus also necessary to fit them in the groups.
Factor 8: Nostalgist (Screenshots and Videos) Users with high nostalgist attribute
have the tendency of restoring their gameplay memories by taking screenshots and
recording videos. They also share their gameplay memories with others in the activ-
ity timeline, so that other players can enjoy the unique scenes and compare to their own
gameplay too. Meanwhile, the ”thumbs up” and appreciation from the others is their
reward.
It is worth noting that the eight factors aim to explore the various attributes of Steam
users instead of arbitrarily categorizing each user into a single type. Generally, each in-
dividual user shall contain certain scores in all given attributes while the attribute value
distribution of different users shall differ. Meanwhile, each user may also contain high
or low score in multiple attributes simultaneously. By reducing the variable dimensions
to one for each attribute and normalizing the value, each individual user shall have a
radar chart illustrating his/her salient attributes. Fig. 1 shows an example of a user who
possesses a salient attribute of improver and is creative with workshop items and also
loves to contribute in improving games by giving reviews. Meanwhile, this particular
GamiFIN Conference 2019, Levi, Finland, April 8-10, 2019 223
user also possesses relevantly salient attributes of elite, achiever, and provider. It indi-
cates that the user also favors gaining levels, badges, and achievements, and providing
guides and artworks to the community.
Table 3. An Example Mapping between Preference Attributes and Motivation Types
Attributes Steam Variables Motivation Types [10, 36] Gameful Elements [40]
Elite Level Mastery Progression
Badges Mastery Incentive
Friends Relatedness Socialization
Profile Customization Autonomy Customization
Achiever Games Mastery Progression
Achievements Mastery Incentive
Perfect Games Mastery Incentive
Provider Guides Mastery, Purpose Altruism
Artwork Autonomy Altruism
Completer Showcases Autonomy, Mastery Customization
Game Completion Rate Mastery Progression
Improver Workshop Items Autonomy, Purpose Altruism
Reviews Autonomy, Purpose Altruism
Trader Items Owned Mastery Incentive
Trades Made Relatedness Socialization
Market Transactions Relatedness Socialization
Belonger Groups Relatedness Socialization
Profile Customization Autonomy Customization
Nostalgist Screenshots Autonomy, Relatedness Socialization
Videos Autonomy, Relatedness Socialization
To apply such a preference framework in gamification design, based on the vari-
ables each attribute is related to, we could find connections between attributes and the
established intrinsic motivation types or other similar gamification design models or
frameworks. With different player motivation and design elements frameworks, the ap-
plication towards personalized gamification design could differ. Table 3 is an example
of connecting the obtained preference attributes with the SDT motivation types [10, 36]
and the gameful design elements categories [40]. Ideally, each Steam variable can be
mapped to a certain type of motivation and a particular gameful design element cate-
gory. Subsequently, the motivation that drives the corresponding preference attributes
and the related gameful design element set can be decided and weighted (e.g., based
on relatedness of the variables to the attributes). However, such presumption of con-
necting attributes, motivation types, and design elements can be subjective, when the
motivation of each user towards each individual Steam variable is unknown and hard to
be dichotomized. For example, ‘Level’ is likely to be driven by the motivation of mas-
tery, when, on the other hand, particularly in Steam, higher level means that the user
will have more badges and showcases to customize. Therefore, the ‘Level’ variable
GamiFIN Conference 2019, Levi, Finland, April 8-10, 2019 224
is driven by the motivation of autonomy, to some extent. Furthermore, a quantifiable
value of ‘Level’, together with ‘Badges’ and ‘Profile Customization’, can be also seen
as the tendency towards social comparison. Such equivocality shall be addressed with
potential ordering or voting schemes.
5 Discussion
Compared with Lim and Harrell’s study on players’ social identity [22], we cover more
perspectives of Steam users’ social behaviors in the gamer community by extending
the data collection to more features. However, different from Sifa et al.’s work [38] our
data covers only the Steam users’ profile information and not users’ in-game behaviors.
