Professional Documents
Culture Documents
2019 Developmentofthe PMGQand Prevalenceof Mobile Gaming Addiction Among Adolescentsin Taiwan Cyberpsychol Behav Soc Netw
2019 Developmentofthe PMGQand Prevalenceof Mobile Gaming Addiction Among Adolescentsin Taiwan Cyberpsychol Behav Soc Netw
Yuan-Chien Pan, MS,1 Yu-Chuan Chiu, MD,2 and Yu-Hsuan Lin, MD, PhD3–6
Downloaded by NATIONAL TAIWAN UNIV. HOSPITAL from www.liebertpub.com at 10/17/19. For personal use only.
Abstract
Mobile gaming has gained popularity among adolescents, and an increase in problematic use has been reported.
The aims of this study are as follows: (a) develop a self-report questionnaire, the Problematic Mobile Gaming
Questionnaire (PMGQ); (b) establish a validated cutoff value using structured interviews; and (c) evaluate the
prevalence of mobile gaming addiction in adolescents. The PMGQ was built as a 12-item questionnaire rated on
a 4-point Likert scale to evaluate the symptoms of problematic mobile gaming (PMG). The construct validity of
the PMGQ was examined using exploratory factor analysis. Overall, 10,775 students with smartphones from
grade 4 to senior high school were recruited to complete the questionnaire. A total of 113 senior high school
students were interviewed using previously developed criteria for PMG to develop an optimal cutoff point
measuring sensitivity, specificity, and diagnostic accuracy. The cutoff point was determined using the Youden
index and optimal diagnostic accuracy. The PMGQ showed good internal consistency (Cronbach’s a = 0.92) and
adequate diagnostic efficiency (area under the receiver operating characteristic curve = 0.802). The items re-
vealed three factors of addiction: compulsion, tolerance, and withdrawal. For the PMGQ, a cutoff point of 29/30
demonstrated the most optimal Youden index and diagnostic accuracy. Demographic data showed that the
proportion of PMG was 19.1 percent among elementary school students, 20.5 percent among junior high school
students, and 19.0 percent among senior high school students. The PMGQ demonstrated appropriate validity
and accuracy in the assessment for PMG.
Keywords: problematic mobile gaming, mobile gaming addiction, Internet gaming disorder, smartphone addiction
1
Department of Psychology, National Taiwan University, Taipei, Taiwan.
2
Department of Psychiatry, MacKay Memorial Hospital, Taipei, Taiwan.
3
Institute of Population Health Sciences, National Health Research Institutes, Miaoli, Taiwan.
4
Department of Psychiatry, National Taiwan University Hospital, Taipei, Taiwan.
5
Department of Psychiatry, College of Medicine, National Taiwan University, Taipei, Taiwan.
6
Institute of Health Behaviors and Community Sciences, College of Public Health, National Taiwan University, Taipei, Taiwan.
662
ASSESSMENT OF PROBLEMATIC MOBILE GAMING 663
games exceeded personal computer (PC) and console game terview; and (c) to survey the prevalence of PMG among
revenue for the first time in 2016.7 Therefore, we use mobile students at elementary, junior, and senior high school levels.
games to represent online smartphone or tablet games. An
official survey by the Institute for Information Industry showed Materials and Methods
that 58.7 percent (12.25 million) of people over the age of 12
Participants and procedure
years owned a smartphone in Taiwan in 2015. The top five
types of apps were those for games (53.7 percent), social We recruited 16,283 students from 169 schools (112 ele-
networking (27.8 percent), tools (27.7 percent), music (27.4 mentary, 39 junior high, and 18 senior high schools) in Tai-
percent), and videos (24.4 percent).8 This report suggested that wan. In total, 10,775 students (6,015 boys and 4,760 girls)
mobile games are a crucial factor in the excessive use of, or the reported owning a smartphone. The students’ ages ranged from
addiction to, smartphones. Portability allows for shorter use 10 to 18 (M = 13.06 years, standard deviation = 1.74) years.
periods of gaming compared with the relatively long periods of After providing informed consent, all participants completed a
PC-based Internet gaming. The short use periods of smart- paper-and-pencil survey that collected demographic data,
phones result in distractions, which might lead to functional characteristics of smartphone use, PMG data, and symptoms of
impairment, which can culminate in incidents such as vehicle IGD. A total of 113 senior high students were further invited to
crashes or near-crashes.9 A case report of tendon rupture as- an interview session. The interviews were conducted indi-
Downloaded by NATIONAL TAIWAN UNIV. HOSPITAL from www.liebertpub.com at 10/17/19. For personal use only.
