Project massive: Self-regulation and problematic use of online gaming

Abstract

A longitudinal design was employed to collect three waves of survey data over a 14 month period from 2790 online gamers. Respondents were asked questions about their gaming activity, motivations, personality, social and emotional environment, and the effect gaming has had on their lives. Prospective analysis was used to establish causal and temporal linkages among the repeatedly measured factors. While the data provide some indication that a player's reasons for playing do influence the development of problematic usage, these effects are overshadowed by the central importance of self-regulation in managing both the timing and amount of play. An individual's level of self-regulatory activity is shown to be very important in allowing them to avoid negative outcomes like problematic use. The role of depression is also discussed. With responsible use, online gaming appears to be a healthy recreational activity that provides millions of people with hours of social entertainment and adaptive diversion. However, failure to manage play behavior can lead to feelings of dependency.

Full text

Project Massive: Self-Regulation and Problematic
Use of Online Gaming
A. Fleming Seay, Ph.D.
School of Information
University of Texas
One University Station, Austin, TX 78712
afseay@ischool.utexas.edu
Robert E. Kraut, Ph.D.
Human Computer Interaction Institute
Carnegie Mellon University
5000 Forbes Ave, Pittsburgh, PA 15213
robert.kraut@cmu.edu
ABSTRACT
A longitudinal design was employed to collect three waves
of survey data over a 14 month period from 2790 online
gamers. Respondents were asked questions about their
gaming activity, motivations, personality, social and
emotional environment, and the effect gaming has had on
their lives. Prospective analysis was used to establish causal
and temporal linkages among the repeatedly measured
factors. While the data provide some indication that a
player’s reasons for playing do influence the development
of problematic usage, these effects are overshadowed by the
central importance of self-regulation in managing both the
timing and amount of play. An individual’s level of self-
regulatory activity is shown to be very important in
allowing them to avoid negative outcomes like problematic
use. The role of depression is also discussed. With
responsible use, online gaming appears to be a healthy
recreational activity that provides millions of people with
hours of social entertainment and adaptive diversion.
However, failure to manage play behavior can lead to
feelings of dependency.
Author Keywords
Online Games, Addiction, Depression, Social Integration,
Self-Regulation, MMORPG, Play Motivation.
ACM Classification Keywords
H.5.0. Information interfaces and presentation (e.g., HCI):
General.
INTRODUCTION
In 2003 approximately 430 million people worldwide, or
7% of the world's population, played video games [35].
Over one quarter of these individuals did so online and that
number as a percentage of total video gamers continues to
grow. In the United States, half of all Americans age six
and older play video games [11]. Worldwide gaming
hardware and software revenues have more than doubled
since 1996, reaching $31.37 billion in 2003. This compares
to $34.2 billion in revenue for the film industry in 2003
[17,15]. The gaming population also continues to diversify.
The average age of the video game player in 2004 was 29,
and 39% of gamers were female [11]. The average 13 - 24
year old in the United States watches 13.6 hours of
television per week compared to 16.7 hours spent using the
internet for activities other than email [15]. The average
adult spends 28 hours a week watching television compared
to average weekly video game play of 7.6 hours [11]. It is
reported that people who play massively multiplayer online
games do so for an average of 15 hours per week, with
weekly play in excess of 30 hours not uncommon [31,33].
One reason for the popularity of online games is that they
meld the fun and challenge of video games with the social
rewards of an online community. Participation in online
communities allows us to stay in touch with old friends,
meet new people, learn, and share information [29]. It also
enables self-exploration and discovery as users extend and
idealize their existing personalities or try out new ways of
relating to one another that can positively affect real life
relationships [32,4]. On the other hand, some fear that
virtual communities detract from social activity and
involvement in the real world, replacing real social
relationships with less robust online substitutes and causing
users to turn away from more traditional media [20,25].
A Question of “Addiction”
It is logical that such a large industry with widening appeal
and an expanding rate of use would have some non-uniform
effects on its participants. Reports in the popular media
continue to suggest that the design and content of certain
games are responsible for the detachment, depression, and
even addiction that some players experience. Some have
estimated that 10% of online game players are addicted to
the activity, an extrapolation from the ABCNEWS.com
survey finding that 10% of all users of the internet are
addicted to it [34,13,16]. An internet search for “gaming
addiction” yields lists of physical and psychological
symptoms from dry-eyes and carpal tunnel syndrome to
“problems with school or work,” offered as indicative of
problem usage behavior [27]. Some clinicians claim that
online game players “don’t have normal social relationships
anymore” and play online games in order to cover feelings
Permission to make digital or hard copies of all or part of this work fo
r
p
ersonal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
b
ear this notice and the full citation on the first page. To copy otherwise,
or republish, to post on servers or to redistribute to lists, requires prio
r
specific permission and/or a fee.
CHI 2007, April 28–May 3, 2007, San Jose, California, USA.
Copyright 2007 ACM 978-1-59593-593-9/07/0004...$5.00
CHI 2007 Proceedings • Games April 28-May 3, 2007 • San Jose, CA, USA
829

of anger, depression and low self-esteem [26]. Under
increasing public and governmental scrutiny, a major
gaming industry group in Korea has laid out a multi-part
initiative aimed at combating overuse of online games
through education, monitoring software and the
establishment of treatment and rehabilitation centers.
Anecdotal evidence also continues to mount. Support
groups and online communities like EverQuest Widows and
Spouses Against EverQuest are available on the web, full of
stories about damaged and destroyed relationships.
