Journal of Adolescent Health 52 (2013) 613e619

www.jahonline.org

Original article

Older Versus Newer Media and the Well-being of United States Youth: ResultsFrom a National Longitudinal Panel

Daniel Romer, Ph.D. *, Zhanna Bagdasarov, Ph.D., and Eian More, M.A.Annenberg Public Policy Center, University of Pennsylvania, Philadelphia, Pennsylvania

Article history: Received July 10, 2012; Accepted November 8, 2012Keywords: Old media; New media; Academic performance; Well-being

A B S T R A C TIMPLICATIONS AND

Purpose: To determine the effects of both older and newer media use on academic, social, andmental health outcomes in adolescents and young adults.Methods: We interviewed a nationally representative panel of youth ages 14e24 years (N ¼ 719)twice 1 year apart to determine time spent with television, the Internet, videogames, and bookreading, as well as the purpose of those uses. A cluster analysis identified major combinations ofmedia use. Regression models tested hypotheses regarding changes in self-reported school grades,participation in clubs and sports, and symptoms of depression, as predicted by recent media useand differences in cluster membership.Results: Use of older media was related to grades, with television inversely and book readingpositively related to performance. Moderate use of the Internet was positively related to partici-pation in both sports and clubs. Although heavy use of the Internet and videogames was associatedwith an increase in depression, increased depression also predicted greater use of these media aswell as withdrawal from sports and clubs. Clusters that used media in moderation with anemphasis on information gathering were most associated with healthy outcomes.Conclusions: Despite concerns that excessive use of new media is harmful to adolescent devel-opment, the findings reinforce previous conclusions that television detracts from academicperformance and book reading supports it. Heavy use of the Internet and video gaming may bemore a symptom of mental health problems than a cause. Moderate use of the Internet, especiallyfor acquiring information, is most supportive of healthy development.

� 2013 Society for Adolescent Health and Medicine. All rights reserved.

* Address correspondence to: Daniel Romer, Ph.D., Annenberg Public PolicyCenter, Universityof Pennsylvania, 202 S. 36th Street, Philadelphia PA19104-3806.

E-mail address: dromer@asc.upenn.edu (D. Romer).

1054-139X/$ e see front matter � 2013 Society for Adolescent Health and Medicine. All rights reserved.http://dx.doi.org/10.1016/j.jadohealth.2012.11.012

CONTRIBUTION

Despite concerns aboutthe ill effects of heavyscreenmedia use onyouth,the present research indi-cates that not all uses ofscreen media are harmful,and practitioners andresearchers should adopta more nuanced approachto both the harmful andbeneficial uses of massmedia by young people.

The development of the Internet and other digital technolo-gies during the past 2 decades has altered how people, especiallyadolescents, usemedia [1,2]. At the same time, widespread use ofthese new technologies has fueled concerns that thesemedia canhave harmful effects on young people’s academic pursuits [3] andoverall well-being [4e6]. Although considerable research hasexamined time spent with television in relation to academic andother outcomes in children and adolescents [7e11], we know

much less about how the use of newer media affects suchoutcomes.

Previous research that has examined the use of newer media,such as videogames [9,12] and the Internet [4,13], has importantlimitations. First, with some exceptions [14], it has largely beencross-sectional and thus precludes assessment of whether mediause is associated with change in outcomes or is merely a conse-quence of those outcomes. Second, research has either focusedon only one or two forms of media use at a time [e.g., 5,12,13] orhas used global measures of screen based media, lumping tele-vision together with computers [15e17]. This makes it difficult toidentify the unique effects of any one medium. Third, it iscommon to evaluate media effects by assessing time spent on

D. Romer et al. / Journal of Adolescent Health 52 (2013) 613e619614

them. However, media can have many different uses, includingentertainment, information gathering, or more recently, socialnetworking. Because different uses can produce unique effects, itis important to assess not only the time spent with media, butalso how those media are used. Finally, because media can beused in different combinations and for different purposes, it isimportant to evaluate howmedia uses might interact rather thanfocusing only on main effects. Studying combinations of mediause should enable a more comprehensive assessment of suchinteractions.

