Maladaptive cognitions predict changes in problematic gaming in highlyengaged adults: A 12-month longitudinal study Cameron J. Forrest, Daniel L. King, Paul H. Delfabbro PII: DOI: Reference:
S0306-4603(16)30372-0 doi:10.1016/j.addbeh.2016.10.013 AB 4987
To appear in:
Addictive Behaviors
Received date: Revised date: Accepted date:
16 November 2015 17 October 2016 21 October 2016
Please cite this article as: Forrest, C.J., King, D.L. & Delfabbro, P.H., Maladaptive cognitions predict changes in problematic gaming in highly-engaged adults: A 12-month longitudinal study, Addictive Behaviors (2016), doi:10.1016/j.addbeh.2016.10.013
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Title: Maladaptive cognitions predict changes in problematic gaming in highly-engaged adults: A 12month longitudinal study
T
Authors’ names and affiliations:
IP
Cameron J. Forrest
SC R
School of Psychology, The University of Adelaide
Corresponding author. School of Psychology, Level 4, Hughes Building, The University of Adelaide, Adelaide, SA 5005, Australia
Daniel L. King
MA
School of Psychology, The University of Adelaide
NU
Email:
[email protected]
Email:
[email protected] Paul H. Delfabbro
CE P
TE
Email:
[email protected]
D
School of Psychology, The University of Adelaide
AC
Conflict of interest: The authors have no conflicts of interest to declare.
ACCEPTED MANUSCRIPT Maladaptive cognitions predict changes in problematic gaming in highly-engaged adults: A 12-month longitudinal study
IP
T
Abstract
SC R
Understanding the role of maladaptive gaming-related cognitions may assist in screening and interventions for problematic gaming, including Internet gaming disorder (IGD). Cognitive-behavioral interventions that target specific cognitions related to gaming may be more effective than more general approaches that focus only on
NU
preoccupation with games. Although past research has identified cross-sectional associations between maladaptive cognitions and problematic gaming, it is less clear whether these cognitions can predict future
MA
changes in problematic gaming behavior. The present study employed an 18-item measure of gaming cognition, assessing perfectionism, cognitive salience, regret, and behavioural salience, to investigate potential changes in problematic gaming over a 12-month period. The sample included 465 Australian adults (84% male, Mage = 26.2
D
years). It was found that individuals who became problematic gamers over 12 months had higher baseline scores
TE
on perfectionism (d = 1.20), cognitive salience (d = 0.74) and regret (d = 0.69) than those who remained nonproblematic gamers. Problematic gamers who became non-problematic gamers had lower baseline
CE P
perfectionism scores (d = 0.62) than those who remained problematic gamers. Cognitive change accounted for an additional 28% of variance in problematic gaming scores beyond gender, age, and frequency of gaming.
AC
These findings suggest that maladaptive gaming-related cognitions could be screened in clinical trials to aid in case formulation and inform decisions on needed interventions to deliver optimal client outcomes.
Keywords: Internet gaming disorder; addiction; cognition; problem video-gaming; Cognitive-behavior therapy; longitudinal
ACCEPTED MANUSCRIPT 1.
Introduction Research interest in problematic gaming (i.e., video gaming) has grown exponentially over the past two
decades. A notable development in the field was the inclusion of ‘Internet gaming disorder’ (IGD) in Section III
T
of the DSM-5 as a condition warranting further study (APA 2013). One aspect of the IGD formulation that has
IP
attracted debate is its cognitive dimension. Specifically, it has been argued that ‘preoccupation’ (i.e., constantly
SC R
thinking about gaming and planning the next gaming activity) may be a useful criterion for screening and diagnosis, but it may not describe the complexity of mental processes that drive problem gaming. Some authors have suggested, for example, that problematic gaming has specific cognitive features such as irrational thinking,
NU
expectancy beliefs, or cognitive biases to gaming stimuli (Haagsma, Caplan, Peters & Pieterse 2013; Kim & Davis 2009; Li & Wang 2013; Wan & Chiou 2007). Some research studies have examined this assumption and
MA
reported a consistent positive association between the presence of maladaptive gaming cognitions and reporting a greater number of IGD symptoms (Liu et al. 2014; Peng & Liu 2010; Zhou, Yuan & Yao 2012). On this basis, it is often argued that interventions tailored to problem gaming should consider the influence of these cognitions
D
to develop more effective treatments (Griffiths & Meredith 2009; King & Delfabbro 2014a; Lemos, De Abreu &
TE
Sougey 2014). However, the literature is currently limited in relation to providing practical guidance on which specific cognitions may be most strongly related to problematic habits, recognising that some gaming-related
CE P
cognitions may operate only temporarily and fail to distinguish healthy from problematic gamers over time. A systematic review by King and Delfabbro (2014b) identified four main categories of maladaptive
AC
cognitions related to IGD, including: (1) beliefs about game rewards and tangibility, (2) maladaptive and inflexible rules about video-gaming behavior, (3) gaming as a source of self-esteem or ego-protection, and (4) gaming as a means of gaining social acceptance. Guided by this framework and other models (e.g., Caplan, 2010; Davis, 2001) a study by Forrest, King and Delfabbro (2016a) developed a scale of 18 items to measure a range of maladaptive gaming-related cognitions. A factor analysis indicated that the items loaded on four distinct factors: (1) Perfectionism or thoughts about wanting to be ‘the best’ at a particular game or games, blaming oneself if unable to play as well as expected, and the inability to cease play if close to completing some objective; (2) Cognitive salience or thoughts related to being unable to function without video-games, ruminating about games when not playing, and using play as a means of distraction from work or other activities; (3) Regret or thoughts related to personal responsibility for the negative consequences of play, and the need to reduce one’s frequency of play; and (4) Behavioural salience or thoughts related to the need to repeat ingame activities, as well as perceptions of the time investment. The four factors correlated positively with
ACCEPTED MANUSCRIPT frequency of play, two measures of problematic gaming habits, and symptoms of psychological distress. Additionally, the author reported significant differences between problematic and non-problematic gamers for all four cognition types, and only behavioural salience was a non-significant predictor of problematic gaming
T
status when controlling for all other variables. A similar pattern of results was reported in a similar study of 824
IP
adolescents, which reported that profiles of maladaptive gaming cognitions were distinct between problematic
SC R
and highly-engaged gamers (King & Delfabbro, 2016). On this preliminary evidence, it seems that maladaptive gaming-related cognitions are a promising research avenue with the potential to inform cognitive-behavioural
NU
therapies.