Communications of the ACM published an editorial on the
deleterious impact online gaming has on undergraduates,
particularly computer science majors, in the United States
[23]. In addition, there do exist truly tragic stories, like that
of the clinically depressed young man, described by many
who knew him as addicted to EverQuest, who killed
himself following an extended session of play [24]. Clearly
there is something here worthy of study; the first challenge
is to determine how to approach it.
A Shift in Terminology
In the popular media, addiction to online games has been
likened to pathological gambling, eating disorders and drug
dependency [26]. In addition, and in spite of the
protestations of leading thinkers in interactive
entertainment, both marketing departments and the critical
media within the games industry also talk about their
games’ “addictive” qualities with pride [1]. More often than
not, statements made about “addictive gameplay” refer to a
desirable quality of the experience marked by incremental
reinforcement, perseverance in adversity, and desire to
continue; to play “just one more”. For most, the experience
of an “addictive” game is much the same as that of a “page
turner” novel; you don’t want to put it down, and it is hard
to keep track of time while engaged with it. Clearly, this
type of an immersive and rewarding entertainment
experience is precisely what the consumer wants and what
the developer wants to create. Addiction of this kind could
easily be recast as engagement, the state of being
delightfully attracted to and enwrapped in an experience
[7].
In contrast, addiction can also be used to describe the state
of powerlessness a person experiences when, despite
attempts to stop or reduce their usage, they are unable to
walk away from a game (or substance, or behavior) even in
the face of persistent and deleterious effects on their life.
Given the various pejorative, disputed, and clinically laden
connotations of the word “addiction”, we have chosen to
refer to self-described pre-occupation with and inability to
withdraw from gaming as problematic use of online games.
We do this not to refer to addiction euphemistically, but to
dissociate the phenomenon under study from the state of
biochemical dependency most closely associated with the
word “addiction.” For the purposes of the present study,
problematic use can be operationalized as consumption of
an entertainment product in such amounts or at such times
that it causes demonstrable problems in the user’s real life
extreme enough to cause an individual to identify and
report them [7]. Under this definition, online gaming would
become problematic when it dominates and displaces other
behaviors, leads to conflict, or when inability to play causes
anxiety. Some players, even those spending upwards of 40
hours a week gaming, might have euphoric gaming
experiences, play for long periods, and think about gaming
even when not doing it, but suffer no ill effects as a result.
We hypothesize that such players might actively manage
their use of entertainment products, ensuring that gaming
remains a positive aspect of their lives. Unfortunately, other
players may not be as successful at managing their
consumption, and allow persistent involvement in online
games to interfere with their everyday life. An individual’s
management of his or her own behavior through techniques
like self-monitoring, evaluation against perceived
standards, and self-administration of behavioral
consequences is referred to as self-regulation.
Rather than presenting a monolithic view that online games
are either bad or good, we predict that different levels of
self-regulatory activity and motivations for play are likely
to yield different consequences for the user. The
individual’s motivation to play and ability to manage their
own behavior promise to be important factors in
determining the outcome of use of online games. This
document examines the relationship of self-regulation and
problematic use, reporting a subset of the findings of a
larger inquiry into the social and psychological impact of
online gaming.
PROBLEMATIC USE
Problematic use of online gaming can be viewed as a
special case of the broader concept of Pathological Internet
Use (PIU). The American Psychological Association
formally recognized Internet Addiction in the late 1990s
and gave it this more clinically precise title. PIU has
become the focus of much interest in recent years. The most
popular definitions and metrics of PIU are adapted directly
from clinical definitions of substance abuse/dependency,
pathological gambling, and impulse control disorders found
in various editions of the Diagnostic and Statistical Manual
of Mental Disorders [12, 5, 34, 13, 14].
Brown referred to problem gambling as a type of behavioral
addiction and developed six general criteria to diagnose
them: Tolerance, the need to engage in the problem
behavior for longer periods of time in order to attain the
desired effect; Euphoria, the high brought on by engaging
in the behavior; Salience, the ongoing dominance of the
behavior in thought and action (sometimes divided into
Behavioral and Cognitive Salience); Conflict, the behavior
causing both psychological and environmental discord;
Withdrawal, negative affect associated with periods of
inability to engage in the behavior; and Relapse, resumption
of the behavior despite efforts to stop [5]. As LaRose
observes, deficient self-regulation is both implicit in the
CHI 2007 Proceedings • Games April 28-May 3, 2007 • San Jose, CA, USA
830

definition of addiction and explicit in the criteria commonly
used to assess it [21].
Charlton discovered that Brown’s six criteria did not
universally load on a Computer Addiction factor [7]. In
fact, only Behavioral Salience, Conflict, Relapse, and
Withdrawal loaded on addiction. The others, Tolerance,
Euphoria, and Cognitive Salience, loaded on a Computer
Engagement factor. This finding suggested that scales based
on Brown’s six factors did not measure a unitary
phenomenon. Instead some of these criteria, commonly
viewed as symptomatic only of clinical dependence, were
more strongly associated with a non-pathological construct,
that of engagement. Engagement can be defined as a state
of deep interest in and involvement with a medium.
Tolerance, euphoria, and cognitive salience are not
inherently pathological in the way that the other four
criteria, behavioral salience, conflict, withdrawal, and
relapse are.