sociated with excessive mobile gaming10 suggested that non- vidually by two psychologists to validate and determine the
pharmacological effects of mobile gaming such as pleasure cutoff point of the PMGQ. The structured interview of PMG
and excitement, similar to those of PC gaming, suppress pain was modified from the criteria of smartphone addiction.25
perception in pediatric patients11 and patients with burns.12
These effects can lead to compulsive–addictive behaviors. Measures
Mobile gaming has been increasingly linked to adolescent
Problematic Mobile Gaming Questionnaire. The PMGQ
health problems. The close combination of gaming and so-
was built based on the SPAI considering the fact that the
cial networking available on mobile phones may play a
definition of PMG or mobile gaming addiction is still a
crucial role in its addictive nature.13 Adolescent gamers can
controversial concept in the field of behavioral addiction.24
readily share scores and achievements on social media and
The 16 candidate items for the PMGQ were revised from the
receive immediate feedback. Inviting friends to join mobile
SPAI. A total of 10 items overlapped with the short form
gaming sessions through social media is easier than for PC-
Smartphone Addiction Inventory (SPAI-SF). Additional
based online games. Although adolescents have potential to
items were included with regard to the uniqueness of mobile
build and enhance existing social relationships, they may
gaming and diagnostic characteristics of IGD in the DSM-5.
also have negative social experiences, such as bullying or
Participants were asked to rate items on a 4-point Likert scale
trolling behaviors.14 Furthermore, in a qualitative study in-
(1, strongly disagree; 2, somewhat disagree; 3, somewhat
vestigating sleep deprivation for adolescents, boys were
agree; and 4, strongly agree).
more likely to report engaging in nighttime gaming on their
smartphones, which may negatively affect sleep quality.15
Ten-Item Internet Gaming Disorder Test
Self-report questionnaires have been developed to assess
smartphone addiction16–19 and IGD.20–22 The portability of The Ten-Item Internet Gaming Disorder Test (IGDT-10)
smartphones has resulted in rapid evolution of mobile gam- comprises 10 items and assesses levels of IGD.20 Each cri-
ing. Mobile games have become a crucial factor in smart- terion of IGD in the DSM-5 was operationalized using a
phone use and warrant a dedicated discussion. The elements single item, except for the last criterion referring to func-
of game design are different between computer and mobile tional impairment, which was operationalized using two
gaming platforms. Gaming is platform specific; online items. Participants were asked to rate items on a 3-point
gaming is usually associated with PC games and requires a Likert scale (0, never; 1, sometimes; and 2, often). The items
higher investment of time and effort relative to mobile of the IGDT-10 were recoded into a yes (1) or no (0) format
games. The psychosocial effect of mobile games may be to resemble the dichotomous structure of IGD. The Cron-
different than that of nonmobile games. Related behaviors bach’s alpha value of the scale was 0.68. The diagnosis of
and gaming contexts may also differ,23 which implies that IGD in the present sample was estimated using IGDT-10
PMG may be a different concept than IGD. To our knowl- with a cutoff threshold of 5.20
edge, no questionnaire examines PMG. Among the scales for
smartphone addiction, only the Smartphone Addiction In- Proposed diagnostic criteria for PMG
ventory (SPAI) has a cutoff value that was determined by
The proposed criteria for PMG were modified from the
clinical interviews with diagnostic criteria.24 This was also
criteria of smartphone addiction25 and consisted of the fol-
the only diagnostic criterion of smartphone addiction for
lowing: (A) six symptom criteria; (B) four functional im-
clinical interviews.25 It is not feasible to conduct diagnostic
pairment criteria; and (C) one exclusion criterion. Participants
interviews for smartphone addiction in a large-scale epide-
who had three (or more) symptoms from criterion group A and
miological study. It would be optimal to conduct a one-stage
two (or more) functional impairment criteria, as assessed in
investigation using a brief self-report instrument with high
the structured interview, were defined as having PMG.25
diagnostic accuracy. The specific aims of this study were as
follows: (a) to develop the Problematic Mobile Gaming
Relevant features of PMG
Questionnaire (PMGQ) and its short form; (b) to establish a
cutoff value for the PMGQ to assess mobile gaming addic- To assess the relevant features of mobile gaming, all
tion based on diagnostic criteria obtained by psychiatric in- participants reported (a) if they ever spent money on mobile
664 PAN ET AL.
games; (b) the average time (hours per week) spent on In- loading; these factors are not listed in Table 1. We selected
ternet gaming during weekdays and weekend days; (c) body– items with highest factor loading in each factor and main-
mass index (BMI); and (d) whether they had myopia. tained the proportion of items in the original three-factor
model of the PMGQ to construct a short form of the PMGQ,
Statistical analysis termed PMGQ-SF. The total scores of the PMGQ and
PMGQ-SF ranged from 12 to 48 and 4 to 16, respectively.
