Research on the Effects of Media

© 2008-2011 Douglas A. Gentile; All Rights Reserved

Douglas A. Gentile, Ph.D.
Research Article
Video game playing, attention problems, and impulsiveness:  Evidence of bi-directional causality
Douglas A. Gentile, Edward L. Swing, Choon Guan Lim, and Angeline Khoo

Click here to view a pdf

How to cite: Gentile, D. A., Swing, E. L., Lim, C. G., & Khoo, A. (2012).  Video game playing, attention problems, and impulsiveness: Evidence of bi-directional causality.  Psychology of Popular Media Culture, 1, 62-70.


The present study examines video game playing as it relates to attention problems and
impulsiveness in a sample of 3,034 children and adolescents from Singapore measured
over 3 years. Consistent with previous research, those who spend more time playing
video games subsequently have more attention problems, even when earlier attention
problems, sex, age, race, and socioeconomic status are statistically controlled. Violent
content may have a unique effect on attention problems and impulsiveness, but total
time spent with video games appears to be a more consistent predictor. Individuals who
are more impulsive or have more attention problems subsequently spend more time
playing video games, even when initial video game playing is statistically controlled,
suggesting bidirectional causality between video game playing and attention problems/

Note: The version presented on this website may differ in small ways from the final published version.


Problems associated with attention disorders such as Attention-Deficit/Hyperactivity Disorder (ADHD) impair a variety of functions, particularly school performance (Barry, Lyman, & Klinger, 2002).  Attention disorders are substantially biologically based but have environmental risk factors as well (Biederman et al., 2008).  Some recent evidence suggests that exposure to screen media may increase attention problems (e.g., Christakis, Zimmerman, DiGiuseppe, & McCarty, 2004; Landhuis, Poulton, Welch, & Hancox, 2007; Swing, Gentile, Anderson, & Walsh, 2010).  Most of the research to date has focused on television as a potential contributor to attention problems (e.g., Acevedo-Polakovich, Lorch, & Milich, 2007; Christakis et al., 2004; Johnson, Cohen, Kasen, & Brook, 2007; Landhuis et al., 2007; Mistry, Minkovitz, Strobino, & Borzekowski, 2007; Zimmerman & Christakis, 2007).  Research examining video games has found similar associations with attention problems, though more research examining video games would be useful (Bioulac, Arfi, & Bouvard; 2008; Chan & Rabinowitz, 2006, Swing et al., 2010).  A few studies have found mixed results (Ferguson, 2010) or no evidence of media effects (Obel et al., 2004; Stevens & Mulsow, 2006) on attention problems.  However, these studies either also found significant bivariate correlations between electronic media and attention problems or did not report such analyses.

There are at least four possible explanations for the association between electronic media and greater attention problems.  

Excitement hypothesis. Electronic screen media may make other activities (e.g., work or school) seem less interesting by comparison.  Many television shows and video games are very exciting, fun, and include potent attention grabbing cues (e.g., violence). Indeed, most shows and video games (especially violent ones) make liberal use of features that trigger an orienting response, such as edits, sound effects, flickering light levels, etc. (Kubey & Csikszentmihalyi, 2002).  These salient features provide a type of continual support for attention.  This is quite different from many of the work and school tasks that are difficult for those with attention problems. Over time, frequently engaging in exciting activities (e.g., playing video games) might change a child's expectations regarding the desired level of stimulation. The greater the contrast between electronic media content and work or school tasks, the more difficult it could become to focus on work or school.  