It is not difficult to argue that few would find any level of
the remaining three criteria desirable. Unlike the others,
conflict, withdrawal, and relapse are by their very
definition, undesirable and pathological entities. Physical
and emotional struggle, separation anxiety, and repeated
inability to disengage from a behavior offer few adaptive
interpretations. It is the stark contrast of these three factors
to tolerance and euphoria that drives the bifurcation of
Brown’s criteria, a distinction mathematically demonstrated
by Charlton and shown in Figure 1, below.
Danforth adapted subscales from Charlton’s
Engagement/Addiction Scale, or EAS, to create the EAS-II,
an instrument designed to measure addiction to and
engagement with massively multiplayer games [7,8,10].
The EAS-II is a 28 item instrument comprised of 15 items
from Charlton’s Engagement subscale (e.g. “I feel a sense
of power when I play EverQuest 2”) and 13 from the
Addiction subscale (e.g. “When I am not playing EverQuest
2, I feel agitated”). Deploying the EAS-II with 442 players
of Microsoft/Turbine’s MMOG Asheron’s Call, Danforth’s
results supported the addiction/engagement dichotomy
pointed out by Charlton [10].
Engaged Use Problematic Use
Tolerance Behavioral Salience
Euphoria Conflict
Cognitive Salience Withdrawal
Relapse
Figure 1. The bifurcation of Brown’s diagnostic criteria for
behavioral addiction.
This research employs the full EAS-II instrument, changing
only the way in which the “addiction” factor is referenced:
as problematic use of online games.
SELF-REGULATION
Bandura’s Social Cognitive theory of personality portrays
the human individual as a proactive, self-organizing, and
self-reflecting agent rather than a reactive organism that is
shaped solely by external events and circumstances [2].
Central to this agentic, sociocognitive perspective is the
concept of self-regulation, the ability of an individual to
manage his own behavior through observation, evaluation,
and consequation. Arguments about the design of
potentially harmful forms of entertainment focus heavily on
the content of these objects, but largely ignore the processes
taking place within the consuming individual. Hence, it is
important that any study addressing problematic use of
online gaming examine the role of an individual’s self-
regulatory abilities in managing gaming behavior. These
self control behaviors are often divided into three
interactive classes: self-monitoring, self-evaluation, and
self-consequation [18,2,19,3]. The literature in both
psychology and communications points to the importance
and effectiveness of self-regulation in the identification,
assessment and treatment of both behavioral excesses and
deficits [19,3,21].
In order to illustrate this self-regulatory framework, let’s
examine how self-regulation fits into a personal videogame
play management paradigm. Self-monitoring, or simple
introspective observation of the amount of time one has
spent playing, would presumably have an effect on
subsequent play in that the individual would recognize that
they have been involved in a particular activity for several
hours and may want to consider other concerns. The
inability to recognize how much time one has spent
involved in an activity would be an example of a failure in
self-monitoring, e.g. losing track of time. Self–evaluation of
play would involve an individual comparing their observed
time allotment for gaming to those made to other activities
or by other individuals. For example, a player might notice
that she has been online twice as often as her in-game
friends, suggesting that she may play twice as much as
these other people. Alternatively, she might consider that
she plays during the day at work, but none of her co-
workers or guild mates seem to be online until the evening
hours. This kind of self-evaluation through the comparison
of one’s activities to external standards builds on the self-
monitoring process by utilizing information gained from
self-monitoring. Self-consequation involves the
development of behavioral contingencies that, based on the
outcome of the self-evaluative process, lead to the self-
administration of reinforcement or punishment. For
example, one might deny one’s self a trip to the movies
given a large amount of time spent playing, or treat play as
a reward for the completion of formerly neglected
responsibilities.
In the current study, Carey et. al’s Short Self-Regulation
Questionnaire (SSRQ) was used to empirically measure
self-regulatory behavior [6]. Items included in the SSRQ
address all three of the dimensions of self-regulatory
CHI 2007 Proceedings • Games April 28-May 3, 2007 • San Jose, CA, USA
831

Table 1. Example items from the SSRQ.
behavior discussed above, but measure the construct of self-
regulation in a mathematically unitary fashion. The SSRQ
contains 31 items that the respondent rates on a 5-point
Likert-type scale from Strongly Agree to Strongly Disagree.
Example items are presented in Table 1.
Note that these items address usage of the general processes
of self-regulation and are not specific to online gaming. It is
important to keep in mind that the same self-regulatory
functions can and do operate at various levels of a person’s
behavioral hierarchy at the same time; in several domains at
once from pursuit of life goals to more granular behaviors.
Finally, a note on what is being regulated. When one
chooses to play is as important as how much, as even a very
short session undertaken at an inappropriate time (perhaps
to delay a responsibility or escape / avoid a stressful
situation) can be problematic. Clearly, avoidance behaviors
can be adaptive, but when undertaken at inappropriate times
or with excessive frequency they can be harmful. As such,
regulation of the impulse to play is just as important as
regulation of the amount of play.
HYPOTHESES
Given the preceding discussion of problematic use and self-
regulation, we are ready to explore some hypotheses
regarding their interrelation.
Hypothesis I – Self-Regulatory deficits will predict the
development of problematic use.
This hypothesis makes the simple claim that deficits in self-
regulation contribute to the development of addiction to
online games. By measuring this relationship
longitudinally, we will examine the temporal relationship of
these factors.
Hypothesis II - Certain play motivation factors will
distinguish players who are more susceptible to problematic
use.