1.1 Longitudinal studies of video-gaming
MA
Prospective studies of electronic media habits (e.g., Bessiere, Pressman, Kiesler & Kraut 2010; Dong, Lu, Zhou & Zhao 2011; Dong, Wang, Yang & Zhao 2013; Ko et al. 2015; Sun et al. 2012; van den Eijnden, Spijkerman, Vermulst, van Rooij & Engels 2010; Yen et al. 2012; Yu & Shek 2013) have often subsumed
D
video-gaming under broad categories such as ‘screen-based behaviour’ (including television viewing) or
TE
‘Internet addiction’, which tends to include other activities such as information browsing, online social networking, online gambling, and viewing online pornographic material. For this reason, there have been
CE P
relatively few studies that have focussed specifically on longitudinal predictors of problematic gaming. One of the most comprehensive longitudinal studies reported by Gentile et al. (2011) followed 3034 Singaporean school
AC
children over two years and reported that the frequency of gaming, impulsivity, and social competence predicted pathological involvement with video-games two years later. Further studies have identified other long-term predictors of problematic gaming, including: parent-child closeness (Choo, Sim, Liau, Gentile & Khoo 2015); social competence, self-esteem, and loneliness (Lemmens, Valkenburg & Peter 2011); perceived behavioural control (Haagsma, King, Pieterse & Peters 2013); peer problems, male gender, low academic self-concept, and playing as a response to encountered problems (Möβle & Rehbein 2013); being in a single parent family, low school well-being, and weaker social interaction (Rehbein & Baier 2013); attention problems (Ferguson & Ceranoglu 2014); lower academic achievement, higher than average height, presence of older siblings, and previous victimisation by traditional bullying (Yang et al. 2014); and RSA withdrawal (i.e., a physiological mechanism comparable to sensation-seeking; Coyne et al. 2015). A feature of these studies is their focus on child or adolescent populations, rather than adult gamers. Young populations may be at greater risk of developing gaming problems (Kuss & Griffiths 2012), however many of these findings may not be
ACCEPTED MANUSCRIPT generalizable to older age groups (Forrest, King & Delfabbro 2016b). Adults are also more likely than adolescents to actively seek treatment for problem gaming and therefore it is helpful to understand the gaming-
T
related cognitive profiles of this population.
IP
1.2 The present study
SC R
The present study aimed to determine whether maladaptive gaming cognition can predict future changes in problematic video-gaming. If cognitive changes are associated with changes in problematic habits over time, then this might suggest that these cognitions warrant attention and modification in interventions (see
NU
King & Delfabbro 2014c). An effective means of identifying changes is to study both healthy and problematic users using a longitudinal design, given its advantages over retrospective designs that may yield inaccurate data
MA
(Scharkow, Festl & Quandt 2014). The present study aimed to recruit a sample of highly-engaged and problematic Australian adults, who completed a survey of their habits and gaming-related cognitions over 12 months. It was predicted that participants who became problematic gamers would score significanty higher on
D
the cognition measures than those who did not become problematic gamers. It was also predicted that
TE
participants who became non-problematic gamers would score lower on the cognition measures than those who remained problematic gamers. Finally, it was predicted that long-term changes in problematic gaming scores
AC
CE P
would be positively associated with long-term changes in cognition scores.
AC
CE P
TE
D
MA
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
Figure 1: Flow chart depicting study recruitment, attrition, and sample sizes T1 – T4
2.
Method
2.1 Participants and procedure The study was conducted as part of a broader research project which followed adult video-game players over the course of 12 months. Participants were recruited using flyers distributed throughout the campuses of two South Australian universities supplemented by online advertisements posted on websites frequented primarily by regular Australian video-gamers. All advertisements were addressed to individuals with strong interests in video-gaming. Participants were then directed to an online survey, asked to provide informed
ACCEPTED MANUSCRIPT consent and an email address enabling the researchers to contact them to participate in subsequent waves of data collection. Data were collected on four occasions at 3-month intervals (i.e., T1 = baseline, T2 = 3 months, T3 = 6 months, T4 = 9 months); in all instances, participants were asked about their gaming experiences during the past
T
three months, and each participant therefore provided data for a 12-month period. Participants were
IP
compensated with a $10 voucher for each completed survey.