Displacement hypothesis. Second, time spent with television or video games might simply displace time that would have otherwise been spent on other activities that would have allowed for greater development of impulse control. These two explanations need not be mutually exclusive. Both are consistent with the Strength Model of Self-Control (Baumeister, Vohs, & Tice, 2007). Specifically, to the extent that electronic media use does not tax self-control resources, time spent with such media may weaken ones' ability to exert self-control. The excitement and displacement hypotheses would be consistent with different associations between electronic media variables and attention problems. If attention problems are simply the result of the displacement of self-control building activities, total time spent with electronic media should predict greater attention problems but the content of that media should not make a difference.  If the contrast between exciting television or video games and work or school tasks is important, then differences in content (e.g., greater violence) should predict greater attention problems in addition to or instead of total media exposure.  There have been few tests of this possibility. Zimmerman and Christakis (2007) found violent television to be most strongly related to attention problems (followed by non-violent television and then educational television), however the difference between violent and non-violent television content was not statistically significant.

Attraction hypothesis. A third possibility is that individuals who have attention problems are more attracted to electronic media.  This explanation need not be mutually exclusive with the first two explanations of a causal effect of electronic media on attention problems.  In fact, it is consistent with the Strength Model of Self-Control that exciting electronic media that do not require self-control would be frequently used by those with lower ability to exert self-control (Baumeister et al., 2007). Those with lower self-control may find the appeal of exciting electronic media too difficult to resist. To date, this hypothesis has not been tested using longitudinal data.

Third variable hypothesis. A fourth possibility is that the observed association between electronic media and attention problems is spurious. A third variable such as sex may explain this association.  Thus far, evidence for third variable explanations is weak.  For example, several studies (e.g., Christakis et al., 2004, Swing et al., 2010) included sex, age, and other individual difference variables as covariates and still found unique associations between electronic media and attention problems.  Ferguson (2010) reported a regression model in which the electronic media effects are non-significant, but this model included four highly correlated electronic media variables entered separately (reducing the variance of each media predictor by approximately 25-50% due to covariates that are not alternative explanations). The model also included 12 other covariates, some of which do not seem to be plausible alternative causal explanations (e.g., antisocial personality).  No model was presented involving only the theoretically relevant covariates, leaving a very strong possibility that the inclusion of improper covariates and an overly conservative model caused results to become non-significant.  Thus, the case for any one of these covariates as alternative explanations remains weak but is nonetheless a valid hypothesis. It is therefore valuable to continue to test several variables (e.g., sex, age, race, and socioeconomic status [SES]) as third variable explanations. For example, boys typically spend more time playing video games than girls (especially violent games) and are also more frequently diagnosed with attention disorders. The link between video games and attention disorders could thus be a spurious effect of these two potentially unrelated facts. Likewise, some other variable such as race or SES might predict both parental permissiveness regarding video game use as well as predisposing other adverse home conditions that are the environmental causes of attention problems. In this case, high levels of video game playing would be just one more consequence of the ineffective parenting that is truly causing attention problems. Another possibility is that both increasing video game use and increasing attention problems are normative changes with age. The inclusion of sex, age, race, and SES as covariates allows these potential alternative explanations to be tested and possibly eliminated.

In order to clarify the support for each of these explanations, we collected longitudinal data from a sample of children and adolescents.  Participants reported their amount of video game playing, violent video game exposure, and completed measures of attention problems and impulsiveness.  Several individual difference covariates were also measured, allowing for tests of some potential third variable explanations.  To the extent that overall video game playing is associated with attention problems and impulsiveness, the displacement hypothesis would be supported. If violent video game exposure is uniquely associated with attention problems and impulsiveness (beyond simply the amount of video game playing), this would support the excitement hypothesis. If attention problems and impulsiveness predict increased subsequent video game playing, even when earlier video game playing is statistically controlled, this would support the attraction hypothesis. If controlling for a demographic covariate reduces the video game and attention problems/impulsiveness link to zero, this would support the third variable hypothesis, particularly if this remains true with only the relevant covariate in the model (without the other covariates and only one media variable).  