The literature on player motivation indicates that certain
player motivations, particularly Escapism are cross-
sectionally associated with increased levels of problematic
use [31,33]. An adaptation of Yee’s Facets scale including
the Achievement, Escapism, Roleplaying, Manipulation,
and Relationship types was employed to determine whether
play motivation is predictive of problematic use in the same
way that social and personality factors predict susceptibility
to depression.
Hypothesis III- The effect of self-regulatory deficits on
problematic use will interact with depression.
This hypothesis is designed to evaluate the role of
depression in the relationship between self-regulation and
problematic use. Specifically, this hypothesis addresses the
effect of depression on the self-regulatory processes. It
hinges on a significant depression by self-regulation
interaction in the presence of a main effect of self-
regulation on problematic use. This result would support
the claim that the effect of self-regulatory deficits on
problematic use may be moderated by depression. In this
model, depression is not a cause or necessary precursor of
problematic use, but its presence may heighten the effects
of deficient self-regulation on the development of
problematic use.
It has been suggested that depression could moderate the
effect of self-regulatory mechanisms on an individual’s
behavior [19]. In general, depressive affect is related to
self-imposed low expectations and unreasonably high
standards for success (e.g. self-doubt about ability to
succeed paired with inability to set reasonable and
attainable goals). In addition, depressed individuals operate
under the illusory belief that other people share these
lowered expectations and unrealistic standards for them.
Under such a paradigm, self-evaluation and self-
consequation can easily break down. Adaptive self-
evaluation is predicated on the identification of useful
standards of comparison. Further, if one makes some form
of reinforcement contingent on meeting an unrealistic
standard, the individual will soon identify the goal as
unattainable and, where possible, circumvent the
contingency, thereby giving up any therapeutic effects it
may have had if performed as designed. Even under
conditions of success, where the individual negotiates the
behavioral contingency as designed, depressed individuals
are less likely to view the outcomes as sufficiently
reinforcing to merit repetition. Depressed individuals tend
to be low in expectancy to achieve goals, and apt to
evaluate themselves negatively. Simply, depression lessens
one’s belief in their ability to manage their own behavior
and blunts the capacity to identify success and enjoy its
rewards.
With respect to online game play, the effectiveness of self-
regulatory activities on amount and timing of play may be
reduced for depressed individuals. This suggests a
moderation model in which depressive affect interacts with
self-regulatory deficits to exacerbate problematic play
behavior. Without question, deficient self-regulatory
behavior can logically lead to problematic use. However, it
is the non-additive effect of higher depression and low self-
Short Self Regulation Questionnaire Sample Items
I usually keep track of my progress toward my goals.
It’s hard for me to notice when I’ve “had enough”
(alcohol, food, sweets).
I have personal standards, and try to live up to them.
When I’m trying to change something, I pay a lot of
attention to how I’m doing.
I usually only have to make a mistake one time in order
to learn from it.
CHI 2007 Proceedings • Games April 28-May 3, 2007 • San Jose, CA, USA
832

regulation that is of principal interest here. Depression was
measured by the Center for Epidemiologic Studies
Depression Scale (CES-D).
METHOD
During the spring of 2002, contextual inquiries with 15
experienced online gamers were conducted to support the
creation of a 69-item web survey. Following a period of
analysis and refinement, a new web survey was created and
deployed repeatedly between September of 2004 and
November of 2005. The three waves of longitudinal data
collection lasted 3-4 months each, and were separated by 3-
4 month periods. Participants were solicited through posts
in popular gaming forums and websites. Most of the
respondents played Massively Multiplayer games on the PC
platform, but the sample included a large number of players
from other genres and platforms (e.g. gaming consoles like
Microsoft’s XBox and Sony’s Playstation 2).
Figure 2. Predictors used in modeling problematic use
Prospective analysis was used to evaluate the longitudinal
effects of the factors contained in the predictor blocks
above on the dependent variable of interest, problematic
use. In prospective analysis, a regression equation is built in
which a lagged predictor variables are used to model a
future value of the dependent variable of interest [9].
Initially a lagged value of the dependent variable itself is
entered into the regression equation alone. For example, a
regression equation modeling problematic use for a given
time period is created using only the participant’s
problematic use score from the previous time period, or
wave of data collection. Next, each predictor block is added
to the regression equation incrementally. If a predictor
block adds nothing to the model, that block is removed. It is
important to note that the dependent variable is modeled
using values of the predictor block variables collected
during the previous wave. This technique allows inspection
of the unique variance in the dependent variable accounted
for by the lagged variables, over and above that accounted
for by the previously measured value of the dependent
variable. Prospective analysis exposes those predictors that
add explanatory power to the model in excess of that
generated by the lagged dependent variable. The values of
all predictors are centered so that the size of their effects
can be compared to one another.
RESULTS
A total of 4490 unique respondents participated in the
Project Massive survey from pilot to final wave.
Information on the number of participants by wave and
when each wave was conducted is presented in Table 2.
Wave Participants Returning Start End
Pilot 1836 n/a Mar-02 Dec-02
1 1503 n/a Sep-04 Dec-04
2 1089 397 Apr-05 Jul-05
3 790 331 Oct-05 Nov-05
Table 2. Number of participants by wave with collection dates
The cross-sectional results presented in this section are
drawn from an aggregate pool of 2790 records containing
data from the first time a given participant responded to the
survey regardless of wave.
Participants ranged in age from 11 to 70 with an average
age of 28 (M=27.98). Males comprised 88% of the sample,
with 327 female respondents making up the other 12%.