In this research, we tested hypotheses about a wide range ofmedia in a national longitudinal panel of adolescents and youngadults interviewed on two occasions 1 year apart. We studiedhow time spent with older media (television [TV] and books) aswell as newer forms (the Internet and video games) is related tochanges in school performance as well as participation inextracurricular activity, including sports. Because media use hasbeen suggested to affect mental health [4], we also examinedsymptoms of depression.

Hypotheses About Media Use in Youth

The time displacement hypothesis [18] makes the straight-forward prediction that media use competes with time thatcould more profitably be spent on other, more adaptive pursuits.Indeed, academic performance has been negatively associatedwith amount of time viewing TV [9e11] and playing videogames[12]. Nevertheless, the hypothesis has not been as clearly sup-ported regarding Internet use [13,19].

Displacement effects are less clear regarding extracurricularactivity. Although time spent on TV may displace beneficial clubparticipation [20], its effect on physical activity is less clear[21,22]. Nevertheless, total time spent with screen-based mediahas been inversely related to physical activity [15,17]. Weexamined this hypothesis using a proxy for exercise: namely,engagement in sports.

An important influence on any medium’s effects is itspotential to enhance knowledge and other skills. The learninghypothesis proposes that media that are more likely to lead toknowledge, such as book reading, should correlate positivelywith academic performance [18,23]. Readers have also beenfound to be more civically active [24]. Nevertheless, one concernabout reading on the Internet is that it encourages multitaskingand a shallower form of processing [3] that may serve as a riskfactor for poorer academic performance [25,26]. We were espe-cially interested to see whether media used to gain informationwould enhance favorable outcomes even among youth who usemedia excessively.

In view of the potential for favorable effects of media use, wealso examined the hypothesis that moderate use of media isadaptive. This hypothesis recognizes that moderate media useenables learning as well as development of social capital [20,27]without necessarily interfering with academic or other outcomes.Some support for this hypothesis has been observed for TV andparticipation in extracurricular activity [20,27]. To test both themoderation and time displacement hypotheses, we examined theeffects of media use by comparingmoderate levels of reported usewith low (< 1 hour per day) and high (� 4 hours per day) levels.

The final hypotheses concerned the effects of media use onsymptoms of depression. A recent study of over 6,000 adoles-cents in Switzerland found that moderate use of the Internet wasassociated with the lowest rates of depression [4]. Other research

has found that heavy Internet use [6] and videogaming [28] areassociated with depression. Although these findings suggest thatheavy media use leads to depression, we also tested the alter-native hypothesis that media use increases as a consequence ofdepression. According to this hypothesis, withdrawal from socialactivity, which is a symptom of depression, leads youth to turn tomedia use as a replacement [27,29]. Thus, media use may berelated to depressive symptoms because it reflects rather thancauses depression. This withdrawal hypothesis implies thatincreases in depressive symptoms predict greater media use aswell as reductions in activities, such as sports and clubs.

Method

Participants

This study used data from a panel of respondents in theNational Annenberg Survey of Youth, a nationally representativetelephone survey of youth ages 14e22 years [30,31]. The panelrepresented a cohort of respondents who were interviewed in2008 and re-interviewed in 2009, with a follow-up rate of 58%(N ¼ 719). The panel was balanced with regard to gender (51%female) and included youth ages 14e24 years, with most underage 22 years (86%). Previous studies using the panel haveconfirmed that the follow-up sample is representative of theoriginal [30,31]. Although the panel matched national CurrentPopulation Survey marginals for race and ethnicity as well asregion of the country and rural versus urban residence, weapplied weights designed to match national marginals foreducation level as well as race/ethnicity and region for alldescriptive statistics. Further details about the panel are availablein other reports [30,31].

Measures

The survey measured participants’ demographics (age;gender; school attendance in the past year; and neighborhoodmedian income, determined from the ZIP code of residence),access to the Internet in the home, hours of the day spent withmajor media (the Internet, TV, and videogames), as well asfrequency of book reading, specific uses of the media, and fourbehavioral outcomes. We conducted all assessments at bothinterviews to permit analyses of change over the period of 1 year.