This study included 3034 children/adolescents from 12 different schools in Singapore with a 99% response rate.  Participants were ages 8-17 (M = 11.2, SD = 2.1) at the first wave of data collection, starting in grades 3, 4, 7, and 8. Children completed questionnaire measures in their classrooms in three waves (W1, W2, & W3), each one year apart. Questionnaire data were available for 3034, 2360, and 2232 participants for W1, W2, and W3, respectively.  Parent consent and child assent were gathered.  Data were collected by classroom teachers and school research coordinators, with direction from trained research personnel.  The questionnaires had been pretested with hundreds of children in three schools (not included in the present sample) to ensure comprehensibility.  

Video Games
At each wave, participants indicated how many hours they played video games during each of three time periods (morning, afternoon, and evening) on a typical school day and on a typical weekend, from which we calculated the average weekly video game playing time. Video game playing showed excellent internal reliability across time periods and days (alphas of .90, .88 and .87 at waves 1, 2, and 3, respectively). Average weekly video game playing also showed strong test-retest correlations (r = .36 and r = .46 from wave 1 to 2 and wave 2 to 3, respectively). This suggests that this measure also has adequate test-retest reliability, although it is certainly likely that amount of gaming could change across time. Participants also listed the three video games that they play the most at each wave and indicated how often they killed creatures in each game on a four point scale (never, seldom, often, almost always) as well as how often they killed players in the game on the same scale.  Violent video game exposure was computed based on the average amount of killing creatures and players in each of the three games. Violent video game exposure also showed adequate internal reliability (alphas of .77, .75, and .76 at waves 1, 2, and 3, respectively). Video game violence exposure scores also showed strong test-retest correlations (r = .38 and r = .46 from waves 1 to 2 and waves 2 to 3, respectively) indicating adequate test-retest reliability, although again there is no need for children to be consistent in their violent game play across years.

Attention Problems
Participants completed the Current ADHD Symptoms Scale Self-Report, an 18 item measure of inattention and hyperactivity symptoms, at W2 and W3 only (University of Massachusetts Medical School, 2011).  This measure requires participants to indicate how often they exhibit symptoms such as “Fail to give close attention to details or make careless mistakes in my work” or “Blurt out answers before questions have been completed”.  Each question is answered on a four point scale (never or rarely, sometimes, often, or very often). Scores on the Current ADHD Symptoms Scale Self-Report showed excellent inter-item reliability in the present sample (alphas of .92 and .93 at waves 2 and 3, respectively). Scores on this scale also showed considerable stability across waves (r = .47 from waves 2 to 3) supporting the test-retest reliability of this scale.

 Participants also completed  14 items from the Barratt Impulsiveness Scale-11, a measure of impulsiveness, at W1, W2, and W3 (Patton, Stanford, & Barratt, 1995).  This included items “I often make things worse because I act without thinking” and “I concentrate easily” (reversed).  Questions were answered on a four point scale (strongly disagree, disagree, agree, or strongly agree). These items showed adequate inter-item reliability in the present sample (alphas of .62, .73, and .65 at waves 1, 2, and 3, respectively). Impulsiveness scores showed strong test-retest correlations (r = .45 and r = .49 from waves 1 to 2 and waves 2 to 3, respectively), indicating adequate test-retest reliability. ADHD symptom scores were strongly correlated with impulsiveness scores (r =.48 and r = .47 at waves 2 and 3, respectively) providing evidence of the convergent validity of each measure.
 School performance, as measured by self-reported scores from the most recent exam in four different school subjects (English, math, science, and second language) also served as a useful outcome for establishing the predictive validity of the measures of attention problems and impulsiveness, given that previous research has shown individuals with ADHD to underperform academically (Barry et al., 2002). Attention problem scores were associated with poorer performance on the recent exams (r = -.24 and r = -.26 at waves 2 and 3, respectively). Impulsiveness scores were also associated with poorer exam performance (r = -.19, r = -.18, and r = -.17 at waves 1, 2, and 3, respectively). These small to moderate negative correlations demonstrate the predictive validity of the attention problems and impulsiveness scores.