74.8% of respondents had jobs or were self-employed. 49%
of the respondents were single, 41% were married, and 21%
of the respondents had children.
The mean number of hours spent playing online games per
week was 21.7. As is shown in Figure 3, the distribution is
skewed to the right, with a sizable minority of players
(~15%) indicating they play more than 54 hours per week.
Figure 3. The average hours played per week
On average respondents spent 36% of their weekly online
time playing by themselves, 33% of it playing with
members of their player organization, 15% playing with
online friends not in their guild, 18% playing with
CHI 2007 Proceedings • Games April 28-May 3, 2007 • San Jose, CA, USA
833

N~2600 Engaged Use Problematic Use
Play Hours .263 .318
Depression .085 .380
Loneliness .076 .284
Self-Regulation -.014 -.345
Game Affinity .404 .086
Table 3. Cross-sectional correlation of certain measures with
Engaged and Problematic Use
strangers, and 6% just “hanging out” logged in with no
intention to play.
Table 3 offers discriminative statistical evidence in favor of
the conceptual distinction between problematic and
engaged use. Though both outcomes are associated with
hours of play, we see a rather stark contrast in the usage
outcomes’ associations with the other measures. Only
problematic use is associated with depression, loneliness,
and self-regulation. Engaged use shows no relationship with
these measures, but does show positive correlation with
game affinity, a simple measure of how much the player
likes the game they play.
Table 4 shows the zero order correlations of problematic
use with the player motivation factors. The correlation with
self-regulation and hours of play per week are also shown.
Problematic Use
Escapism .370 (2669)
Achievement .297 (2657)
Manipulation .228 (2656)
Relationship .102 (2663)
Roleplaying -.006 (2648)
Hrs of play / week .318 (2677)
Self-Regulation -.345 (2489)
Table 4. Correlates of Problematic Use
A Longitudinal Model of Problematic Use
The number of respondents participating in at least two
waves of the survey was 499. This is the pool of
participants from which we are able to make longitudinal
models, since we have at least two waves of data from each
of them.
Table 5 shows a longitudinal regression model of change in
problematic use over time. This model accounts for roughly
53% of the variance in problematic use (adjusted R2 (372)
=0.525). The relatively strong negative main effect
associated of self-regulation demonstrates a longitudinally
negative relationship between self–regulation and
problematic use. That is, individuals reporting high levels
of self regulatory activity at a given time period report
lower levels of problematic usage of video games at the
next time period.
Estimate S.E. p value
Intercept 2.940 0.045 0.000
Problematic Use (Lagged) 0.677 0.049 0.000
Block 1 - Controls
Female -0.011 0.109 0.918
Age -0.009 0.040 0.822
Intelligence -0.064 0.039 0.108
Attractiveness 0.071 0.046 0.122
Block 2 – Types of Play
Achievement 0.039 0.040 0.327
Escapism 0.075 0.044 0.086
Roleplaying 0.046 0.040 0.255
Manipulation -0.092 0.045 0.044
Relationship -0.074 0.042 0.080
Hours 0.071 0.050 0.157
Affinity 0.072 0.040 0.072
Block 3 – Types of Use
Engaged Use -0.061 0.049 0.210
Self-Regulation -0.132 0.053 0.013
Block 4 – Social Dimensions
Play w/ RL Friends 0.011 0.039 0.787
Block 5 - Depression
Depression 0.077 0.051 0.138
Interactions
Self-Reg * Depression 0.113 0.032 0.000
Hours * Game Affinity -0.080 0.038 0.038
Table 5. Regression model predicting Problematic Use
Conversely, it indicates that those individuals reporting
lower levels of self-regulatory activity are likely to report
increases in problematic use the next time data is collected.
This result fully supports the prediction made in Hypothesis
I, clearly indicating that individuals who actively monitor
and manage their behavior in general are less likely to allow
their involvement in online gaming to cause them real life
problems.
While there still appear to be near significant trends for a
collection of player motivations (Escapism, Manipulation,
and Roleplaying), these effects are altered in both size and
significance by the introduction of Self-Regulation and the
interactions. Overall, Escapism is associated with increases
in problematic use while Relationship play and
Manipulation are associated with decreases. These effects
are changed slightly with the introduction of Self-
Regulation into the model.
CHI 2007 Proceedings • Games April 28-May 3, 2007 • San Jose, CA, USA
834

Figure 4. A plot of the Self-Regulation by Depression
Interaction on Problematic Use
The effect of Manipulation remains significant, but the
positive effect of Escapism and the negative effect of
Relationship play are dampened, becoming only marginally
significant. The negative effect of Manipulation play on
problematic use indicates that players who are motivated to
play by their enjoyment of harassing and annoying others
are likely to report lower levels of problematic use at a
second time period than those players less inclined to
behave in such a manner.
The interaction of self-regulation and depression indicates
that depression moderates the effect of self-regulatory
behavior on problematic use. At lower levels of depression,
self-regulation has the negative effect on problematic use
illustrated by its main effect. However, as depression
increases above mean levels, the effect of self-regulation on
problematic use is eliminated. Simply, depressive affect
reduces the effectiveness of the self-regulatory processes,
rendering the individual less able to manage their own
behavior and more likely to experience problematic use.
This result offers specific support for Hypothesis III, which
posits the exact moderating relationship obtained here. This
interaction is plotted in Figure 4.