The construct validity of the PMGQ was examined by
The PMGQ and PMGQ-SF demonstrated good internal
exploratory factor analysis (EFA) using the maximum like-
consistency (Cronbach’s a = 0.92 and 0.79, respectively).
lihood estimation method and oblique promax rotation. EFA
was conducted using the sample of senior high students
(n = 1,455, which was regarded as an appropriate sample Determination of cutoff point
size). A receiver operating characteristic (ROC) analysis was A structured interview based on proposed diagnostic cri-
conducted to examine the diagnostic ability of the PMGQ. teria for PMG was used as the gold standard for diagnosis.
The area under the ROC curve (AUC) was used to measure Table 2 shows the ROC analysis for the PMGQ. The AUC
the diagnostic efficacy of the PMGQ. The diagnostic accu- was 0.802, indicating the high diagnostic value of the
racy indicated the percentage of all correct decisions. PMGQ. A cutoff point of 29/30 was optimal for discrimi-
Downloaded by NATIONAL TAIWAN UNIV. HOSPITAL from www.liebertpub.com at 10/17/19. For personal use only.
The cutoff point for the PMGQ was optimized for diag- nating cases of PMG from diagnostic negatives. Table 3
nosis when the score was accompanied by the highest shows the ROC analysis for the PMGQ-SF. The AUC was
Youden index and optimal diagnostic accuracy. Participants 0.736, indicating that the PMGQ-SF had acceptable diag-
were further classified into a PMG group and non-PMG nostic efficiency. A cutoff point of 9/10 was best for dis-
group according to the cutoff point of the PMGQ. Demo- criminating cases of PMG from diagnostic negatives.
graphic data and relevant characteristics of mobile gaming Furthermore, we conducted an exploratory analysis
were further compared between these two groups by using (n = 6,996) to determine the cutoff point of the PMGQ based
the chi-squared test or t test. A one-way analysis of variance on the diagnosis of IGD as the gold standard. The results
compared the group differences in terms of relevant features revealed that an AUC of 0.854 and a cutoff point of 29/30
of PMG among students with three levels of education. were still optimal for distinguishing between cases of IGD
and diagnostic negatives.
Results
Factor analysis of the PMGQ and development Demographic data and relevant characteristics of PMG
of the short form PMGQ
All participants were divided into three groups as follows:
Table 1 shows the PMGQ and results of the EFA. Results (a) fourth to sixth grade, (b) junior high school, and (c) senior
revealed that the PMGQ contained three potential factors, high school. Each group was further classified into a PMG
namely compulsion, tolerance, and withdrawal. We deleted group and a non-PMG group with a cutoff point of 29/30 in
four items because of apparent cross-loading and low factor the PMGQ. Overall, the proportion of people with PMG was
Table 1. Factor Loading of the Problematic Mobile Gaming Questionnaire
Factor 1, Factor 2, Factor 3,
compulsion tolerance withdrawal
(1) I have often experienced dry/sore eyes, muscle aches, or other physical 0.751 -0.092 0.060
discomforts from playing mobile games for a long duration.a
(2) I often do not plan on playing mobile games, but cannot resist picking up 0.692 0.052 -0.012
my phone to play (swipe).a
(3) I have a habit of playing mobile games before going to sleep and therefore 0.689 -0.059 0.077
reducing sleep time or sleep quality.
(4) Mobile games have had a negative impact on my academic performance or 0.670 0.034 0.084
occupation.
(5) Despite the negative consequence of mobile games, I have not decreased 0.600 0.171 -0.062
playing mobile games.
(6) I have attempted to reduce my mobile gaming, but did not succeed. 0.580 0.173 0.007
(7) Compared to 3 months ago, I am averaging more time every week playing -0.074 0.852 0.084
mobile games.a
(8) I find that I am spending more time playing mobile games. 0.082 0.850 -0.047
(9) I have been told multiple times that I am spending more time playing 0.117 0.530 0.150
mobile games.
(10) If I cannot play a mobile game, I feel restless and irritable.a 0.014 -0.002 0.905
(11) If I have a period of not playing mobile games, I start to feel -0.004 0.100 0.814
uncomfortable.
(12) I had an experience where I was playing a mobile game while 0.216 0.025 0.427
simultaneously walking, crossing the street, riding a motorcycle, or
driving and almost had a dangerous accident.