Media measures. We assessed time spent using the Internetand TV with items that asked for the approximate number ofhours spent on a typical weekday and weekend using eachmedium (“< 1 hour, 1e2 hours, 3e5 hours, 6e8 hours, or >

8 hours”).We converted these responses tomedian hourly values(e.g., we coded 3e5 hours as 4 hours) and weight averaged themto provide a single estimate of weekly use. We assessed video-game usewith a single item asking for time spent on a typical day.

We assessed various uses of the Internet on a 4-point scale byasking whether the respondent used the Internet for thatpurpose “most days, some days, less often, or never during theweek.” Factor analysis indicated that these uses could be classi-fied into two types: (1) communicating and social networking(“instant messaging or chatting with friends,” “updating anonline journal or blog,” “using online social networking sites likeMySpace or Facebook,” or “visiting streaming video sites such asYouTube or Hulu”); and (2) gathering information about “local ornational news,” “politics or government,” “entertainment like

D. Romer et al. / Journal of Adolescent Health 52 (2013) 613e619 615

music or movies,” or “something for schoolwork.” We took theaverage values of these reports as measures of Internetcommunication or information gathering, respectively.

We also assessed various uses of other media using the same4-point scale, including the main measure of book reading aswell as “watching national or local news on television,” “readingnewspapers or magazines,” “following shows on television,” and“watching movies on television or at theaters.” An item assessingengagement in online multiplayer videogames was highlyrelated to time spent on videogames, and thus was not includedas a separate indicator.

Behavioral outcomes. We assessed four outcomes, each of whichhad relatively stable reliabilities over the 1-year period: reportedgrade point average assessed on a 4-point scale (A, B, C, and D orless) for all respondents who had attended school in the previoussemester and who reported their grade point average in 2008(78%) (r ¼ .56); participation in (1) sports (r ¼ .64) or (2) clubsand other extracurricular activity (r ¼ .50) (“Most days, somedays, or never”); and an item taken from the Youth Risk BehaviorSurvey to assess two symptoms of depression: the numberof times one had experienced “� 2 weeks of “sadness or hope-lessness that interfered with daily activities in the past12 months” (“Once, twice, three times or more”) (r ¼ .41).

Analysis plan

Initial analyses focused on low, moderate, and high levels oftime spent with TV, the Internet, videogames, and book reading(Table1) aspredictorsof change in the fourbehavioral outcomesofinterest over the 1-year study period using regression models(binary logistic regression for the highly skewed measure ofdepression). Thesemodels controlled fordemographic differencesas well as status on the outcome 1 year earlier. Because we wereinterested in past-year media use as a predictor of change inoutcomes over theprevious year,we focusedon recent rather thanyear-agomedia uses as predictors of change in outcomes. Holdingconstant prior-year media use yielded the same pattern of resultsbut with weaker parameter estimates for recent use. We alsolooked for possible interactions between age and media use, butthese consistently showed no differences in media relations. Totest the withdrawal hypothesis, we regressed media use and highlevels of participation in sports and clubs on change in depression.

To identify potential interactions of media use, we clusteredrespondents according to time spent with the four major media

Table 1Distributions for major media use variables and outcomes (N ¼ 719)

Characteristic % Low % Medium % High

Television 24.5 50.0 25.0Internet 26.4 40.7 32.6Videogames 23.2 65.0 11.2Book reading 12.5 58.7 28.8Grade point averagea 15.7 46.9 37.4Sports 34.8 28.2 37.0Clubs 36.8 34.1 29.1Depression 61.6 19.3 19.1

a Scores based on 606 reports of grades. Distributions based on hours fortelevision, Internet, and videogames (Low ¼ < 1 hr/day; Medium ¼ between 1and < 4 hr/day; High � 4); book reading (Low ¼ never; Medium ¼ some days orless; high ¼ most days); grade point average (Low ¼ C or less; Medium ¼ B;High ¼ A); sports and clubs (Low ¼ Never; Medium ¼ some days; High ¼ mostdays); and depression (Low ¼ never; Medium ¼ once; High > once).