The Hours by Affinity interaction is plotted in Figure 5. It
illustrates that how much a player likes a game will
moderate the effect that hours of weekly play has on
problematic use. At low levels of affinity, hours of play has
a strong positive effect on problematic use. However, at
higher levels of game affinity, increases in hours of play
have no effect on problematic use levels. This suggests that
players who enjoy and have high regard for the game that
they play can play it for many hours each week without
feeling that the activity is causing them any problems.
However, individuals who continue to play a game that they
view negatively or do not like for many hours each week
report higher levels of problematic use.
Figure 5. A plot of the Hours of Play by Game Affinity
Interaction on Problematic Use.
It is important to note that even though cross-sectional
analyses show an association between hours of play and
problematic use, hours of play (or amount of consumption)
in and of itself is not predictive of problematic use. This
result further discounts simple “media effects” models in
which amount of exposure is determinant of the outcome of
use. Along with the zero order correlation described earlier,
this indicates that while hours of play may have a positive
cross-sectional relationship with reports of problematic use,
it does not have longitudinally predictive power. Simply, a
large amount of play is certainly associated with
problematic use cross-sectionally, but is not predictive of
future problematic usage issues, particularly in situations
where the player enjoys the game that they are playing (e.g.
high game affinity).
As discussed earlier, Danforth used a seven-factor
personality inventory, including the Big Five plus
Attractiveness and Negative Valence, which showed little
predictive value with respect to the development of
“addiction” [10]. This result has been replicated in the
current study. The final model presented in Table 5 reflects
the removal of five of the seven personality factor from the
model due to their lack of contribution to the model fit.
DISCUSSION
Hypothesis I
Hypothesis I stated that self-regulatory deficits would
predict the development of problematic use. The obtained
significant negative effect of self-regulation on changes in
problematic use supports the prediction made in Hypothesis
I. Those individuals who report low levels of self-
regulatory activity tend to go on to report significantly
higher levels of future problematic use. On the other hand,
those individuals who actively regulate the timing and
amount of their play behavior through self-monitoring, self-
evaluation and self-consequation report significantly lower
levels of future problematic use than their counterparts.
Further, the effect of self-regulation on problematic use is
CHI 2007 Proceedings • Games April 28-May 3, 2007 • San Jose, CA, USA
835

the largest and most robust of all predictive factors
measured. Clearly, the self-regulatory processes are
essential in allowing online gaming to remain a benign and
enjoyable pass-time rather than an obstructive pre-
occupation. Active self-regulation appears to be a player’s
best defense.
Hypothesis II
Hypothesis II predicted that certain play motivation factors
would distinguish players who are more susceptible to
problematic use. The significant negative effect of
Manipulation offers only minimal support for Hypothesis
II, as it was expected that Escapism and Achievement
would have a significant positive effect.
The significant negative effect of Manipulation on
problematic use is an interesting one. In general, we might
consider this type of “grief play” to arise from a situation in
which the core game mechanic has failed to engage the
player and forced him to pursue other avenues of enjoyment
within the game’s confines. The fact that grief play has a
cross-sectionally positive relationship with problematic use
suggests that in the near term it offers enough reward to
compel some players to over-indulge in its pursuit.
However, the longitudinally negative relationship of grief
play with problematic use supports this notion that, with
time, grief players tend to report lower levels of
problematic use, perhaps due to the diminished enjoyment
associated with repeated manipulative exploits within the
same game or upon the same set of victims and their desire
to find a game they find more enjoyable in general. Such
diminishing returns would not reward a player for repeated
and extended pursuit of manipulation play compared to
escapism and achievement play which are more regularly, if
not almost continuously, reinforced. Given the age and
gender profile of players scoring high on the Manipulation
dimension (young males), one can hope that parental
supervision plays some role in constraining their usage (if
not their in-game behavior) such that it does not become
problematic.
The Escapism motivation does have a near significant
positive effect on problematic use. This would suggest that
players who are high in the Escapism motivation tend to
report increases in problematic use. What this means is that
the use of online gaming as a curative respite from real
world stressors, while adaptive in moderation, can have
deleterious effects on those who use it in this manner. The
fact that there is no interaction with weekly hours of play
means that responsibility for increased levels of
problematic use lies with the Escapism motivation and not
the hours spent pursuing it. That is, one need not spend
many hours “escaping” but rather resort to the escape
behavior at inappropriate times or in unsuitable situations in
order to feel that the behavior has begun to have negative
effects on their real life. This is not to suggest that
Escapism is a necessarily insidious and maladaptive way to
go about using online games. On the contrary, it is hard to
argue against the relaxing and restorative effects of pursuit
of any recreational activity used to release or relieve
feelings of stress and anxiety. Again, this effect is not quite
significant in the final model, since the variance for which
it accounts is more robustly explained by the self-regulation
factor and the significant interactions.
Another nearly significant effect is that of the Relationship
play motivation. This result suggests that players who use
online gaming as a medium in which to meet people and
interact with them in meaningful social ways report lower
levels of problematic use than those less socially motivated.
It seems that those who view playing as an adaptive social
activity that rounds out their existing social life are less
likely to later report that they feel the activity has been
causing real life difficulty for them.
In sum, the fact that the player motivations were less
predictive of problematic use than expected is not so much
indicative of their descriptive weakness as it is a testament
to the centrality of self-regulation. Without self-regulation
and its interaction with depression in the model,
Manipulation, Escapism, and Relationship all significantly
contribute to the model.