Thus, with the current dataset, mapping from the obtained user preferences towards the
gameful design elements regarding heavily in-game behaviors, such as, immersion or
risk/reward, is not possible [40]. Furthermore, based on the goal of this study to study
users’ preference regarding gamification design, the data limits generalization towards
all gamification users instead of only gamers. Despite the above limitations, the data
(similar to other product-oriented social media profiles, e.g. Amazon profiles) can be
seen as more generalized rather than focusing on gamers from specific games or genres.
Compared with previous studies on gamification user types [40,42], such data collected
from user profiles can be more objective than self-reported survey data.
This study presents a data-driven approach to investigating users’ preferences to-
wards game design elements. The resulting axes of variation among players can be in-
spected and used in gamification. In future work the results can also be used as a basis
for categorization of players; data-driven approaches [16] can improve efficiency and
representativeness compared to manually designed categories. One follow-up direction
is to build a collaborative filtering recommender system based on similarity of users’
preference towards various game design elements, allowing a personalized gamifica-
tion design based on the recommendation for each user [41]. Another future direction
is to validate the user preference framework with empirical analysis. For example, the
user preference scale of Tondello et al. [42] can be adopted as a reference, with Steam
users as participants. Furthermore, the data volume can be enlarged with more users,
e.g., by crawling from multiple seed users; our data could further be combined with
additional data regarding, e.g., players’ in-game behaviors, preference on game genres,
and reviews on games. After validation, the proposed user preference framework can
be applied to future data-driven player studies. Together with previous gamification de-
sign methods [27], the framework will facilitate gamification design and provides an
efficient way to address key issues in the user analysis phase [28].
6 Conclusion
We presented an exploratory way of analyzing user presences towards game design ele-
ments using Steam user profile data. Using EFA, eight factors/attributes are gained, the
value of which can be used to define each individual user’s preference regarding behav-
iors in the Steam community. Together with the connection between such behaviors and
the underlying motivation types and gameful design elements, each user’s preference
GamiFIN Conference 2019, Levi, Finland, April 8-10, 2019 225
regarding gamification systems can be also perceived. Due to the quantifiable and ob-
jective nature of the data, such estimation of the users’ preference can be more precise.
It will contribute to the future work of personalized gamification design and creation of
recommender systems for personalized gamification in a data-driven manner.
Acknowledgments. This research was supported by the Academy of Finland project
Centre of Excellence in Game Culture Studies (CoE-GameCult, 312395).
References
1. Barata, G., Gama, S., Jorge, J.A., Gonçalves, D.J.: Relating gaming habits with student per-
formance in a gamified learning experience. In: Proceedings of the first ACM SIGCHI annual
symposium on Computer-human interaction in play. pp. 17–25. ACM (2014)
2. Bartle, R.: Hearts, clubs, diamonds, spades: Players who suit muds. Journal of MUD research
1(1), 19 (1996)
3. Bateman, C., Lowenhaupt, R., Nacke, L.: Player typology in theory and practice. In: DiGRA
Conference (2011)
4. Baumann, F., Emmert, D., Baumgartl, H., Buettner, R.: Hardcore gamer profiling: Results
from an unsupervised learning approach to playing behavior on the steam platform. Procedia
Computer Science 126, 1289–1297 (2018)
5. Becker, R., Chernihov, Y., Shavitt, Y., Zilberman, N.: An analysis of the steam community
network evolution. In: Electrical & Electronics Engineers in Israel (IEEEI), 2012 IEEE 27th
Convention of. pp. 1–5. IEEE (2012)