(using original unbinned scores) as well as their different uses(a total of 10 media use indicators). We used a cluster analysisprogram in SPSS (version 18.0; SPSS Inc, Chicago, IL) that iden-tifies groupings (or clusters) based on Euclidian distancesbetween cases [32]. The program uses standardized scores asinput to remove scale differences inmeasures and uses Schwarz’sBayesian Information Criterion (BIC) as the cluster-fit criterion.To identify clusters that were not just attributable to differencesin age and gender, we partialled out this variation from all mediause variables.

Among the solutions provided by the clustering algorithm,the four-cluster solution produced the best BIC. However, we alsoexamined the five- and six-cluster solutions, which had BICvalues that did not differ dramatically from the four-clustersolution and that split two of the clusters into meaningfulsubgroups. The six-cluster solution successfully categorized allbut 11 cases (1.4%), with the resulting profiles sufficiently infor-mative that we retained this solution. We subjected the clustermeans on the 10 indicators to a multidimensional scaling (MDS)analysis also implemented in SPSS. This analysis placed theclusters into a three-dimensional space that facilitated theirdescription.

To assess the ability of the clusters to predict change inoutcomes, we analyzed the behavioral outcomes using eitherlinear regression or binary logistic regression (for the highlyskewed hopelessness score) with each cluster as a dummy-codedpredictor (1¼ yes; 0¼ no), controlling for demographic variables(age, gender, ethnicity, neighborhood income, and urbanicity)and status of the dependent variable in 2008. We tested modelswith and without inclusion of time spent with major media.

Results

Table 1 shows the distribution of major media use andoutcome measures, and Figure 1 shows the significant relationsbetween outcomes and use of each medium controlling fordemographic differences and prior-year outcomes. Based on thedisplacement hypothesis, we expected that increasing media usewould be disruptive to adaptive functioning. However, we sawthat the only clear support for the hypothesis was in the case ofgrades and TV. Other screen media uses were not related togrades. In support of the learning hypothesis, increasing bookreading was favorably related to grades.

The moderation hypothesis received support in the relationbetween Internet use and both sports and clubs. Moderate use ofthe Internet was positively related to both outcomes, with lowerand higher levels associated with less participation. In addition,contrary to the displacement hypothesis, heavier book readingwas associated with greater participation in clubs.

Finally, for mental health, Internet and videogame use wereassociated with increased reports of depression. We tested thewithdrawal hypothesis by determining whether change in recentdepression symptoms predicted change in use of the Internet andvideogames. Indeed, controlling for past symptoms and mediause, recent depressionwas associatedwith greater Internet use, B(standard error [SE]) ¼ .119 (.058), p ¼ .039, and videogameplaying, B (SE) ¼ .144 (.044), p ¼ .001. Consistent with thehypothesis that depressed persons withdraw from social andphysical activity, increase in depression symptoms predictedreduction in high levels of sport participation, adjusted oddsratio¼ .57, p¼ .020, and clubs, adjusted odds ratio¼ .60, p¼ .020.

Figure 1. Significant relations between media use and changes in outcomes over 1-year follow-up period. (A) Television and GPA: Blin ¼ �.122 (SE ¼ .037), p ¼ .001. (B)Reading and GPA: Blin ¼ .070 (SE ¼ .049), p ¼ .001; reading and clubs: Blin ¼ .131 (SE ¼ .047), p ¼ .005. (C) Internet and sports: Bquad ¼ .041 (SE ¼ .017), p ¼ .015; Internetand clubs: Bquad ¼ .047 (SE ¼ .017), p ¼ .007; Internet and hopelessness: Blin ¼ .335 (SE ¼ .122), p ¼ .006. (D) Games and hopelessness: Blin ¼ .549 (SE ¼ .173), p ¼ .001.

D. Romer et al. / Journal of Adolescent Health 52 (2013) 613e619616

Media use clusters

Table 2 lists media use patterns for each cluster. Table 3presents the demographics. The six clusters all had about thesame mean age; however, there was some variation in gender.The MDS revealed that the six clusters were located in a three-dimensional space defined by (1) watching TV shows versususing the Internet and playing videogames; (2) watching TV ingeneral versus reading books and gathering information on the

Table 2Mean standardized media use scores for clusters, controlling for age and gender

Cluster (N) TV hrs Internet hrs Game hrs Book Reading News Reading TV

Low (91) L.52 L.69 L.24 L.79 L.18 L

MI (163) L.77 .05 L.25 .04 L.53 L

HI (161) L.20 L.14 L.24 .80 .70HTV (93) .84 L.64 L.17 L.19 L.13IC (123) .52 .04 L.03 L.67 L.05Gamers (76) .74 1.75 1.59 .19 .19

HI ¼ high information users; HTV ¼ heavy TV users; IC ¼ Internet communicators; LBolded scores are significantly different from 0, p < .05.