Hypothesis III
Hypothesis III specifically predicted that the effect of self-
regulatory deficits on problematic use would interact with
depression. Explicit support for this hypothesis is provided
by the significant Self-Regulation by Depression interaction
obtained in the problematic use model. This interaction
indicates that depressive affect moderates the effect of the
self-regulatory processes on the development of
problematic use. At lower levels of depression, the self-
regulatory processes work, as indicated by their main effect,
to lower problematic use levels. With increased levels of
depression the negative effect of the self-regulatory
processes are blunted and they become less effective in
preventing problematic use. In this model, depression does
not cause problematic use, but its presence may catalyze
and accelerate the effects of deficient self-regulation on the
development of problematic use.
Recall our earlier discussion of depressive affect and its
relation to self-imposed lower expectations and
unreasonably high standards for success. This can be
characterized as a brutal pairing of self-doubt about one’s
ability to succeed and a tendency to set unreasonable and
unattainable goals. Lack of self-belief paired with inability
to set and evaluate progress toward reasonable goals
undermines the basis of the self-regulatory processes.
Specifically, an overarching tendency to view one’s self
negatively hampers the self-evaluative process. Further, if
one is unable to identify and place value upon self-
evaluative successes, then self-consequation becomes
impossible. Even when one does register a self-evaluative
success, depressed individuals are less likely to view the
rewards of self-consequation as sufficiently reinforcing to
merit repetition [19]. Depression lessens one’s belief in
CHI 2007 Proceedings • Games April 28-May 3, 2007 • San Jose, CA, USA
836

their ability to manage their own behavior and blunts the
capacity to identify success and enjoy its rewards. The
results obtained in support of Hypothesis III provide
empirical evidence in favor of the notion that depression
undermines the self-regulatory processes through this
mechanism and in doing so makes even those individuals
who do self-regulate vulnerable to problematic usage of
online games.
Limitations
There are several important limitations to this research that
deserve discussion. Foremost among them are the standard
caveats associated with survey research of this type. All of
the data collected for this study were self-reports. As such,
issues of social desirability and accuracy of response need
to be taken into account. Further, though we have a good
picture of who we did get as respondents, we do not know
anything about those people that were not involved or chose
not to participate in Project Massive. Every effort was made
to ensure that users of a wide number of platforms and
games were included in the study, however most came from
the massively multiplayer genre.
Only one psychological outcome of gaming was addressed in
this study, depression. Though there may be several other
possible negative outcomes (e.g. aggression), one should also
consider the various positive outcomes of play. Happiness,
self-esteem and assertiveness all would make valid additions
to a more general inquiry addressing the psychological
impact of online gaming.
CONCLUSION
Now that we have reviewed the individual hypotheses, let us
turn to what we have learned more broadly. It seems safe to
say that the data provide no indication that online gaming is a
broadly negative activity. On the contrary, the overwhelming
majority of those surveyed indicate no elevation in
loneliness, depression, or problematic use. This seems to
indicate that, for most, online gaming is an adaptive and
enjoyable, or at least benign, activity.
This study has identified and demonstrated the central
importance of self-regulation in changing problematic
gaming behavior or preventing it from developing altogether.
The results clearly indicate that self-regulation is important in
shielding the user from problematic use and reducing or
eliminating problematic use once it arises.
We are now be able to speak with some confidence about
what does not cause problematic use and what helps prevent
it, leveraging this information to explain why some players
describe themselves as “addicted” while others remain
adaptively engaged. The results of Project Massive indicate
that self-regulatory activity is essential in addressing
problematic usage. These self-regulatory findings can inform
the design of informal personal strategies and formal
software systems aimed at helping players and developers
alike manage play behavior and protect against problematic
use. Further, these findings and their implications are
applicable to the more general case of internet dependency.
If they are not already an important part of our present,
online communities like those that exist in and around online
games will become an immense force in our future. They will
come to affect many aspects of our lives; how we
communicate, how we learn, how we relax, what we buy,
and even whom we trust. Understanding the effects that
participation in these digital communities has on the day-to-
day lives and well being of those who participate in them is
imperative as we strive to ensure that humanity is
empowered and not ensnared by the technologies that we
create. Project Massive is a small but important step in that
direction.
ACKNOWLEDGMENTS
First and foremost, the authors would like to express their
sincere gratitude to the 4490 survey participants without
whom Project Massive would never have been possible. The
first author is forever in the debt of the following individuals
for their contributions to this work: John Bailey, Katie
Bessiere, K.C. Blackburn, Matt Blythe, Ed Castronova,
Richard Catrambone, Darren Gergle, Larry Hodges, Scott
Hudson, William Jerome, Sara Kiesler, Cliff Lampe, Kevin
Sang Lee, Jeff Nichols, Amy Ogan, Sarah Pressman, Paul
Resnick, Michael Scheier, Jesse Schell, Irina Shklovski,
Dmitri Williams, Jake Wobbrock, and Nick Yee.
REFERENCES
1. Adams, E. (2002) “Stop Calling Games Addictive.”
http://www.designersnotebook.com
2. Bandura, A. (1999). “A sociocognitive analysis of
substance abuse: An agentic perspective.” Psychological
Science 10(3): 214-217.
3. Bandura, A., Caprara, G.V., Barbaranelli, C., Verbino,
M., Pastorelli, C. (2003). “Role of affective self-
regulatory efficacy in diverse spheres of psychosocial
functioning.” Child Development 74(3): 769-782.
4. Bessière, K., Kiesler, S., Kraut, R., & Boneva, B. (Under
Review). Longitudinal Effects of Internet Uses on
Depressive Affect: A Social Resources Approach.