SC R
Figure 1 presents each stage of the study, summarising drop-out between waves. Inclusion criteria were being aged 18 or older and playing games for at least 7 hours per week (i.e., an average of one hour per day) at baseline. Of the 657 initial responses received, 192 were excluded on the basis of these criteria or for providing
NU
incomplete or disingenuous responses. Final participation rates were: T1 N = 465, T2 N = 374, T3 N = 329, T4 N = 290. Attrition analysis (see Table 1) revealed that participants who dropped out of the study between T 1 and T4
MA
were on average younger, played video-games more frequently, had higher scores on the problem gaming measure and three of the four cognitions measures than those who remained in the study, although effects were
D
small.
TE
Table 1: Baseline (T1) comparisons between participants who did and did not remain in the study Did not complete T4
(N = 290)
(N = 175)
CE P
Completed T4
AC
Age
Weekly play (hours)
244 (84.1)
149 (85.1)
M (SD)
0.08 (1)
t(df)
d *
26.77 (7.59)
25.27 (6.57)
2.16 (463)
0.21
31.09 (19.34)
35.60 (21.83)
2.25* (333)
0.22
*
GAS
43.14 (13.26)
46.11 (15.09)
2.15 (331)
Perfectionism
11.50 (4.66)
12.13 (4.94)
1.37 (463)
Cognitive salience
10.90 (4.04)
11.91 (4.82)
2.33* (318)
Regret Behavioural salience Note: *p < .05, **p < .01.
2.2 Measures
Effect size
Χ2 (df)
n (%)
Gender=male
Test statistic
5.30 (2.61) 6.81 (2.73)
6.05 (2.97) 7.37 (2.96)
0.21
0.23
**
0.27
*
0.20
2.78 (331) 2.06 (463)
ACCEPTED MANUSCRIPT Participants completed an online survey requesting demographic information, frequency of gaming during a typical week in the last 3 months (in hours), and additional measures listed below. The survey took between 30-45 minutes to complete.
T
Maladaptive gaming cognitions: This survey contained 18 items used to measure maladaptive
IP
cognitions related to video-gaming, separated into four subscales: Perfectionism (6 items, T1 α = .83) measured
SC R
thoughts such as wanting to be a better player than others, blaming oneself or feelings of failure if unable to complete or solve parts of a game, and being unable to cease play until objectives have been completed. Cognitive salience (6 items, T1 α = .81) measured thoughts related to being unable to function or focus on other
NU
activities due to ruminating about video-games, as well as the importance of gaming to one’s identity. Regret (3 items, T1 α = .81) measured thoughts about feeling guilty or wanting to reduce one’s frequency of play due to
MA
negative consequences. Behavioural salience (3 items, T1 α = .74) measured thoughts about repetitive in-game behaviours, as well as difficulty in ceasing play due to time investment. Responses were given on a 5-point scale where 1 = ‘Never’, 2 = ‘Rarely’, 3 = ‘Some of the time’, 4 = ‘Most of the time’, and 5 = ‘Always’. Further
D
information on how these scales were constructed has been reported elsewhere (Forrest et al. 2016a).
TE
The Game Addiction Scale (GAS; Lemmens, Valkenburg & Peter 2009; T1 α = .93) is a 21-item measure of the extent to which an individual’s video-gaming habits have become problematic in terms of
CE P
symptoms related to salience, tolerance, mood modification, relapse, withdrawal, conflict, and problems caused by play. Responses are given on a 5-point Likert scale (‘never’ to ‘very often’), providing a potential minimum score of 21 and a potential maximum of 105. Each of the 7 symptom types is measured by individual subscales
AC
which are considered ‘met’ if the average score on their items is 3 (‘sometimes’) or greater. Participants were classified as problematic gamers if they met at least 4 of the 7 criteria. Rates of problematic involvement were: T1 15.7%, T4 9.2%.
2.3 Analytical strategy Descriptive statistics were produced using all available data from completed surveys across the four waves of measurement. For the longitudinal analyses only (T1 – T4), multiple imputation was used to provide values where data were missing. Multiple imputation uses all available data to provide estimations for any missing data, the final values of which are calculated as the averages of five such estimations. Imputing data allows analyses to be performed on responses from all participants, rather than excluding cases listwise or pairwise. Little’s test indicated that data could be considered missing completely at random for the purposes of
ACCEPTED MANUSCRIPT analyses, χ2(366) = 410.76, p > .05. Because data were only missing where participants had dropped out between waves, the number of participants for whom values were imputed was equivalent to the number of drop outs (i.e. for N = 175 cases or 37.6% of the baseline sample).
T
As the purpose of the study was to investigate whether baseline cognitions could be used to predict
IP
changes in behaviour at 12 months, comparisons were focussed on differences between T1 and T4, to reduce the
SC R
complexity of group assignment and in line with research by Ko et al. (2007). First, participants were classified into one of four groups according to whether they could be classified as problematic gamers at T1 and T4. Participants who were non-problematic gamers at both T1 and T4 formed the ‘Never’ group. Those who were
NU
non-problematic gamers at T1 but were problematic gamers at T4 formed the ‘Starts’ group. Those who were problematic gamers at T1 but were non-problematic gamers at T4 formed the ‘Stops’ group. Finally, those who
MA
were problematic gamers at both T1 and T4 formed the ‘Stays’ group. For ease of comparability, these labels were consistent with those used by Ko et al. (2007) and Gentile et al. (2011). These groups were compared using chi-square analyses and one-way ANOVA with planned comparisons. Comparisons between the Never
D
and Starts groups were used to identify potential risk factors of problematic gaming, whereas comparisons
TE
between the Stops and Stays groups were used to identify potential remission factors. A hierarchical regression model was then constructed predicting changes in problematic gaming scores (calculated as T4–T1). For
CE P
consistency with past literature, this regression was based on analyses used by Ko et al. (2007), Meerkerk et al. (2006) and Scharkow et al. (2014). Gender, age, frequency of play, and baseline problematic gaming scores
AC
were entered in the first step, and baseline and changes in cognition scores were entered in the second step. All analyses were conducted using SPSS for Windows, version 21.
