Although video games have been traditionally associated

Transcription

1 CYBERPSYCHOLOGY, BEHAVIOR, AND SOCIAL NETWORKING Volume 14, Number 11, 2011 ª Mary Ann Liebert, Inc. DOI: /cyber Is Playing Exergames Really Exercising? A Meta-Analysis of Energy Expenditure in Active Video Games Wei Peng, Ph.D., Jih-Hsuan Lin, M.A., and Julia Crouse, M.A. Abstract This article reports a meta-analysis of energy expenditure (EE) of playing active video games (AVGs). In particular, heart rate (HR), oxygen consumption (VO2), and EE were assessed and three moderators for the effects of AVGs types of AVG, player age, and player weight status were analyzed. The results show that playing AVGs significantly increased HR, VO2, and EE from resting. The effect sizes of playing AVGs on HR, VO2, and EE were similar to traditional physical activities. AVG type and player age were significant moderators for the effects of AVGs. The finding suggests that AVGs are effective technologies that may facilitate light- to moderate-intensity physical activity promotion. Introduction Although video games have been traditionally associated with sedentary lifestyles, 1 an emerging type of active video gaming, dubbed as exergames, has been embraced by the public as a fun and engaging form of exercise. In 2003, Konami of Japan reported > 6.5 million sales of its rhythmic dancing game Dance Dance Revolution Ò (DDR). 2 The Wii Fit and Wii Fit Plus combined have sold > 11 million copies worldwide as of Hoping to capitalize on the success of Nintendo s active video game (AVG) market, both Sony and Microsoft debuted new motion capture hardware and exergames in late We adopt the term active video games to describe these commercial off-the-shelf video games that incorporate physical activity into game play by requiring players body movements to interact with the game system. In the past decade, researchers have utilized different types of AVGs to examine the energy expended during game play to determine whether AVGs qualify as a new form of physical activity. Overwhelmingly, studies have found that energy expenditure (EE) a of playing AVGs using the dance pad, 4 6 the Playstation 2 EyeToy, 6 8 the Wii console system, and the Wii balance board 9 is equivalent to light- to moderate-intensity physical activity. Additionally, many of these studies found EE differences between different groups. For example, several studies found significant gender differences; most commonly, males were found to have higher EE than females. 4,10 12 Other studies found differences between adults and children 6 or overweight and lean participants The main goal of the present study was to take a metaanalysis approach to synthesize current literature of AVGs. Specifically, the objectives were (a) to provide a summary effect of EE and related physiological measurements of AVG studies, (b) to compare the effects of playing AVGs and traditional physical activities, and (c) to examine whether AVG type and sample characteristics influence the effects of AVGs. Three outcome variables were included in this study: heart rate (HR), oxygen consumption (VO2), and EE. b Methods Search strategy To ensure a comprehensive inclusion of relevant studies in this meta-analysis, four search strategies were adopted. First, the authors conducted systematic literature searches using seven databases (Communication and Mass Media Complete, Cochrane Library, Medline, PsychInfo, PubMed, SPORTDiscus, and Web of Science) in English to retrieve peer-reviewed articles that focus on investigating the effects of AVG play on players physiological responses. Two authors searched articles from January 1995 to October 2010 using the following keywords in combination: physical activit*, exercise, weight loss, obesity, motor activit*, leisure activit*, physical fitness, physical education, physical training, sports, games, video game, computer game, active video game, and exergame. Two authors examined the abstracts of all articles (n = 8944 c ) from this search to determine the extent of relevancy and 47 studies were identified for further selection. Two authors then read the full papers and selected studies that meet the following criteria. College of Communication Arts and Sciences, Michigan State University, East Lansing, Michigan. 681