Internet; and (3) communicating via the Internet versus readingbooks. We used the locations of the clusters in this space tocharacterize their salient features.

Low users. The group of low users (13%) was located in the centerof the space, with low levels of all media use. Its members livedin neighborhoods with the lowest median income and also hadthe highest proportion of non-Hispanic whites (83%). Itsmembers were among the least likely to still be in school (76%).

News Internet Information Internet Communication TV Shows Movies

.66 L.98 L.85 L1.18 L.15

.45 .24 .35 L.62 L.52

.50 .74 .00 .34 .09

.27 L.85 e1.09 .67 L.29

.16 L.02 .70 .68 .70

.18 .26 .49 L.03 .37

ow ¼ low users; MI ¼ moderate information users; TV ¼ television.

Table 3Mean demographic scores for clusters

Characteristic Cluster

Low (N ¼ 91) Mod Info (N ¼ 163) Heavy Inf (N ¼ 161) HTV (N ¼ 93) Int Comm (N ¼ 123) Gamers (N ¼ 76) Total (N ¼ 708)

Age (yrs) 17.0 17.0 16.9 16.8 16.7 16.7 16.9% Male 54.6 41.3 54.7 50.0 48.0 66.8 51.1% In school 75.7 92.0 87.2 74.7 86.0 85.1 84.8Neighborhood income (000) 39.1 48.9 51.7 43.6 46.5 44.2 46.7% Internet at home 83.7 95.8 89.2 73.3 87.1 84.7 87.1% Non-Hispanic white 82.6 70.5 77.2 62.3 51.9 51.8 67.1% Hispanic 20.4 16.9 11.1 24.2 20.0 15.8 17.4% Non-Hispanic black 7.5 8.7 12.3 23.9 21.5 28.5 15.8% Urban 20.6 30.1 27.8 23.2 23.3 52.1 28.6% Suburban 49.7 53.7 58.0 43.1 56.2 34.7 51.1% Rural 29.7 16.2 14.2 33.7 20.6 13.2 20.2

Heavy Inf ¼ heavy information users; HTV ¼ heavy TV users; Int Comm ¼ Internet communicators; Low ¼ low users; Mod Info ¼ moderate information users; TV ¼television.Characteristics that were significantly different from the mean (p < .05) are in bold.

D. Romer et al. / Journal of Adolescent Health 52 (2013) 613e619 617

Moderate information users. The large group of moderate infor-mation users (23%) tended toward reading books and using theInternet for information on the second dimension. Its memberswere disproportionately female and tended to live in higher-income neighborhoods.

Heavy information users. Youth in the large group of heavyinformation users (23%) were extreme on the second dimension,reading books at a high rate and using TV and the Internetprimarily for information. They lived in neighborhoods with thehighest median income and had the highest rate of Internetaccess in the home.

Heavy TV users. The group of heavy TV users (13%) was extremeon both TV dimensions. It also had the lowest rate of Internetaccess in the home and the highest representation in rural areas.It is therefore not surprising that the group used the Internetsparingly for all purposes.

Internet communicators. Youth in the group of Internetcommunicators (17%) were extreme on the third dimension ofusing the Internet for communication. Aside from having rela-tively higher representation of nonwhites, the group did notstand out on other demographic characteristics.

Gamers. Youth in thedisproportionatelymale gamers group (11%)stood out on the second dimension for using videogames and theInternet at very high rates. They were overrepresented in low-income urban neighborhoods and were less likely to be white.