3.
Results
Descriptive statistics by measurement occasion are displayed in Table 2. Significant differences were found between all four groups for perfectionism (F(3, 461) = 50.31, p < .001), cognitive salience (F(3, 461) = 75.67, p < .001), regret (F(3, 461) = 60.37, p < .001) and behavioural salience (F(3, 461) = 17.65, p < .001), but not for age (F(3, 461) = 1.80, p > .05). Table 3 presents the results of baseline comparisons (chi-square and planned contrasts) between the Never/Starts and Stops/Stays groups. Participants who became problematic gamers during the study had significantly higher perfectionism, cognitive salience, and regret scores at baseline than those who remained non-problematic gamers, and those who became non-problematic gamers had
ACCEPTED MANUSCRIPT significantly higher perfectionism scores at baseline than those who remained problematic gamers. These findings indicated that perfectionism, cognitive salience, and regret were the most relevant cognitions when
T
comparing problematic and non-problematic gamers longitudinally.
T2 (N = 374)
T3 (N = 329)
SC R
T1 (N = 465)
IP
Table 2: Descriptive statistics by measurement occasion
T4 (N = 290)
n (%)
Gender=male
393 (84.5)
312 (83.4)
276 (83.9)
244 (84.1)
Age Weekly
play
NU
M (SD) 26.21 (7.25)
26.63 (7.28)
27.10 (7.45)
27.55 (7.64)
32.79 (20.40)
31.36 (18.77)
28.01 (16.31)
29.31 (16.52)
42.56 (13.93)
41.06 (13.63)
40.20 (13.06)
11.28 (4.57)
10.82 (4.21)
10.56 (4.25)
11.12 (4.34)
10.98 (4.23)
10.92 (4.19)
5.35 (2.63)
5.18 (2.57)
5.11 (2.61)
6.68 (2.72)
6.29 (2.54)
6.48 (2.65)
Perfectionism
11.74 (4.78)
Cognitive salience
11.28 (4.37)
Regret
5.58 (2.77)
Behavioural
7.02 (2.83)
D
44.26 (14.04)
TE
GAS
MA
(hours)
AC
CE P
salience
ACCEPTED MANUSCRIPT
Starts
(N = 374)
(N = 18)
Test statistic
1.33
Perfectionism Cognitive salience
26.54 (7.33) 10.61 (3.86) 10.11 (3.25)
(1)
(N = 25)
t (df)
23.11 (5.74) 15.33 (5.36) 12.56 (4.10)
1.96 (461) 3.69
**
(18)
2.49
*
(18)
*
(18)
4.89 (2.15)
6.39 (2.52)
2.48
Behavioural salience
6.59 (2.64)
7.67 (2.95)
1.66 (461)
TE D
Regret
d
AC
CE P
Note: *p < .05, **p < .01.
Test statistic
Effect size
Χ2 (df)
N (%)
37 (77.1)
US
17 (94.4)
M (SD) Age
Stays
CR
Χ (df)
23 (92.0)
M (SD)
2.50
(1)
t (df)
25.42 (6.85)
24.96 (7.40)
0.26 (461)
1.20
15.50 (4.68)
18.72 (6.19)
2.28* (39)
0.74
16.48 (4.84)
17.88 (5.04)
1.14
(47)
0.69
8.83 (3.22)
9.08 (3.09)
0.32
(51)
9.23 (2.86)
8.72 (2.85)
0.77 (461)
MA N
316 (84.5)
Stops (N = 48)
2
N (%) Gender=male
Effect size
IP
Never
T
Table 3: Baseline comparisons (planned contrasts) between Never, Starts, Stops and Stays groups using GAS criteria (N = 465)
d
0.62
ACCEPTED MANUSCRIPT Table 4 presents a summary of the hierarchical regression models predicting T1–T4 changes in problematic gaming scores. Higher baseline perfectionism, cognitive salience and regret scores, and increases in cognitive salience, regret, behavioural salience and weekly frequency of play predicted increases in problematic
SC R
IP
scores accounted for a further 28% of changes in problematic gaming scores.
T
gaming scores over 12 months. When controlling for all other variables, baseline and changes in cognition
Table 4: Hierarchical regression predicting changes in problem gaming scores from baseline and changes in gaming-related cognitions over 12 months (N = 465)
NU
B (SE)
Step 1
β
ΔR2
8.15** (2.19)
Constant
MA
Gender=male Age (baseline) Age change
D
Frequency of play (baseline) Frequency of play change
TE
GAS (baseline)
CE P
Step 2
-0.24
(0.86)
-0.01
0.02
(0.05)
0.01
-0.29
(0.32)
-0.02
(0.02)
0.05
(0.03)
0.12
-0.70** (0.04)
-0.77
0.03 0.09
**
.459
Perfectionism (baseline)
0.26* (0.12)
0.10
Perfectionism change
0.09
(0.13)
0.03
0.90
**
(0.15)
0.31
1.34
**
(0.14)
0.43
0.88
*
(0.20)
0.19
Regret change
1.32
**
(0.21)
0.25
Behavioural salience (baseline)
0.15
(0.18)
0.03
(0.19)
0.09
Cognitive salience (baseline)
AC
Cognitive salience change Regret (baseline)
Behavioural salience change
0.46
*
.280 Note: *p < .05, **p < .001.