Bold indicates values most closely related to their given factors.
a
The items of the Problematic Mobile Gaming Questionnaire-Short Form.
ASSESSMENT OF PROBLEMATIC MOBILE GAMING 665
Table 2. Sensitivity, Specificity, Positive cation levels. A significant difference in BMI was noted
Predictive Rate, Negative Predictive Rate, between the PMG and non-PMG groups among elementary
Diagnostic Accuracy, and Youden Index school and junior high students, but not senior high students.
of Cutoff Points in the Problematic Mobile The proportion of children with myopia or severe myopia
Gaming Questionnaire Between Diagnostic was not significantly different between the PMG and non-
Positive and Negative Groups (N = 98)
PMG groups.
Youden Regarding students who played both PC-based and
Cutoff Sensitivity Specificity PPR NPR DA index mobile-based online games (n = 6,996), Figure 1 shows the
point (%) (%) (%) (%) (%) (%) proportion of those with positive diagnoses of IGD and
PMG. The results demonstrated that 77.8 percent of students
26 88.2 43.8 45.5 87.5 59.2 32.0 with possible IGD also met the PMG criteria (determined
27 82.4 50.0 46.7 84.2 61.2 32.4 using the PMGQ). However, only 12.6 percent of students
28 76.5 59.4 50.0 82.6 65.3 35.8
who met the PMG criteria also met the IGD criteria. Our data
29 70.6 68.8 54.5 81.5 69.4 39.3
30 67.6 82.8 67.6 82.8 77.6 50.5 showed that students with IGD also have high probability of
31 47.1 89.1 69.6 76.0 74.5 36.1 having a diagnosis of PMG, but not vice versa.
Downloaded by NATIONAL TAIWAN UNIV. HOSPITAL from www.liebertpub.com at 10/17/19. For personal use only.
Table 4. Comparison Between Problematic Mobile Gaming and Nonproblematic Mobile Gaming Groups According to the Cutoff Point
of 29/30 in the Problematic Mobile Gaming Questionnaire
Fourth to sixth grade Junior high school Senior high school
(N = 4,948) (N = 4,372) (N = 1,455)
PMG Non-PMG PMG Non-PMG PMG Non-PMG
(N = 946) (N = 4,002) (N = 898) (N = 3,474) (N = 277) (N = 1,178)
n % n % v2 P n % n % v2 p n % n % v2 p
Gender 64.366 <0.001 1.165 0.280 13.693 <0.001
Male 613 64.8 2,014 50.3 539 60.0 2,016 58.0 186 67.1 647 54.9
666
Female 333 35.2 1,988 49.7 359 40.0 1,458 42.0 91 32.9 531 45.1
Spend money on mobile 202 21.4 345 8.6 126.147 <0.001 326 36.3 628 18.1 138.943 <0.001 117 42.2 364 30.9 13.028 <0.001
gaming
Myopia 336 35.5 1,443 36.1 0.097 0.756 449 50.0 1,786 51.4 0.568 0.451 177 63.9 769 65.3 0.188 0.664
M SD M SD t P M SD M SD t p M SD M SD t p
Time spent on Internet gaming 29.39 15.12 20.53 11.15 16.96 <0.001 33.13 14.43 23.12 11.16 19.34 <0.001 30.94 14.00 23.13 10.46 8.72 <0.001
(hours per week)
BMI 19.78 5.38 19.00 4.87 4.12 <0.001 20.39 4.76 20.03 4.31 2.09 0.037 21.35 4.90 21.21 4.35 0.44 0.658
PMGQ 35.24 5.22 19.39 5.59 82.80 <0.001 34.77 4.57 20.10 5.50 82.11 <0.001 34.47 4.30 20.85 5.48 44.83 <0.001
IGDT-10 2.06 2.62 0.45 1.04 15.40 <0.001 1.49 2.34 0.36 0.88 11.97 <0.001 0.98 1.97 0.37 0.89 4.35 <0.001
BMI, body–mass index; IGDT-10, Ten-Item Internet Gaming Disorder Test; PMG, problematic mobile gaming; PMGQ, Problematic Mobile Gaming Questionnaire; SD, standard deviation.