6. Bertens, P., Guitart, A., Chen, P.P., Periáñez, Á.: A machine-learning item recommendation
system for video games. arXiv preprint arXiv:1806.04900 (2018)
7. Blackburn, J., Simha, R., Kourtellis, N., Zuo, X., Long, C., Ripeanu, M., Skvoretz, J.,
Iamnitchi, A.: Cheaters in the steam community gaming social network. arXiv preprint
arXiv:1112.4915 (2011)
8. Crawford, A.V., Green, S.B., Levy, R., Lo, W.J., Scott, L., Svetina, D., Thompson, M.S.:
Evaluation of parallel analysis methods for determining the number of factors. Educational
and Psychological Measurement 70(6), 885–901 (2010)
9. Cronbach, L.J.: Coefficient alpha and the internal structure of tests. psychometrika 16(3),
297–334 (1951)
10. Deci, E.L., Eghrari, H., Patrick, B.C., Leone, D.R.: Facilitating internalization: The self-
determination theory perspective. Journal of personality 62(1), 119–142 (1994)
11. Denisova, A., Cairns, P.: First person vs. third person perspective in digital games: Do player
preferences affect immersion? In: Proceedings of the 33rd Annual ACM Conference on Hu-
man Factors in Computing Systems. pp. 145–148. ACM (2015)
12. Deterding, S., Dixon, D., Khaled, R., Nacke, L.: From game design elements to gameful-
ness: defining gamification. In: Proceedings of the 15th international academic MindTrek
conference. pp. 9–15. ACM (2011)
13. Hair, J.F., Black, W.C., Babin, B.J., Anderson, R.E., Tatham, R.L., et al.: Multivariate data
analysis (vol. 6) (2006)
14. Hamari, J., Koivisto, J., Sarsa, H.: Does gamification work?–a literature review of empirical
studies on gamification. In: 2014 47th Hawaii international conference on system sciences
(HICSS). pp. 3025–3034. IEEE (2014)
15. Hamari, J., Tuunanen, J.: Player types: A meta-synthesis. Transactions of the Digital Games
Research Association 1(2), 29 (2014)
16. Han, J., Pei, J., Kamber, M.: Data mining: concepts and techniques. Elsevier (2011)
GamiFIN Conference 2019, Levi, Finland, April 8-10, 2019 226
17. Hendrickson, A.E., White, P.O.: Promax: A quick method for rotation to oblique simple
structure. British journal of statistical psychology 17(1), 65–70 (1964)
18. Horn, J.L.: A rationale and test for the number of factors in factor analysis. Psychometrika
30(2), 179–185 (1965)
19. Hsu, S.H., Kao, C.H., Wu, M.C.: Factors influencing player preferences for heroic roles in
role-playing games. CyberPsychology & Behavior 10(2), 293–295 (2006)
20. Kaiser, H.F.: The varimax criterion for analytic rotation in factor analysis. Psychometrika
23(3), 187–200 (1958)
21. Kallio, K.P., Mäyrä, F., Kaipainen, K.: At least nine ways to play: Approaching gamer men-
talities. Games and Culture 6(4), 327–353 (2011)
22. Lim, C.U., Harrell, D.F.: Developing social identity models of players from game telemetry
data. In: AIIDE (2014)
23. Lin, D., Bezemer, C.P., Hassan, A.E.: Studying the urgent updates of popular games on the
steam platform. Empirical Software Engineering 22(4), 2095–2126 (2017)
24. Lin, D., Bezemer, C.P., Hassan, A.E.: An empirical study of early access games on the steam
platform. Empirical Software Engineering 23(2), 771–799 (2018)
25. Lin, D., Bezemer, C.P., Zou, Y., Hassan, A.E.: An empirical study of game reviews on the
steam platform. Empirical Software Engineering pp. 1–38 (2018)
26. Marczewski, A.: Even ninja monkeys like to play: Gamification, game thinking & motiva-
tional design. Gamified UK (2015)
27. Mora, A., Riera, D., Gonzalez, C., Arnedo-Moreno, J.: A literature review of gamification
design frameworks. In: 2015 7th International Conference on Games and Virtual Worlds for
Serious Applications (VS-Games). pp. 1–8. IEEE (2015)
28. Morschheuser, B., Hassan, L., Werder, K., Hamari, J.: How to design gamification? a
method for engineering gamified software. Information and Software Technology 95, 219–
237 (2018)
29. Nacke, L.E., Bateman, C., Mandryk, R.L.: Brainhex: preliminary results from a neurobio-
logical gamer typology survey. In: International Conference on Entertainment Computing.
pp. 288–293. Springer (2011)
30. O’Neill, M., Vaziripour, E., Wu, J., Zappala, D.: Condensing steam: Distilling the diversity of
gamer behavior. In: Proceedings of the 2016 Internet Measurement Conference. pp. 81–95.