Participants reported sex, age, and race (coded as majority vs. minority).  Participants reported the educational achievement of their mother and father as well as the type of home they lived in. Housing type, which is classified by the size of residence (eg, 1-2 room public housing, 3 room public housing, etc), is a standard demographic characteristic in studies on Singaporean youth  as a proxy indicator of socioeconomic status in the Singaporean context (Ho & Yip, 2002).  Mother's and father's educational achievement were each standardized and combined to compute parental education.  Mother's and father's educational achievement were strongly correlated with each other (r = .65) and also with housing type (r = .29 and r = .33 for mothers and fathers, respectively), indicating that these measures have good internal reliability. Socioeconomic status (SES) was computed based on the parental education (standardized) and the rank order of their housing type (standardized).  


Bivariate correlations were computed for video game exposure, video game violence, impulsiveness, attention problems, sex, age, race, and SES at all applicable waves (see Table 1).

The weighted averages of the bivariate correlations of video game exposure and video game violence with impulsiveness and attention problems from all relevant waves are reported in Table 2.  These bivariate correlations (from r = 0.14 to r = 0.22) are in the small to moderate range (as would be predicted for an environmental risk of attention problems).

In order to compare total video game playing and video game violence as predictors of attention problems and impulsiveness, four general linear models were computed (see Tables 3 and 4).  Models 1 and 2 compare total video game exposure (VGE) and video game violence (VGV) as predictors of attention problems with sex, age, race, and SES included as covariates. Model 1 uses the average of W2 and W3 for VGE, VGV, and attention problems. Model 2 tests the time lagged video game effects by using the W2 measure of VGE and VGV, W2 attention problems as a covariate, and W3 attention problems as the outcome.

Models 3 and 4 are similar but impulsiveness is the outcome (W1, W2, and W3 averages in Model 3 and W1 impulsiveness as a covariate and W3 impulsiveness as the outcome in Model 4).  These models are a conservative test of video game effects on attention problems and impulsiveness.

 In Models 1 and 3, both VGE and VGV uniquely predict attention problems and impulsiveness, respectively, providing some support for the excitement and displacement hypotheses. However in Models 2 and 4, VGE (but not VGV) remains significant suggesting that total video game exposure is a more robust predictor of attention problems and impulsiveness than violent gaming.

A path analysis was computed using Amos 7 to test the time lagged effects of total VGE on attention problems as well as the effect of attention problems on total VGE (see Figure 1).  This model showed a good fit to the data, χ2(6) = 14.86, p = .021, normed fit index (NFI) = .989, comparative fit index (CFI) = .993, root means square error of approximation (RMSEA) = .029, 90% CI: .011, .049.  

A similar path analysis was computed to test the time lagged effects of total VGE on impulsiveness as well as the effect of impulsiveness on total VGE (see Figure 2).  This model also showed good fit to the data, χ2(12) = 33.67, p = .001, NFI = .969, CFI = .980, RMSEA = .031, 90% CI: .019, .044.

Figures 1 and 2 show a small effect of VGE on subsequent attention problems and impulsiveness even when sex, age, race, SES, and earlier attention problems or impulsiveness are controlled.

Figures 1 and 2 also show a small effect of attention problems and impulsiveness on subsequent VGE.  This indicates potentially bi-directional causality between video game playing and attention problems/impulsiveness. All four general linear models and both path models include at least one significant video game predictor, suggesting that the association of VGE with impulsiveness and attention problems cannot simply be explained by sex, age, race, or SES as third variables.


Consistent with most previous research, the present study found video game playing to be associated with greater subsequent attention problems, even when earlier attention problems were statistically controlled.  There was some evidence that violent video game content added uniquely to predicting attention problems beyond the total amount of time played.  Specifically, video game violence exposure was uniquely associated with attention problems and impulsiveness when sex, age, race, and SES were statistically controlled. However, total time spent playing video games was the more robust predictor in this sample, predicting attention problems and impulsiveness even when earlier attention problems/impulsiveness were statistically controlled as well, providing stronger support for the displacement hypothesis than for the excitement hypothesis.  Future research should continue to examine potential content effects. The current study is also the first to test the attraction hypothesis (i.e., individuals who are impulsive or have attention problems seek out video games).  The data were consistent with this hypothesis as well.

