Skip to main content

Depression, Internet Gaming Disorder, and the Moderating Effect of the Gamer-Avatar Relationship: an Exploratory Longitudinal Study


Research into Internet gaming disorder (IGD) literature largely uses cross-sectional designs and seldom examines gaming context-related factors. Therefore, the present study combined a cross-sectional and longitudinal design to examine depression and the gamer-avatar relationship (GAR) as risk factors in the development of IGD among emerging adults. IGD behaviors of 125 gamers (64 online gamers, Mage = 23.3 years, SD = 3.4; 61 offline gamers, Mage = 23.0 years, SD = 3.4) were assessed using the nine-item Internet Gaming Disorder Scale Short Form (IGDS-SF9; Pontes and Griffiths Revista Argentina de Ciencias del Comportamiento, 7, 102–118, 2015a; Computers in Human Behavior, 45, 137–143, 2015b). The Self-Presence Scale (Ratan and Dawson Communication Research, 2015) and the Beck Depression Inventory (Beck et al. 1996) were also used to assess gamers’ levels of GAR and depressive symptoms, respectively. Regression and moderation analyses revealed that depression and the GAR act as individual risk factors in the development of IGD over time. Furthermore, the GAR exacerbates the IGD risk effect of depression.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2


  1. The current study is part of a wider project of Federation University Australia that addresses the interplay between individual, Internet, and proximal context factors in the development of Internet gaming disorder symptoms among emerging adults. Instruments used in the data include the following: (1) Internet Gaming Disorder 9-Short Form (Pontes and Griffiths 2015a, b), (2) Beck Depression Inventory—second edition (21 items) (Beck et al. 1996), (3) Beck Anxiety Inventory (21 items) (Beck and Steer 1990), (4) Hikikomori-Social Withdrawal Scale (5 items) (Teo et al. 2015), (5) Attention Deficit Hyperactivity Self-Report Scale (18 items) (Kessler et al. 2005), (6) Ten Item Personality Inventory (Gosling et al. 2003), (7) The Balanced Family Cohesion Scale (7 items) (Olson 2000), (8) Presence Questionnaire (10 items) (Faiola et al. 2013), (9) Online Flow Questionnaire (5 items) (Chen et al. 2000), (10) Self-Presence Questionnaire (Ratan and Hasler 2010), (11) The Gaming-Contingent Self-Worth Scale (12 items) (Beard and Wickham 2016), and (12) Demographic and Internet Use Questions. The battery of questionnaires was utilized for both online and face-to-face data collection. The use of the fitness tracker (Fitbit flex) was used only for face-to-face data collection. Data have not been used in any previous published studies.

  2. To ensure that there were no significant differences between the online and face-to-face samples considering their demographic and Internet use characteristics, as well as the variables used in the present study, independent sample t-tests and chi-square analyses were conducted. Findings did not indicate any significant differences in regards to gender (χ 2 = 0.21, df = 1, p = .89), the type of game genre (i.e., MMOs vs MMORPGs) played (χ 2 = 2.59, df = 1, p = .61), the age of the participants (t = − 0.54, df = 120, p = .59), their years of Internet use (t = 2.35, df = 122, p = .06), their reported level of avatar self-presence (t = 1.09, df = 119, p = .28), and their assessed psychopathological symptoms (depression t = − 0.38, df = 111, p = .70; IGD t = − 0.14, df = 111, p = .89). Therefore, online and face-to-face data (i.e., TP1) were combined (i.e., analyzed together) to investigate cross-sectional questions.

  3. The longitudinal design was assessed for attrition. Assessments’ frequency for each participant varied within a range of 1–3 (Maverage = 2.57). Time point 1 comprised 61 participants, time point 2 comprised 56 participants (8.20% attrition), and time point 3 comprised 43 participants (29.51% attrition). In line with literature recommendations, attrition, in relation to the studied variables, was assessed using Little’s Missing Completely At Random test (MCAR), which was insignificant (MCAR, χ 2 = 1715.79, p = 1.00; Little and Rubin 2014). In order to avoid list-wise deletion, which would reduce the sample’s power, maximum likelihood imputation (five times) of values was applied (Gold and Bentler 2000).