Table 4Unstandardized regression coefficients for clusters with (Bþ) and without (B) inclusio

Outcome Low Users Mod Info Heavy Info

B Bþ B Bþ B BþGPA L.167* L.196* .089 .042 .119* .043Clubs L.117 L.047 .126* .061 .151* .144Sports L.040 L.059 .108 .054 L.002 L.031Depression L.060 .121 L.196 .045 L.669** L.807

GPA ¼ grade point average; Heavy Info ¼ heavy information users; HTV ¼ heavy telinformation users.

* p < .05.** p < .01.

*** p < .001.

Media clusters and developmental outcomes

Controlling for demographic differences, membership in themedia groups was related to change in several outcomes(Table 4). It is encouraging that the two largest groups were theones with the most adaptive outcomes. The moderate informa-tion users were more active in clubs and consistently but notstatistically advantaged across other outcomes. The heavyinformation users had higher grades, participated in clubs moreoften, and were lowest in depression. These differences largelyremained after controlling for use of the four major media.

Based on the relation between TV use and grades, one wouldexpect the heavy TV users to have lower grades. However, thisgroup fared noworse than others. Nevertheless, the groupwouldhave done better were it not for its heavy TV use. The Internetcommunicators experienced lower grades, a difference that wasremoved after controlling for their lower level of book readingand higher use of TV. The low media users also experiencedlower grades, but their poorer performance was not attributableto interference from media.

The gamers, who were the heaviest media users, did notexperience lower grades; neither did controlling for time spentwith major media increase their grades. Nevertheless, theyparticipated less often in sports. Controlling for their Internet useremoved this deficit.

Two groups with heavy Internet use, the Internet communi-cators and gamers, also experienced greater depression, a changethat declined after controlling for use of the Internet andvideogames.

n of television, Internet, videogame, and book reading main effects

HTV Internet Comm Gamers

B Bþ B Bþ B Bþ.132 .225* L.223** L.094 L.014 L.031

* L.127 L.142 L.116 L.065 L.074 L.063.022 .013 L.001 .038 L.169* L.066

** L.055 .314 .429* .360 .935*** .281

evision users; Internet Comm ¼ Internet communicators; Mod Info ¼ moderate

D. Romer et al. / Journal of Adolescent Health 52 (2013) 613e619618

Discussion

The results from this national panel of over 700 US youthstudied over the course of 1 year confirmed that it is important todistinguish among different uses of media. Consistent with pastresearch, use of TV detracted from grades [9,12,18], whereas bookreading was supportive of school performance [18,23]. Indeed,the MDS analysis indicated that use of TV competed directlywith book reading and use of the Internet. However, heavy useof the Internet and games was not harmful to academic perfor-mance. Although heavy use of the Internet interfered withclub and sports participation, moderate use of the Internet wassupportive. Finally, consistent with past research, heavy use ofthe Internet [4,6] and videogames [28] was positively related toincreases in depression. However, this relation was also consis-tent with the withdrawal hypothesis that young people whoexperience depressive symptoms increase their use of mediaand desist from social and physical activity. Thus, factors thatlead to depression are likely to be drivers of both media use anddepression symptoms.

The finding that moderate use of the Internet was associatedwith increases in clubs and sports is consistent with previousresearch indicating that shared media experience enables youngpeople to build community with peers [20,27]. It is not surprisingthat the two groups that used media at moderate levels (theinformation users) were also more engaged in clubs and no lessinvolved in sports than average. Nevertheless, these youth tendedto live in more advantaged neighborhoods, which suggests thattheir favorable outcomes as well as healthy media use benefitedfrom resources available in those settings. Hence, concerns aboutthe digital divide are not assuaged by these results [33].

The implications of our findings for health promotion areclear. It is important to look at the whole pattern of media usebefore drawing conclusions about the well-being of youth. Thecluster analysis demonstrated that not all youth who use mediaat excessive levels will experience adverse outcomes. The gamersand heavy TV users did not experience lower grades thanaverage. The group that experienced adverse effects of TVviewing most strongly (Internet communicators) also tended toread books less frequently than average.