4.
Discussion This study examined whether maladaptive gaming-related cognitions may predict changes in
problematic gaming behaviour over time. In a sample of highly-engaged gaming adults, it was found that
ACCEPTED MANUSCRIPT normal individuals who became problematic gamers over the course of 12 months scored significantly higher on measures of perfectionism, cognitive salience, and regret at baseline than those who remained non-problematic gamers. Gamers whose habits became non-problematic during this period had lower baseline perfectionism
T
scores than those who remained problematically involved. These findings therefore suggest that there may exist
IP
longitudinal associations between changes in maladaptive gaming-related cognitions and problematic gaming.
SC R
Past research and reviews (Choo et al. 2015; Davis 2001; Winkler et al., 2013) have emphasised that CBT may be an effective treatment for IGD. However, there is potential for these interventions to be more tailored to the specific cognitive dimensions of gaming, and include outcome measures of gaming cognition
NU
(King & Delfabbro 2014c; Peng & Liu, 2010). This study identified four types of gaming-related cognitions which appear to change as a function of problematic gaming. Differences in cognitive salience and regret were
MA
found to be particularly influential on levels of gaming problems. Although changes in behavioural salience scores significantly predicted changes in problematic gaming scores, planned contrasts found non-significant differences between the Never/Starts and Stops/Stays groups. These findings lend support to models of gaming
D
that emphasise these cognitive features (Caplan 2010; Davis 2001). However, this does not rule out the
TE
possibility of other important cognitive aspects of gaming, particularly as gaming itself continues to evolve via new technologies, such as virtual reality devices. For example, competition and social interaction (Caplan 2003;
CE P
Forrest et al. 2016b; Sherry, Greenberg, Lucas & Lachlan 2006) have been associated with problematic gaming and there may be useful cognitions associated with these motivations to play.
AC
Longitudinal regression analyses found that cognitive salience and regret were the strongest predictors of long-term changes in problematic gaming scores. Higher baseline regret, higher baseline cognitive salience, and increases in regret, cognitive salience and behavioural salience predicted increases in problematic gaming scores over 12 months. Furthermore, the combination of baseline and changes in all four cognition types accounted for a further 28% of variance in changes in problematic gaming than the combination of gender, age, frequency of play, and baseline problematic gaming. These variables had been noted as the most consistent predictors of problematic habits in the IGD literature, but only changes in frequency of play remained a significant predictor when controlling for cognition scores. Interestingly, although baseline perfectionism scores were a significant predictor of problem gaming, changes in perfectionism were not. This supports previous suggestions that certain types of cognitions are more relevant to problematic habits than others (Forrest et al. 2016b). Discrepancies between the regression model and planned contrasts likely reflect the use of problematic
ACCEPTED MANUSCRIPT gaming scores as a continuous measure versus using cut-off scores to classify individuals as problematically involved. A relatively high proportion of problematic gamers became non-problematic gamers during the study.
T
Using GAS criteria, 66% of participants who were classified as problematic gamers at T1 became non-
IP
problematic gamers by T4. This finding was consistent with studies that have identified declines in symptom
SC R
severity among adults over time (Konkolÿ et al., 2015; Möβle & Rehbein 2013; Scharkow et al. 2014). In a similar study of Australian adult regular gamers, King et al. (2013) reported that normal and problematic gamers both experienced decreases in problem gambling symptoms over 18 months. In line with their suggestions, these
NU
findings may indicate maturation or spontaneous recovery effects which have been observed for other addictions such as pathological gambling (Anglin, Brecht, Woodward & Bonett 1986; Toneatto et al. 2008). Further
MA
research could investigate the variables that contribute to these declines in problematic gaming over time, with a particular focus on modifiable psychological, social and environmental variables that might inform education
Limitations and future directions
TE
5.
D
for individuals seeking to moderate their gaming activity.
This study had some limitations that warrant discussion. The cognition measure developed for this
CE P
study is only preliminary at this stage and would benefit from further validation using other samples. The measure may benefit from item refinement using cognitive testing, which may identify other cognition types
AC
associated with problematic gaming. The study sample was recruited from multiple sources and is unlikely to be representative of the general gaming population, especially since highly-engaged gamers were targeted in order to provide higher rates of problematic involvement. Similarly, the GAS was developed using an adolescent population and comparatively few studies have examined its psychometric properties in adult populations. Furthermore, despite recruiting a sample with high rates of problematic gaming, sample sizes for the Starts, Stops and Never groups were small. Small sample sizes increase risk of type II error. The group classifications – ‘Starts, Stops, Stays, and Never’ – referred to problematic gaming status at the two measurement occasions, and did not take into account possible dynamic changes in the interim. Some participants classified as ‘Never’ may have met some problem criteria during the 9 months between T1 and T4, and this represents a limitation of using only cut-off scores for classifying individuals as problematic gamers. However, it is also the case that such minor changes in problematic gaming status would have been transient. The attrition rate was also relatively high despite offering rewards for participation. Another issue is the analyses carried the assumption that
ACCEPTED MANUSCRIPT maladaptive cognitions are a cause of problematic habits. However, it is likely that the presence of one symptom or condition reinforces the other. For example, Gentile et al. (2011) found that impulsivity predicted pathological gaming, but also increased after participants became pathological gamers. These finding should not
T
be interpreted as an attempt to resolve the ‘chicken or egg’ problem of whether maladaptive cognitions or
IP
problematic behaviours arise first. To this end, case studies and qualitative data would be especially useful in
6.