2 682 PENG ET AL. Selected studies for this meta-analysis must (a) employ at least one AVG condition (interactive multimedia such as Web sites, s, and sedentary video games that do not require physical activities were excluded), (b) have normal population instead of rehabilitation patients, (c) report both resting d and AVG playing data in terms of HR, VO2, or EE, (d) report data that can be converted to a standard unit of measurement, and (e) report both means and standard deviations, allowing the authors to calculate the effect sizes. One study was excluded because of insufficient information to calculate the effect sizes. 16 Another study 17 was excluded following authors suggestion that EE was not measured from indirect calorimetry. e Therefore, 16 studies were obtained from this systematic database search. Second, we examined the reference lists of these 16 reports using the above inclusion criteria. The search generated two additional research reports. 18,19 However, one report 18 was further excluded as it reported the same results as in another study. 17 Third, a forward citation search was conducted in the Social Science Citation Index on all the 16 studies identified from the systematic database search. One additional study was found. 14 Finally, we examined relevant literature review papers for potential studies. No additional articles were identified. From the above four approaches, a total of 18 studies (n = 354) were included in this meta-analysis. Table 1 lists the outcome variables measured in each study. Effect size extraction and calculation As all the studies employed within-subjects design, in which the participants were assessed at rest, while playing a number of AVGs, or while engaging in other traditional physical activities, we treated resting as the pretest condition and AVG playing and other physical activities as the posttest conditions. Given the nature of the within-subjects design and the small sample size in each study, effect size was calculated using Hedge s g To calculate the effect size in a pretest posttest design, means and standard deviations from both pretest and posttest conditions, sample size, and pretest posttest correlations for the outcome variables (HR, VO2, and EE) were needed. 23,25 All the calculations were performed on the random-effects model assumption using the computer software Comprehensive Meta-Analysis 2.2, which was adopted in recent meta-analyses. 26,27 None of the studies reported the pretest (resting) and posttest (AVG playing or other physical activity) correlation for the outcomes variables. Authors of nine studies provided the pretest posttest correlations 5,10,11,14,17,28 31 based on request. A meta-analysis was conducted with these acquired correlations to obtain a summary value of the pretest posttest correlation for each outcome variable 25,27 to replace the missing correlations in the remaining nine studies. Nine studies employed multiple AVGs and physical activities or reported data in multiple time points. For example, Willems and Bond 31 reported brisk walking three times and three types of AVGs (Wii tennis, Wii baseball, and Wii boxing). For the meta-analysis of the effect size of AVGs, each study was treated as the unit of analysis so that a study that employed more AVGs as treatment conditions will not be assigned more weight than a study that has fewer such conditions. 23 Therefore, a combined effect across the different AVG conditions was calculated in each study. The same strategy was adopted to calculate a combined effect across the different traditional physical activities in each applicable study. Meta-analysis approach Once the effect size and variance of AVG playing and physical activity for each outcome variable in each study were calculated, we computed a random-effects model metaanalysis of the effects of playing AVGs on HR, VO2, and EE and the effects of traditional physical activity on HR, VO2, and EE. The Q and I 2 statistics were used to examine whether significant heterogeneity existed. f Table 1. Description of Outcome Variables Study name n HR: AVG HR: PA VO2: AVG VO2: PA EE: AVG EE: PA Barkley and Penko X X X X Fawkner et al X X X Graf et al Outlier Outlier Outlier Outlier X X Graves et al X X X Haddock et al X X Haddock et al X X X X X X Lanningham-Foster et al X Lanningham-Foster et al X Maddison et al X X X Mellecker and McManus X X Penko and Barkley X X X X Ridley and Olds X X X Sell et al X X Siegel et al X X X Straker and Abbott 8 20 X X X Tan et al X X Unnithan et al. 15,a 22 X X Willems and Bond X X X X X X Number of studies (total sample size) 18 (354) 13 (229) 4 (66) 14 (266) 5 (88) 13 (237) 3 (53) a Resting HR and EE were not reported. Thus, effect sizes of AVG on HR and EE were not calculated. The outlier is defined in endnote h. n, sample size; AVG, active video game; PA, physical activity; EE, energy expenditure.