Consistent with recommendations regarding media use [34],book reading and informational use of the Internet are mediauses that should be encouraged. Youth who use media in theseways appear to experience better outcomes across the board.Indeed, the MDS analysis indicated that both TV and Internetcommunication compete with book reading. Recommendationsthat limit screenmedia use to< 2 hours of entertainment per dayfor adolescents [35]may be unrealistic andmay overlook some ofthe benefits of moderate screen media use. Indeed, grouping allscreen media together overgeneralizes and should be avoided.Finally, youth who engage in heavy Internet use and video-gaming and who withdraw from social activity may be at risk formental health problems, but this may be as much a marker ofdepression as a cause of it. Nevertheless, youth who exhibit thispattern should be encouraged to seek help if they show signs ofdepression.

Some of the findings invite the need for further research. Wedo not understand why book reading is supportive of clubparticipation, but it is also a common finding in adults [24]. Wealso did not have explicit information about multitasking,although the heavier media use groups likely engaged in it. Inaddition, we did not probe for use of different videogame genres,

some of which may be more harmful to health than others[36,37]. We also relied heavily on self-reports of media use.However, a study that compared self-reports of TV use with dailylogs found relatively high overlap even in youth ages 11e15 years(r ¼ .47) [38]. Finally, we did not measure Internet or gamingaddiction, which might have identified youth at greater risk ofmental health and other problems [39,40]. Despite these limi-tations, our results provide helpful guidance that has not been asclearly available in previous research.

Acknowledgments

This research was supported by the Annenberg Public PolicyCenter. An earlier version of this article was presented at themeetings of the International Communication Association in2010. The authors sadly acknowledge the unfortunate death ofDr. Bagdasarov while the article was under review.

References

[1] Pew Research Center. Pew Internet and American Life Project, 2012,http://pewinternet.org/Data-Tools/Get-the-Latest-Statistics.aspx. AccessedDecember 31, 2012.

[2] Rideout VJ, Foehr UG, Roberts DF. Generation M2: Media in the Lives of8- to 18-Year-Olds, 2010, http://www.kff.org/entmedia/upload/8010.pdf.Accessed December 31, 2012.

[3] Carr N. The shallows: what the Internet is doing to our brains. New York:W. W. Norton & Co.; 2010.

[4] Belanger RE, Akre C, Berchtold A, et al. A U-shaped association betweenintensity of Internet use and adolescent health. Pediatrics 2011;127:e330e5.

[5] Cummings HM, Vanderwater EA. Relation of adolescent video game play totime spent in other activities. Arch Pediatr Adolesc Med 2007;161:684e9.

[6] Ybarra ML, Alexander C, Mitchell KJ. Depressive symptomatology, youthInternet use, and online interactions: A national survey. J AdolescentHealth 2005;36:9e18.

[7] Dennison BA, Edmunds LS. The role of television in childhood obesity.Progr Pediatr Cardiol 2008;25:191e7.

[8] Weicha JL, Sobol AM, Peterson KE, et al. Household television access:Associations with screen time, reading, and homework among youth.Ambul Pediatr 2001;1:244e51.

[9] Sharif I, Sargent JD. Association between television, movie, and video gameexposure and school performance. Pediatrics 2006;118:e1061e70.

[10] Gentile DA, Walsh DA. A normative study of family media habits. Appl DevPsychol 2002;23:157e78.

[11] Barr-Anderson DJ, van den Berg P, Neumark-Sztainer D, et al. Character-istics associated with older adolescents who have a television in theirbedrooms. Pediatrics 2008;121:718e24.

[12] Anderson CA, Dill KE. Video games and aggressive thoughts, feelings, andbehavior in the laboratory and life. J Pers Soc Psychol 2000;78:772e90.

[13] Kim JH, Lau CH, Cheuk K, et al. Predictors of heavy Internet use andassociations with health-promoting and health risk behaviors among HongKong university students. J Adolescence 2010;33:215e20.

[14] Gentile DA, Swing EL, Lim CG, et al. Video game playing, attention prob-lems, and impulsiveness: Evidence of bidirectional causality. Psychol PopMedia Culture 2012;1:62e70.

[15] Iannotti RJ, Kogan MD, Janssen I, et al. Patterns of adolescent physicalactivity, screen-based media use, and positive and negative health indi-cators in the U.S. and Canada. J Adolescent Health 2009;44:493e9.