SC R
explaining their development.
Conclusion
NU
Problematic gaming is thought to be maintained by specific maladaptive beliefs about gaming activities. Past longitudinal studies of problematic gaming have often focussed on consequences rather than
MA
causes of play. This study suggests that changes in gaming-related cognitive salience, regret, and perfectionism, may explain longer term changes in problematic gaming habits. This study presents a preliminary measure of these cognitions which demonstrate cross-sectional and longitudinal associations with problematic gaming
D
involvement. It is hoped that these findings may inform the development of interventions for clients seeking to
TE
address negative impacts of gaming on other areas of functioning (e.g., sleep, work productivity, interpersonal relationships), by enabling individuals to identify the types of maladaptive thought patterns that initiate and
CE P
sustain problematic gaming sessions.
References
AC
American Psychiatric Association (2013). Diagnostic and statistical manual of mental disorders (DSM-5). 5th ed. Washington, DC: Author. Anglin, M.D., Brecht, M.L., Woodward, J.A., & Bonett, D.G. (1986). An empirical study of maturing out: Conditional
factors.
International
Journal
of
the
Addictions,
21(2),
233–246.
http://dx.doi.org/0.3109/108260880609063452 Bessiere, K., Pressman, S., Kiesler, S., & Kraut, R. (2010). Effects of Internet use on health and depression: A longitudinal study. Journal of Medical Internet Research, 12(1). http://dx.doi.org/10.2196/jmir.1149 Caplan, S.E. (2003). Preference for online social interaction: A theory of problematic Internet use and psychosocial
well-being.
Communication
http://dx.doi.org/10.1177/0093650203257842
Research,
30(6),
625–648.
ACCEPTED MANUSCRIPT Caplan, S.E. (2010). Theory and measurement of generalized problematic Internet use: A two-step approach. Computers in Human Behavior, 26(5), 1089–1097. Carli, V., Durkee, T., Wasserman, D., Hadlaczky, G., Desplains, R., Kramarz, E., … Kaess, M. (2013). The
T
association between pathological Internet use and comborbid psychopathology: A systematic review.
IP
Psychopathology, 46(1), 1–13. http://dx.doi.org/10.1159/000337971
SC R
Choo, H., Sim, T., Liau, A.K.F., Gentile, D.A., & Khoo, A. (2015). Parental influences on pathological symptoms of video-gaming among children and adolescents: A prospective study. Journal of Child and Family Studies, 24, 1429–1441. http://dx.doi.org/10.1007/s10826-014-9949-9
NU
Coyne, S.M., Dyer, W.J., Densley, R., Money, N.M., Day, R.D., & Harper, J.M. (2015). Physiological indicators of pathological video-game use in adolescence. Journal of Adolescent Health, 56(3), 307–
MA
313. http://dx.doi.org/10.1016/j.adohealth.2014.10.271
Davis, R.A. (2001). A cognitive-behavioural model of pathological Internet use. Computers in Human Behavior, 17(2), 187–195. http://dx.doi.org/10.1016/S0747-5632(00)00041-8
D
Dong, G., Lu, Q., Zhou, H., & Zhao, X. (2011). Precursor or sequela: Pathological disorders in people with
TE
Internet addiction disorder. PLoS ONE: 6(2): e14703. http://dx.doi.org/10.1371/journal.pone.0014703 Dong, G., Wang, J., Yang, X., & Zhou, H. (2013). Risk personality traits of Internet addiction: A longitudinal
CE P
study of Internet-addicted Chinese university students. Asia-Pacific Psychiatry, 5(4), 316–321. http://dx.doi.org/10.1111/j.758-5872.2012.00185.x Ferguson, C.J., & Ceranoglu, T.A. (2014). Attention problems and pathological gaming: Resolving the ‘chicken egg’
in
AC
and
a
prospective
analysis.
Psychiatric
Quarterly,
85(1),
103–110.
http://dx.doi.org/10.1007/s11126-013-9276-0 Forrest, C.J., King, D.L., & Delfabbro, P.H. (2016a). The measurement of maladaptive cognitions underlying problematic video-game playing among adults. Computers in Human Behavior, 55, 399–405. http://dx.doi.org/10.1016/j.chb.2015.09.017 Forrest, C.J., King, D.L., & Delfabbro, P.H. (2016b). The socialising patterns and social networks of adults with and without Internet gaming disorder. Manuscript submitted for publication. Gentile, D.A., Choo, H., Liau, A., Sim, T., Li, D., Fung, D., & Khoo, A. (2011). Pathological video game use among
youths:
A
two-year
longitudinal
http://dx.doi.org/10.1542/peds.2010.1353
study.
Pediatrics,
127(2),
e319–e329.
ACCEPTED MANUSCRIPT Griffiths, M.D., & Meredith, A. (2009). Videogame addiction and its treatment. Journal of Contemporary Psychotherapy, 39(4), 247–253. http://dx.doi.org/10.1007/s10879-009-9118-4 Haagsma, M.C., Caplan, S.E., Peters, O., & Pieterse, M.E. (2013). A cognitive-behavioral model of problematic
T
online gaming in adolescents aged 12–22 years. Computers in Human Behavior, 29(1), 202–209.