3 META-ANALYSIS OF ACTIVE VIDEO GAME 683 Subgroup analysis 23 was used to compare the differences between playing AVGs and traditional physical activity. Only the studies that included both AVGs and traditional physical activities as the treatment conditions were included in the subgroup analysis. Within each of the included studies, a combined effect across the different types of AVGs and a combined effect across different types of physical activities were calculated. 23 Moderator analysis approach The moderators tested in this meta-analysis were AVG type, age, and weight status of participants. These three variables have been shown to influence EE. 4,6,15 In addition, most of the included studies explored EE differences based on these variables. Gender was not included as a moderator because only three studies reported raw data for male and female separately. AVG type was coded based on the main body movements required to play the games: upper body movements (e.g., Wii sports), lower body movements (e.g., DDR), or whole body movements (e.g., PS2 Eyetoy Final Furlong). We coded these games based on the authors descriptions of participants game play and videos showing individuals playing the game. Within a study, a combined effect was calculated for multiple AVGs categorized into the same AVG type. Each AVG type was treated as a unit of analysis. Age was coded as children (6 17 years old) or adult (18 years old or above). Only studies that had an exclusively child sample or an exclusively adult sample and studies that had Table 2. Study Characteristics Study names Age Gender Weight Active game systems (game titles) Concentration Country Barkley and Penko 10 Adult 6 M Overweight Nintendo Wii (Wii sports boxing) Upper body United States 6F Fawkner et al. 5 Children 20 F Normal Zigzag Dance Mats (Xer-Dance) Lower body United Kingdom Graf et al. 11 Children 14 M Both a. Sony Playstation 2 (DDR) Lower body United States 9 F b. Nintendo Wii (Wii sports Upper body bowling and boxing) Graves et al. 12 Children 7 M Normal Nintendo Wii (Wii sports bowling, Upper body United Kingdom 6F tennis, and boxing) Haddock et al. 28 Children 18 M Overweight XaviX ( Jackie Chan Studio Fitness) Lower body United States 5F Haddock et al. 29 Children 13 M Overweight CatEyeÔ Gamebike + Sony Lower body United States 7F Playstation 2 (Disney s Cars) Lanningham- Children 11 M Both Nintendo Wii (Wii sports boxing) Upper body United States Foster et al F Lanningham- Adult 10 M Both Foster et al F Maddison et al. 7 Children 12 M Normal and a. Sony Playstation 2 (Eyetoy) Upper body United States 13 F overweight b. Xbox (DDR Ultramix 2) Lower body Mellecker and Upper body New Zealand McManus 36 Children 11 M Normal a. Sony Playstation 2 (Eyetoy KnockOut) 10 F b. Sony Playstation 2 (Dance UK) Lower body c. Sony Playstation 2 (Homerun, Whole body Groove, and AntiGrav) Penko and Barkley 14 Children 11 M Normal a. XaviX (bowling) Upper body Hong Kong 7 F b. XaviX J-Ma Jackie s Action Run Lower body Ridley and Olds 37 Children 12 M Both Nintendo Wii (Wii sports boxing) Upper body United States 12 F Sell et al. 35 Children 5 M Normal a. Arcade (Daytona, Air Hocky, and Mini Dunxx) Upper body Australia 5 F b. Arcade (Final Furlong) Whole body Siegel et al. 30 Adult 19 M Both Sony Playstation 2 (DDR) Lower body United States Straker and Abbott 8 Adult 6 M Both a. Arcade (3-kick) Whole body United States 7 F b. XaviX (Jackie Chan Studio Upper body Fitness boxing) c. Stationary bike + Sony Lower body Playstation 2 (Disney s Cars) Tan et al. 19 Children 12 M Normal Sony Playstation 2 Whole body Australia 8F (Eyetoy Cascade) Unnithan et al. 15 Both 21 M Normal Sony Playstation 2 (DDR 3 rd Mix) Lower body Singapore 19 F Willems and Bond 31 Children 16 M Normal and Sony Playstation 2 (DDR) Lower body United Kingdom 6F overweight Number of studies Adults 7 M Both Nintendo Wii (Wii sports tennis, Upper body United Kingdom (total sample size) 3F boxing, and bowling) M, male sample size; F, female sample size; children, 6 17 years old; adults, 18 years old and above.

4 684 PENG ET AL. Table 3. Statistics for Each Study Regarding the Impact of Active Video Game Compared with Resting on Heart Rates Study name Hedges s g Standard error Variance Lower limit Upper limit z value p value Relative weight Barkley and Penko Fawkner et al Graves et al Haddock et al Maddison et al Mellecker and McManus Penko and Barkley Ridley and Olds Sell et al Siegel et al Straker and Abbott Tan et al Willems and Bond Total mixed age groups yet reported the data of each subgroup were included in the analysis, using age as a moderator. Within a study, a combined effect across multiple AVG conditions was calculated for each age group. Each age group was treated as a unit of analysis. Weight status was coded as normal weight (body mass index between the 5th and 84th percentile) or overweight or obese (body mass index between the 85th and 99th percentile). Only the studies that had an exclusively normal weight or an exclusively overweight or obese sample and studies that had mixed weight status groups yet reported the data of each subgroup were included in the analysis, using weight status as a moderator. Within a study, a combined effect across multiple AVG conditions was calculated for each weight group. Each weight status group was treated as a unit of analysis. Results Sample A total of 354 participants were included in these 18 studies. The sample size in each study ranged from 10 to 40, with a mean of 20 subjects. The sample characteristics are listed in Table 2. Efficacy of AVGs The meta-analysis results indicated that AVG play resulted in higher heart beat per minute than resting (k = 13 g, g = 2.63, SE = 0.24, 95% confidence interval (CI) = , Q 12 = 40.86, p < 0.001, I 2 = 70.63), higher oxygen consumption than resting (k = 14 h, g = 2.97, SE = 0.28, 95% CI = , Q 13 = 74.05, p < 0.001, I 2 = 82.44), and higher EE than resting (k = 13, g = 2.49, SE = 0.17, 95% CI = , Q 12 = 32.76, p < 0.001, I 2 = 63.37). The results of the analyses are presented in Tables 3 5. Comparing AVG play and traditional physical activity The subgroup analysis showed that the effect size of AVG play on HR (k = 4 i, g = 2.30, SE = 0.47, 95% CI = ) did not differ significantly (Q = 1.66, p > 0.05) from traditional physical activities (k = 4, g = 1.46, SE = 0.45, 95% CI = ). Similarly, although AVG play (k = 5 j, g = 2.53, SE = 0.39, 95% CI = ) yielded smaller effects on VO2 (Q = 1.65, p > 0.05) than traditional physical activities (k = 5, g = 3.36, Table 4. Statistics for Each Study Regarding the Impact of Active Video Game Compared with Resting on VO2 Study name Hedges s g Standard error Variance Lower limit Upper limit z value p value Relative weight Barkley and Penko Fawkner et al Graves et al Haddock et al Haddock et al Maddison et al Penko and Barkley Ridley and Olds Sell et al Siegel et al Straker and Abbott Tan et al Unnithan et al Willems and Bond Total