ASSESSMENT OF PROBLEMATIC MOBILE GAMING 667
emerging addictive behavior, the diagnostic criteria of PMG rienced drivers. New England Journal of Medicine 2014;
should be confirmed by international consensus. The issue of 370:54–59.
whether the criteria of IGD should be modified for PMG 10. Gilman L, Cage DN, Horn A, et al. Tendon rupture asso-
warrants further research. Finally, the assessment of PMG ciated with excessive smartphone gaming. JAMA Internal
was determined solely on the basis of participants’ self- Medicine 2015; 175:1048–1049.
report questionnaires and responses to interviews. Supple- 11. Das DA, Grimmer KA, Sparnon AL, et al. The efficacy of
mentary information or app-recorded data may assist in playing a virtual reality game in modulating pain for chil-
confirmation of these symptoms27,29 and correct underesti- dren with acute burn injuries: a randomized controlled trial
mation of time spent due to the time distortion effect.30 [ISRCTN87413556]. BMC Pediatrics 2005; 5:1.
In conclusion, the PMGQ identifies three constructs of 12. Haik J, Tessone A, Nota A, et al. The use of video capture
virtual reality in burn rehabilitation: the possibilities.
behavioral addiction and the characteristics of PMG. The
Journal of Burn Care & Research: Official Publication of
cutoff point determined by psychiatrists’ diagnostic inter-
the American Burn Association 2006; 27:195–197.
views will be useful for conducting clinical screening and 13. Liu C-H, Lin S-H, Pan Y-C, et al. Smartphone gaming and
investigating the prevalence of PMG in children and ado- frequent use pattern associated with smartphone addiction.
lescents. Future studies should explore cross-cultural char- Medicine 2016; 95:e4068.
acteristics and the links to physical, social, and mental health
Downloaded by NATIONAL TAIWAN UNIV. HOSPITAL from www.liebertpub.com at 10/17/19. For personal use only.
27. Lin YH, Lin SH, Yang CCH, et al. (2017) Psychopathology 35. Wong FY. Influence of Pokémon Go on physical activity
of everyday life in the 21st century: smartphone addiction. In levels of university players: a cross-sectional study. Inter-
Montag C, Reuter M, eds. Internet addiction: neuroscientific national Journal of Health Geographics 2017; 16:8.
approaches and therapeutical implications including smart- 36. Althoff T, White RW, Horvitz E. Influence of Pokemon Go
phone addiction. Cham: Springer International Publishing, on physical activity: study and implications. Journal of
pp. 339–358. Medical Internet Research 2016; 18:e315.
28. Kwon M, Kim DJ, Cho H, et al. The Smartphone Addiction 37. Annie A. (2016) App Annie—the app analytics and app
Scale: development and validation of a short version for data industry standard. www.appannie.com/cn (accessed
adolescents. PLoS One 2013; 8:e83558. July 31, 2017).
29. Lin YH, Lin YC, Lin SH, et al. To use or not to use? 38. Alim-Marvasti A, Bi W, Mahroo OA, et al. Transient
Compulsive behavior and its role in smartphone addiction. smartphone ‘‘blindness.’’ New England Journal of Medi-
Translational Psychiatry 2017; 7:e1030. cine 2016; 374:2502–2504.
30. Lin YH, Lin YC, Lee YH, et al. Time distortion associated 39. Irshad F, Adhiyaman V. Transient smartphone blindness.
with smartphone addiction: identifying smartphone addic- Canadian Journal of Ophthalmology Journal Canadien
tion via a mobile application (App). Journal of Psychiatric D’Ophtalmologie 2017; 52:e107–e108.
Research 2015; 65:139–145. 40. Sathiamoorthi S, Wingerchuk DM. Transient smartphone
Downloaded by NATIONAL TAIWAN UNIV. HOSPITAL from www.liebertpub.com at 10/17/19. For personal use only.
31. Kaptsis D, King DL, Delfabbro PH, et al. Withdrawal blindness: relevance to misdiagnosis in neurologic practice.
symptoms in internet gaming disorder: a systematic review. Neurology 2017; 88:809–810.
Clinical Psychological Review 2016; 43:58–66.
32. Yen CF, Hsiao RC, Ko CH, et al. The relationships be-
tween body mass index and television viewing, internet use
and cellular phone use: the moderating effects of socio-
demographic characteristics and exercise. The International Address correspondence to:
Journal of Eating Disorders 2010;43:565–571. Dr. Yu-Hsuan Lin
33. Howe KB, Suharlim C, Ueda P, et al. Gotta catch’em all! Institute of Population Health Sciences
Pokemon GO and physical activity among young adults: National Health Research Institutes
difference in differences study. BMJ (Clinical Research Ed) 35 Keyan Road, Zhunan
2016; 355:i6270. Miaoli County 35053
34. Xian Y, Xu H, Xu H, et al. An initial evaluation of the Taiwan
impact of Pokemon GO on physical activity. Journal of the
American Heart Association 2017; 6:e005341. E-mail: yuhsuanlin@nhri.edu.tw