IP
http://dx.doi.org/10.1016/j.chb.2012.08.006
SC R
Haagsma, M.C., King, D.L., Pieterse, M.E., & Peters, O. (2013). Assessing problematic video gaming using the theory of planned behavior: A longitudinal study of Dutch young people. International Journal of Mental Health and Addiction, 11(2), 172–185. http://dx.doi.org/10.1007/s11469-012-9407-0
NU
Kim, H., & Davis, K.E. (2009). Toward a comprehensive theory of problematic Internet use: Evaluating the role of self-esteem, anxiety, flow, and the self-rated importance of Internet activities. Computers in Human
MA
Behavior, 25(2), 490–500. http://dx.doi.org/10.1016/j.chb.2008.11.001 King, D.L., & Delfabbro, P.H. (2014a). Internet gaming disorder treatment: A review of definitions of diagnosis and
treatment
outcome.
Journal
of
Clinical
Psychology,
70(10),
942–955.
D
http://dx.doi.org/10.10002/jclp.22097
TE
King, D.L., & Delfabbro, P.H. (2014b). The cognitive psychology of Internet gaming disorder. Clinical Psychology Review, 34(4), 298–308. http://dx.doi.org/10.1016/j.cpr.2014.03.006
factors
CE P
King, D.L., & Delfabbro, P.H. (2014c). Is preoccupation an oversimplification? A call to examine cognitive underlying
Internet
gaming
disorders.
Addiction,
109,
1566–1567.
AC
http://dx.doi.org/10.1111/add.12547 King, D. L. & Delfabbro, P. H. (2016). The cognitive psychopathology of Internet gaming disorder in adolescence. Journal of Abnormal Child Psychology. DOI: 10.1007/s10802-016-0135-y. King, D.L., Delfabbro, P.H., & Griffiths, M.D. (2013). Trajectories of problem video gaming among adult regular gamers: An 18-month longitudinal study. Cyberpsychology, Behavior, and Social Networking, 16(1), 72–76. http://dx.doi.org/10.1089/cyber.2012.0062 Ko, C.-H., Wang, P.-W., Liu, T.-L., Yen, C.-F., Chen, C.-S., & Yen, J.-Y., (2015). Bidirectional associations between family factors and Internet addiction among adolescents in a prospective investigation. Psychiatry and Clinical Neurosciences, 69(4), 192–200. http://dx.doi.org/10.1111/pcn.12204 Ko, C.-H., Yen, J.Y., Yen, C.F., Lin, H.C., & Yang, M.J. (2007). Factors predictive for incidence and remission of Internet addiction in young adolescents: A prospective study. Cyberpsychology and Behavior, 10(4), 545–551. http://dx.doi.org/10.1089/cpb.2007.9992
ACCEPTED MANUSCRIPT Konkolÿ Thege, B., Woodin, E.M., Hodgins, D.C., & Williams, R.J. (2015). Natural course of behavioural addictions: A 5-year longitudinal study. BMC Psychiatry, 15(4). http://dx.doi.org/10.1186/s12888-0150383-3
research.
Journal
of
Behavioral
1(1),
3–22.
SC R
http://dx.doi.org/10.1556/JBA.1.2012.1.1
Addictions,
IP
empirical
T
Kuss, D.J., & Griffiths, M.D. (2012). Online gaming addiction in children and adolescents: A review of
Lemmens, J.S., Valkenburg, P.M., & Peter, J. (2009). Development and validation of a game addiction scale for adolescents. Media Psychology, 12(1), 77–95. http://dx.doi.org/10.1080/15213260802669458
NU
Lemmens, J.S., Valkenburg, P.M., & Peter, J. (2011). Psychosocial causes and consequences of pathological gaming. Computers in Human Behavior, 27(1), 144–152. http://dx.doi.org/10.1016/j.chb.2010.07.015
MA
Lemos, I.L., De Abreu, C.N., & Sougey, E.B. (2014). Internet and video game addictions: A cognitivebehavioral approach. Revista de Psiquiatria Clinica, 41(3), 82–88. http://dx.doi.org/10.1590/010160830000000016
Children
and
TE
adolescents.
D
Li, H., & Wang, S. (2013). The role of cognitive distortion in online gaming addiction among Chinese Youth
Services
Review,
35(9),
1468–1475.
http://dx.doi.org/j.childyouth.2013.05.021
CE P
Liu, G.C., Yen, J.Y., Chen, C.Y., Yen, C.F., Chen, C.S., Lin, W.C., & Ko, C.H. (2014). Brain activation for response inhibition under gaming cue distraction in Internet gaming disorder. Kaohsiung Journal of Medical Sciences, 30(1), 43–51. http://dx.doi.org/10.1016/j.kjms.2013.08.005
AC
Meerkerk, G.J., van den Eijnden, R.J.J.M., & Garretsen, H.F.L. (2006). Predicting compulsive Internet use: It’s all about sex! Cyberpsychology and Behavior, 9(1), 95–103. http://dx.doi.org/10.1089/cpb.2006.9.95 Möβle, T., & Rehbein, F. (2013). Predictors of problematic video game usage in childhood and adolescence. Sucht, 59(3), 153–164. http://dx.doi.org/10.1024/0939-5911.a000247 Peng, W., & Liu, M. (2010). Online gaming dependency: A preliminary study in China. Cyberpsychology, Behavior, and Social Networking, 13(3), 329–333. http://dx.doi.org/10.1089/cyber.2009.0082 Rehbein, F., & Baier, D. (2013). Family-, media-, and school-related risk factors of video game addiction: A 5year longitudinal study. Journal of Media Psychology, 25(3), 118–128. http://dx.doi.org/10.1027/18641105/a000093 Salguero, R.A.T., & Morán, R.M.B. (2002). Measuring problem video game playing in adolescents. Addiction, 97(12), 1601–1606. http://dx.doi.org/10.1046/j.1360-0443.2002.00218.x
ACCEPTED MANUSCRIPT Scharkow, M., Festl, R., & Quandt, T. (2014). Longitudinal patterns of problematic computer game use among adolescents
and
adults–a
2-year
panel
study.