ACM (2016)
31. Orji, R., Nacke, L.E., Di Marco, C.: Towards personality-driven persuasive health games
and gamified systems. In: Proceedings of the 2017 CHI Conference on Human Factors in
Computing Systems. pp. 1015–1027. ACM (2017)
32. Orji, R., Tondello, G.F., Nacke, L.E.: Personalizing persuasive strategies in gameful systems
to gamification user types. In: Proceedings of the 2018 CHI Conference on Human Factors
in Computing Systems. p. 435. ACM (2018)
33. Pontes, H.M., Griffiths, M.D.: Measuring dsm-5 internet gaming disorder: Development and
validation of a short psychometric scale. Computers in Human Behavior 45, 137–143 (2015)
34. Reilly, A., Eaves, R.C.: Factor analysis of the minnesota infant development inventory based
on a hispanic migrant population. Educational and psychological measurement 60(2), 271–
285 (2000)
35. Ryan, R.M., Deci, E.L.: Intrinsic and extrinsic motivations: Classic definitions and new di-
rections. Contemporary educational psychology 25(1), 54–67 (2000)
36. Ryan, R.M., Deci, E.L.: Self-determination theory and the facilitation of intrinsic motivation,
social development, and well-being. American psychologist 55(1), 68 (2000)
37. Sifa, R., Bauckhage, C., Drachen, A.: The playtime principle: Large-scale cross-games in-
terest modeling. In: CIG. pp. 1–8 (2014)
38. Sifa, R., Drachen, A., Bauckhage, C.: Large-scale cross-game player behavior analysis on
steam. Borderlands 2, 46–378 (2015)
GamiFIN Conference 2019, Levi, Finland, April 8-10, 2019 227
39. Stewart, B.: Personality and play styles: A unified model. Gamasutra, September 1 (2011)
40. Tondello, G.F., Mora, A., Nacke, L.E.: Elements of gameful design emerging from user
preferences. In: Proceedings of the Annual Symposium on Computer-Human Interaction in
Play. pp. 129–142. ACM (2017)
41. Tondello, G.F., Orji, R., Nacke, L.E.: Recommender systems for personalized gamification.
In: Adjunct Publication of the 25th Conference on User Modeling, Adaptation and Person-
alization. pp. 425–430. ACM (2017)
42. Tondello, G.F., Wehbe, R.R., Diamond, L., Busch, M., Marczewski, A., Nacke, L.E.: The
gamification user types hexad scale. In: Proceedings of the 2016 annual symposium on
computer-human interaction in play. pp. 229–243. ACM (2016)
43. Tseng, F.C.: Segmenting online gamers by motivation. Expert Systems with Applications
38(6), 7693–7697 (2011)
44. Xu, Y., Poole, E.S., Miller, A.D., Eiriksdottir, E., Kestranek, D., Catrambone, R., Mynatt,
E.D.: This is not a one-horse race: understanding player types in multiplayer pervasive health
games for youth. In: Proceedings of the ACM 2012 conference on computer supported co-
operative work. pp. 843–852. ACM (2012)
45. Yee, N.: Motivations for play in online games. CyberPsychology & behavior 9(6), 772–775
(2006)
46. Zwick, W.R., Velicer, W.F.: Comparison of five rules for determining the number of compo-
nents to retain. Psychological bulletin 99(3), 432 (1986)
GamiFIN Conference 2019, Levi, Finland, April 8-10, 2019 228