 These findings provide evidence for bi-directional causality: children with greater impulsiveness and attention problems spend more time playing video games, which in turn increases subsequent attention problems and impulsiveness.  This finding does not alter the cause for concern about the potential for video games to contribute to the development of attention problems.

The longitudinal design allowed the present study to provide stronger evidence of causality than a single time point study, but these data were nonetheless observational. Thus it remains possible that some third variable not included accounts for one or both directions of apparent causality.   However, several third variables have been ruled out thus far and to date there is no substantial evidence in favor of any particular third variable explanation.  Furthermore, controlling for earlier attention problems or impulsivity also controls for all of the variables that caused them in the first place, including all prior genetic and environmental factors. This further weakens the third variable hypothesis as an alternative explanation. The study would also have been strengthened if we had been able to include the ADHD symptoms scale at Wave 1 also, but were unable to due to time constraints in the classroom.

The effects observed in the present study are admittedly small in a statistical sense (β = .05), but several facts should be considered regarding these effects.  First, these effects come from conservative analyses that likely underestimate the true effect just as the bivariate correlations are likely to be overestimates. Some children with attention problems are also likely to underestimate their problems in self-reports, compared to the reports of adults such as parents and teachers (Owens, Goldfine, Evangelista, Hoza, & Kaiser, 2007). This would lead estimates of the link between video games and attention problems to be underestimated when based on child self-reports.  Future research might address these self-report issues by including parent reports or child time logs for assessing video game playing as well as parent and teacher reports of attention problems or impulsiveness. It should be noted, however, that small effects such as those obtained here can be important when they apply to a large population as is the case with video games.  In fact, even the most conservative estimates of electronic media effects on attention are similar in magnitude to specific genes, such as LPHN3, the markers for which increase the odds of inattention by approximately 1.23 yet are considered to have great practical importance (Arcos-Burgos et al., 2010).  Finally, the size of the effect is consistent with theoretical predictions, as environmental factors should explain only a small amount of variance in attention problems.

As with all nature-nurture questions, the answer ultimately is that both matter.  For the past 30 years, most of the research on attention problems has focused on biological and genetic factors rather than on environmental factors.  This allowed for rapid advances in drug therapies, but has also caused many researchers and members of the general public to assume that impulsivity and attention problems were not modifiable by experience.  This is unfortunate, as it means we have only focused on part of the solution.  Furthermore, many problems with genetic bases are clearly enhanced by environmental triggers.  By understanding some of the environmental influences, we can develop more effective solutions for children and parents.  More research is clearly needed on the environmental factors, especially factors that are easily modified by parents, such as screen time.