[16] Ramirez ER, Norman GJ, Rosenberg DE, et al. Adolescent screen time andrules to limit screen time in the home. J Adolescent Health 2011;48:379e85.

[17] Sisson SB, Broyles ST, baker BL, et al. Screen time, physical activity, andoverweight in U.S. youth: National Survey of Children’s Health 2003.J Adolescent Health 2010;47:309e11.

[18] Thompson FT, Autin WP. Television viewing and academic achievementrevisited. Education 2003;124:194e202.

[19] Jackson LA, von Eye A, Biocca FA, et al. Does home Internet use influencethe academic performance of low-income children? Dev Psychol 2006;42:429e35.

[20] Pasek J, Kenski K, Romer D, et al. America’s youth and communityengagement: How use of mass media is related to political knowledge andcivic activity among 14 to 22 year olds. Commun Res 2006;33:115e35.

[21] Gross EF, Juvonen J, Gable SL. Internet use and well-being in adolescence.J Soc Issues 2002;58:75e90.

D. Romer et al. / Journal of Adolescent Health 52 (2013) 613e619 619

[22] Van den Bulck J. Is television bad for your health? Behavior and body imageof the adolescent “couch potato.” J Youth Adolescence 2000;29:273e88.

[23] Bauerlain M. The dumbest generation: How the digital age stupefies youngAmericans. New York: Penguin Group; 2007.

[24] Bradshaw T, Nichols B, Bauerlein M. Reading at risk: A survey of literaryreading inAmerica. Alexandria (VA):National Endowment for theArts; 2004.

[25] Junco R, Cotten SR. No A 4 U: The relationship between multitasking andacademic performance. Computers Educ 2012;59:505e14.

[26] Ophir E, Nass C, Wagner AD. Cognitive control in media multitaskers. ProcNatl Acad Science U S A 2009;106:15583e7.

[27] Romer D, Jamieson KH, Pasek J. Building social capital in young people: Therole ofmassmedia and life outlook. Political Communication 2009;26:65e83.

[28] Mentzoni RA, Brunborg GS, Molde H, et al. Problematic video game use:Estimated prevalence and associations with mental and physical health.Cyberpsychol Behav Soc Netw 2011;14:591e6.

[29] Potts R, Sanchez D. Television viewing and depression: No news is goodnews. J Broadcast Electron Media 1994;38:79e90.

[30] Dunlop SM, More E, Romer D. Where do youth learn about suicides on theInternet, and what influence does it have on suicidal ideation? J ChildPsychol Psychiatry 2011;52:1073e80.

[31] Khurana A, Romer D. Modeling the distinct pathways of influence of copingstrategies on youth suicidal ideation: A national longitudinal study. PrevSci 2012;13:644e54.

[32] Norusis M. SPSS 12.0 guide to data analysis. Upper Saddle River (NJ):Prentice Hall; 2004.

[33] Zickuhr K, Smith A. Digital differences: For some groups, digital disparitiesstill remain. Pew Research Center, http://pewinternet.org/Reports/2012/Digital-differences.aspx. Accessed December 31, 2012.

[34] American Academy of Pediatrics. Children, adolescents, and television.Pediatrics 2001;107:423e6.

[35] American Academy of Pediatrics. Policy statementdMedia education.Pediatrics 2010;126:1012e7.

[36] Anderson CA, Gentile DA, Buckley KE. Prosocial, antisocial, and othereffects of recreational video games. In: Singer DG, Singer JL, editors.Handbook of children and the media. 2nd ed. Thousand Oaks (CA): Sage;2012:249e72.

[37] Papastergiou M. Exploring the potential of computer and video games forhealth and physical education: A literature review. Comput Educ 2009;53:603e22.

[38] Schmitz KH, Harnack L, Fulton JE, et al. Reliability and validity of a briefquestionnaire to assess television viewing and computer use by middleschool children. J Sch Health 2004;74:370e7.

[39] Gentile DA, Choo H, Liau A, et al. Pathological video game use amongyouths: A two-year longitudinal study. Pediatrics 2011;127:e319e29.

[40] Lam LT, Zi-Wen P. Effect of pathological use of the Internet on adolescentmental health. Arch Pediatr Adolesc Med 2010;164:901e6.