  4. In line with the approval received by the ethics committee of Federation University, the flyers: a) indicated that participants were required to participate on three separate measurement occasions approximately one month apart; b) included an email address to contact the investigators; and c) clearly described the process and stages of the data collection (face-to-face and online). MMO and MMORPG players, aged between 18-29 years old, interested in the study received the Plain Language Information Statement (PLIS). The PLIS clearly indicated that participation was voluntary and that participants could independently decide to withdraw from the study at any point. Individuals who choose to participate were required to provide informed consent.


Download references


There was no funding.

Author information

Authors and Affiliations



TB contributed to the literature review, hypothesis formulation, data collection and analyses, and the structure and sequence of theoretical arguments.

VS contributed to the literature review, hypothesis formulation, data collection and analyses, and the structure and sequence of theoretical arguments.

LL contributed to the data collection and analyses.

BA contributed to the data collection and analyses.

MG contributed to the theoretical consolidation of the current work, and revised and edited the final manuscript.

Corresponding author

Correspondence to Tyrone L. Burleigh.

Ethics declarations

Conflict of Interest

The authors declare that they have no conflict of interest.

Ethical Standards—Animal Rights

All procedures performed in the study involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. This article does not contain any studies with animals performed by any of the authors.

Informed Consent

Informed consent was obtained from all individual participants included in the study.



A1. Analytical assumptions

The presence of outliers in the data was assessed by calculating the Mahalanobis distance (MD). In regard to the cross-sectional analysis, three outlier values were detected with a MD = 21.47, p = .00001; MD = 13.42, p = .00025; and MD = 11.57, p = .00067, respectively. In regard to the longitudinal analysis, two outlier values were detected with a MD = 17.44, p = .00003, and MD = 14.90, p = .00011, respectively. Following literature recommendations, any MD that referred to a probability lower than 0.001 was defined as an outlier, was treated as a missing value, and was substituted with maximum likelihood (five times) imputation based on all the available variables (Lee and Xia 2006).

Assumptions of linearity, multivariate normality, and homoscedasticity were further assessed (Rao and Toutenburg 1995). Given that only one independent variable was used, multicollinearity was not assessed. Linearity assumption was tested with probability-probability plot (PP). All variables used in both cross-sectional and longitudinal data did not violate the assumption of linearity. Normality was checked by measuring skewness and kurtosis with values below − 2 or above + 2 addressed with bootstrapping-1000 resamples (Berkovits et al. 2000). In order to assess homoscedasticity, scatterplots of regression standardized residuals, and the regression standardized predicted values were implemented (George and Mallery 2010). Both cross-sectional and longitudinal data scatterplots did not indicate any violations to the assumption of homoscedasticity.

A2. Correlations and descriptive statistics

Descriptive statistics and bivariate correlations of cross-sectional data can be found in Table 2. The results revealed that higher depression scores and GAR scores are correlated with higher IGD scores.

Descriptive statistics and bivariate correlations, from longitudinal data, can be found in Table 3. The results indicate that higher depression scores are correlated with higher IGD scores. Both depression scores and IGD scores were significantly related to each other on all time points. In addition to this, both GAR scores and IGD scores are also significantly related across all time points.

Table 2 Descriptive statistics and correlations among study variables (cross-sectional data)
Table 3 Descriptive statistics and correlations among study variables (longitudinal data)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Burleigh, T.L., Stavropoulos, V., Liew, L.W.L. et al. Depression, Internet Gaming Disorder, and the Moderating Effect of the Gamer-Avatar Relationship: an Exploratory Longitudinal Study. Int J Ment Health Addiction 16, 102–124 (2018).

Download citation

  • Published:

  • Issue Date:

  • DOI:


  • Internet gaming disorder
  • Video gaming
  • Gamer-avatar relationship
  • Online addiction
  • Depression