Acevedo-Polakovich, I. D., Lorch, E. P., & Milich, R. (2007). Comparing television use and reading in  children with ADHD and non-referred children across two age groups. Media Psychology, 9,  447-472.
Arcos-Burgos, M., Jain, M., Acosta, M. T., Shively, S., Stanescu, H., Wallis, D., ... Muenke, M.  (2010). A common variant of the latrophilin 3 gene, LPHN3, confers susceptibility to ADHD  and predicts effectiveness of stimulant medication. Molecular Psychiatry, 15, 1053-1066.
Barkley, R. A. (1997). Behavioral inhibition, sustained attention, and executive functions: Constructing  a unifying theory of ADHD. Psychological Bulletin, 121, 65-94.
Barry, T. D., Lyman, R. D., & Klinger, L. G. (2002). Academic underachievement and attention- deficit/hyperactivity disorder: The negative impact of symptom severity on school performance.  Journal of School Psychology, 40, 259-283.
Baumeister, R. F., Vohs, K. D., & Tice, D. M. (2007). The strength model of self-control. Current  Directions in Psychological Science, 16, 351-355.
Biederman, J., Petty, C. R., Wilens, T. E., Fraire, M. G., Purcell, C. A., Mick, E., Monuteaux, M. C., &  Faraone, S. V. (2008). Familial risk analyses of attention deficit hyperactivity disorder and  substance use disorders. The American Journal of Psychiatry, 165, 107–115.
Bioulac, S., Arfi, L., & Bouvard, M. P. (2008). Attention deficit/hyperactivity disorder and video  games: A comparative study of hyperactive and control children. European Psychiatry, 23,  134-141.
Bonett, D. G. (2007). Transforming odds ratios into correlations for meta-analytic research. American  Psychologist 62, 254-255.
Chan, P. A., & Rabinowitz, T. (2006). A cross-sectional analysis of video games and attention deficit  hyperactivity disorder symptoms in adolescents. Annals of General Psychiatry, 5(16).
Christakis, D. A., Zimmerman, F. J., DiGiuseppe, D. L., & McCarty, C. A. (2004). Early television  exposure and subsequent attentional problems in children. Pediatrics, 113, 708-713.
Green, C. S., & Bavelier, D. (2003). Action video game modifies visual selective attention. Nature,  423, 534-537.
Ho, K.C. & Yip, J. (2003). The State of Youth in Singapore. Singapore: National Youth Council.
Johnson, J. G., Cohen, P., Kasen, S., & Brook, J. S. (2007). Extensive television viewing and the  development of attention and learning difficulties during adolescence. Archives of Pediatrics  and Adolescent Medicine, 161, 480-486.
Kubey, R. & Csikszentmihalyi, M. (2002, February).  "Television Addiction Is No Mere Metaphor." Scientific American.
Landhuis, C. E., Poulton, R., Welch, D., & Hancox, R. J. (2007). Does childhood television viewing  lead to attention problems in adolescence?: Results from a prospective longitudinal study.  Pediatrics, 120, 532-537.
Mistry, K. B., Minkovitz, C. S., Strobino, D. M., & Borzekowski, D. L. (2007). Children's television  exposure and behavioral and social outcomes at 5.5 years: Does timing of exposure matter?  Pediatrics, 120, 762-769.
Murphy, K. R. & Davidshofer, C. O. (1988). Psychological testing. Engelwood Cliffs, NJ: Prentice  Hall.  
Obel, C., Henriksen, T. B., Dalsgaard, S., Linnet, K. M., Skajaa, E., Thomsen, P. H., & Olsen, J.  (2004). Does children's watching of television cause attention problems? Retesting the  hypothesis in a Danish cohort. Pediatrics, 114, 1373-1374.
Owens, J. S., Goldfine, M. E., Evangelista, N. M., Hoza, B., & Kaiser, N. M. (2007). A critical review  of self-perceptions and the positive illusory bias in children with ADHD. Clinical Child and  Family Psychology Review, 10, 335-351.
Patton, J. H., Stanford, M. S., & Barratt, E. S. (1995). Factor structure of the Barratt Impulsiveness
Scale. Journal of Clinical Psychology, 51, 768-774.
Stevens, T., & Mulsow, M. (2006). There is no meaningful relationship between television exposure  and symptoms of attention deficit/hyperactivity disorder. Pediatrics, 117, 665-672.
Swing, E. L., Gentile, D. A., Anderson, C. A., & Walsh, D. A. (2010). Television and video game  exposure and the development of attention problems. Pediatrics, 126, 214-221.
University of Massachusetts Medical School. (2011). ADHD Self-Report. Retrieved July 8, 2011  from:
Zimmerman, F. J., & Christakis, D. A. (2007). Associations between content types of early media  exposure and subsequent attentional problems. Pediatrics, 120, 986-992.