Unpublished manuscript, Carnegie Mellon University,
Pittsburgh, PA.
5. Brown, R. I. F. (1993). Some contributions of the study
of gambling to the study of other addictions. Gambling
behavior and problem gambling. W. R. C. J. A.
Eadigton. Reno, NV, University of Nevada: 241-272.
6. Carey, K. B., D. J. Neal, et al. (2004). “A psychometric
analysis of the self-regulation questionnaire.” Addictive
Behaviors 29(2): 253-260.
7. Charlton, J. P. (2002). “A factor-analytic investigation of
computer 'addiction' and engagement.” British Journal of
Psychology 93: 329-344.
8. Charlton, J. P., & Danforth, I.D.W. (2004).
Differentiating computer-related addictions and high
CHI 2007 Proceedings • Games April 28-May 3, 2007 • San Jose, CA, USA
837

engagement. Psychology & IT 2004, University of
Cadiz.
9. Cohen, J. & Cohen, P. (1983). Applied Multiple
Regression/Correlation Analysis for the Behavioral
Sciences, Second Edition. Lawrence Erlbaum
Associates, New Jersey, USA.
10. Danforth, I. D. W. (2004). “Addiction to Online
Games.”
11. ESA (2004). “Top Ten Industry Facts.”
http://www.theesa.com/pressroom.html.
12. Goodman, A. (1990). “Addiction: Definition and
implications.” British Journal of Psychiatry 85: 1403-
1408.
13. Greenfield, D. N. (1999). Virtual Addiction. Oakland,
CA, New Harbinger Publications, Inc.
14. Griffiths, M. D., Davies, N.O., Chappell, D. (2004).
“Online Computer Gaming: A Comparison of
Adolescent and Adult Gamers.” Journal of Adolescence
27: 87-96.
15. Harris Interactive (2003). “Born to Be Wired:
Understanding the First Wired Generation.”
http://us.i1.yimg.com/us.yimg.com/i/promo/btbw_2003/
btbw_execsum.pdf.
16. IGDA, S., D., Staehlin, C., McShaffry, M., Meunier, V.
(2003). “IGDA Austin Chapter Panel on Game
Addiction.”
www.igda.org/articles/austin_addiction.php.
17. Kagan (2004). “The Business of Movie Production and
Distribution 2003.” http://www.kagan.com/cgi-
bin/pkcat/prd03.htm.
18. Kanfer, S. H. (1970). Self-regulation: Research, issues,
and speculations. Behavior modification in clinical
psychology. C. M. Neuringer, J.L. New York, Apple-
ton-Century-Crofts: 178-220.
19. Kocovski, N. L. and N. S. Endler (2000). “Self-
regulation: Social anxiety and depression.” Journal of
Applied Biobehavioral Research 5(1): 80-91.
20. Kraut, R., Lundmark, V., Patterson, M., Kiesler, S.,
Mukopadhyay, T., & Scherlis, W. (1998). “Internet
paradox: A social technology tht reduces social involve-
ment and psychological well-being?” American
Psychologist 53: 1017-1031.
21. LaRose, R., Lin, C.A., Eastin, M.S. (2003).
“Unregulated Internet Usage: Addic-tion, Habit, or
Deficient Self-Regulation?” Media Psychology 5(3):
225-253.
22. Medsearch (2004). “Medsearch Medical Reference.”
http://www.gulflink.osd.mil/medsearch/glossary/glossar
y_d.shtml.
23. Messerly, J. G. (2004). “How Computer Games Affect
CS(and other)Students' School Performance.” 47: 29-31.
24. Miller, S. A. (2002). “Death of a game addict.”
http://www.jsonline.com/news/state/mar02/31536.asp.
25. Nie, N.H., & Hillygus, S. (2001). Education and
democratic citizenship. In Making good citizens:
Education and civil society, Diane Ravitch and Joseph
Viteritti (eds.). New Haven: Yale Univ. Press.
26. O' Dwyer, S. (2002). “Online Gaming Addiction on the
Rise.” http://www.media52.net/archives/000032.html.
27. Orzack, M. H. (2004). “Computer Addiction Services.”
http://www.computeraddiction.com/.
28. Radloff, L. S. (1991). “The use of the center for
epidemiologic studies depression scale in adolescents
and young adults.” Journal of Youth and Adolescence
20: 149-166.
29. Rheingold, H. (2000). The Virtual Community:
Homesteading on the Electronic Frontier. MIT Press.
30. Salguero, R. A. T., & Moran, R.M.B. (2002).
“Measuring Problem Video Game Playing in
Adolescents.” Addiction 97: 1601-1606.
31. Seay, A. F., Jerome, W.J., Lee, K.S., & Kraut, R.E.
(2004). Project Massive: A Study of Online Gaming
Communities. Proceedings of the CHI 2004 Conference
on Human Factors in Computing Systems, Vienna,
Austria.
32. Turkle, S. (1995). Life on the screen: Identity in the age
of the Internet. Simon and Schuster, New York, USA.
33. Yee, N. (2004). “The Daedalus Project.”
http://www.nickyee.com/daedalus/.
34. Young, K. S. (1998). Caught in the Net. New York, John
Wiley & Sons, Inc.
35. Zona, I. (2004). “Online Console Gaming, 2004.”
http://www.researchandmarkets.com/reports/225173/.
CHI 2007 Proceedings • Games April 28-May 3, 2007 • San Jose, CA, USA
838