Addiction,
109(11),
1910–1917.
http://dx.doi.org/10.1111/add.12662
T
Sherry, J., Greenberg, B., Lucas, S., & Lachlan, K. (2006). Video game uses and gratifications as predictors of
IP
use and game preference. In P. Vorderer and J. Bryant (Eds.), Playing computer games: Motives,
SC R
responses and consequences (pp. 213–224). Mahwah, New Jersey: Erlbaum.
Sun, P., Johnson, C.A., Palmer, P., Arpawong, T.E., Unger, J.B., Xie, B., … Sussman, S. (2012). Concurrent and predictive relationships between compulsive Internet use and substance use: Findings from
NU
vocational high school students in China and the USA. International Journal of Environmental Research and Public Health, 993), 660–673. http://dx.doi.org/10.3390.ijerph9030660
MA
Toneatto, T., Cunningham, J., Hodgins, D., Adams, M., Turner, N., & Koski-Jannes, A. (2008). Recovery from problem gaming without formal treatment. Addiction Research & Theory, 16(2), 111–120. http://dx.doi.org/10.1080/16066350801923638
D
van den Eijnden, R.J.J.M., Spijkerman, R., Vermulst, A.A., van Rooij, T.J., & Engels, R.C.M.E. (2010).
TE
Compulsive Internet use among adolescents: Bidirectional parent-child relationships. Journal of Abnormal Child Psychology, 38(1), 77–89. http://dx.doi.org/10.1007/s10802-009-9347-8
CE P
van Rooij, A.J., Schoenmakers, T.M., van den Eijnden, R.J.J.M., & van de Mheen, D. (2010). Compulsive Internet use: The role of online gaming and other Internet applications. Journal of Adolescent Health,
AC
47(1), 51–57. http://dx.doi.org/10.1016/j.adohealth.2009.12.021 Wan, C.-S., & Chiou, W.-B. (2007). The motivations of adolescents who are addicted to online games: A cognitive perspective. Adolescence, 42(165), 179–197. Winkler, A., Dörsing, B., Rief, W., Shen, Y., & Glombiewski, J.A. (2013). Treatment of internet addiction: A meta-analysis.
Clinical
Psychology
Review,
33(2),
317–329.
http://dx.doi.org/10.1016/j.cpr.2012.12.005 Yang, S.J., Stewart, R., Lee, J.Y., Kim, J.M., Kim, S.W., Shin, I.S., & Yoon, J.S. (2014). Prevalence and correlates of problematic Internet experiences and computer-using time: A two-year longitudinal study in
Korean
school
children.
Psychiatric
Investigation,
11(1),
24–31.
http://dx.doi.org/10.4306/pi.2014.11.1.24 Yen, J.-Y., Yen, C.-F., Chen, C.-S., Chang, Y.-H., Yeh, Y.-C., & Ko, C.-H. (2012). The bidirectional interactions between addiction, behaviour approach and behaviour inhibition systems among
ACCEPTED MANUSCRIPT adolescents
in
a
prospective
study.
Psychiatry
Research,
200(2-3),
588–592.
http://dx.doi.org/10.1016/j.psychres.2012.03.015 Yu, L., & Shek, D.T.L. (2013). Internet addiction in Hong Kong adolescents: A three-year longitudinal study. of
Pediatric
and
Adolescent
Gynecology,
26(3),
S10–S17.
T
Journal
IP
http://dx.doi.org/10.1016/j.pag.2013.010
deficits
in
individuals
with
an
Internet
SC R
Zhou, Z., Yuan, G., & Yao, J. (2012). Cognitive biases toward Internet game-related pictures and executive game
AC
CE P
TE
D
MA
NU
http://dx.doi.org/10.1371/journal.pone.0048961
addiction.
PLoS
ONE,
7(11):
e48961.
ACCEPTED MANUSCRIPT Role of Funding Sources All authors were employed by the University of Adelaide at the time of writing and no external funding was
T
used in support of this study.
IP
Contributors
SC R
All authors contributed to the study design, analysis, and interpretation of data. Author A wrote the first draft of the manuscript and all authors contributed to and have approved the final manuscript.
NU
Conflict of Interest
AC
CE P
TE
D
MA
All authors declare that they have no conflicts of interest.
ACCEPTED MANUSCRIPT Highlights
CE P
TE
D
MA
NU
SC R
IP
T
Longitudinal study of cognitions related to problematic video-game playing We use a recently developed measure of gaming-related cognitions Cognitions are perfectionism, cognitive salience, regret, and behavioural salience These cognitions act as both risk and remission factors for problematic behaviour We discuss implications for interventions and clinical trials
AC