5 META-ANALYSIS OF ACTIVE VIDEO GAME 685 Table 5. Statistics for Each Study Regarding the Impact of Active Video Game Compared with Resting on Energy Expenditure Study name Hedges s g Standard error Variance Lower limit Upper limit z value p value Relative weight Fawkner et al Graf et al Graves et al Haddock et al Haddock et al Lanningham-Foster et al Lanningham-Foster et al Maddison et al Mellecker and McManus Ridley and Olds Siegel et al Straker and Abbott Willems and Bond Total SE = 0.51, 95% CI = ), and AVG play (k = 3, g = 2.09, SE = 0.36, 95% CI = ) yielded smaller effects on EE (Q = 0.99, p > 0.05) than traditional physical activities (k = 3, g = 2.77, SE = 0.58, 95% CI = ), the difference was not statistically significant. Moderator analyses The moderator analyses results are presented in Table 6. Specifically, AVG type was not a significant moderator for the outcome variables HR and VO2. However, AVG type was a Table 6. Mean Weighted Effect Sizes of Active Video Games by Categorical Moderating Variables Variable Moderators n k g SE Variance 95% CI Q HR Age Adult Children Total between 0.86 Weight Normal weight Overweight Total between 1.01 Systems Upper body Lower body Whole body Total between 0.63 VO2 Age Adult Children Total between 0.00 Weight Normal weight Overweight Total between 0.13 Systems Upper body Lower body Whole body Overall 4.13 EE Age Adult Children Total between 4.31* Weight Normal weight Overweight Total between 1.28 Systems Upper body Lower body Whole body Overall 7.19* *Significant at 0.05 level. n, sample size; k, number of studies; g, Hedge s g, mean weighted effect size; SE, standard error; CI, confidence interval; Q, Q value for heterogeneity.

6 686 PENG ET AL. significant moderator for EE (Q = 7.19, p < 0.05). Lower body and whole body movement systems produced more EE than upper body systems. The moderator analysis also suggested that the effects of playing AVGs did not differ significantly between children and adults in terms of HR and VO2. However, children yielded larger effects than adults in terms of EE (Q = 4.31, p < 0.05). Weight status was not a significant moderator for HR, VO2 or EE. Discussion The meta-analysis demonstrated that playing AVGs significantly increased HR, VO2, and EE. However, AVGs were not found to have significantly different effect sizes in increasing HR, VO2, and EE than traditional moderate-intensity physical activities. This finding suggests that playing AVGs is not only fun because it also produces similar magnitude of effect as lightto moderate-intensity physical activity. Nevertheless, this result needs to be interpreted in a cautionary fashion as the subgroup analyses involved only three to five individual studies that included both AVG and traditional PA. On the basis of this result, we cannot state that playing AVG alone meets the recommended daily moderate to vigorous physical activity recommended to Americans. 33 On the other hand, the metaanalysis demonstrated that these AVGs seem to have good potential to improve aerobic fitness 34 as they yielded similar effects in increasing HR as traditional physical activities. The Q and I 2 statistics indicated that effect sizes among the included studies were not homogenous. Therefore, moderator analyses were conducted. The moderator analysis demonstrated that AVGs that mainly involved upper body movements had significantly smaller effects than AVGs that mainly involved lower or whole body movements. The findings suggest that AVGs that only involved upper body movements were not the best candidates to be used for physical activity promotion, especially the promotion of moderate to vigorous physical activities. AVG designers who intend to use AVGs for physical activities should intentionally build more lower body movements in the AVGs. Another advantage of emphasizing the lower body movements in the AVGs is that players may use energy-saving strategies in AVG games focusing on upper body movements. For instance, instead of actually swinging their arms to play tennis in Wii Sports, players might just shake their wrists. However, it is much more difficult to adopt the energy-saving strategies in AVGs involving lower body movements. The moderator analysis demonstrated that AVGs had greater impact for children than adults in terms of EE. It seems that it is promising to use AVGs to promote physical activity among children. However, cautionary interpretation is needed for the moderation analysis as the number of studies involving adults (k = 3) is really small. More AVGbased physical activity studies are needed to establish the evidence base for the adult population. In addition, because of the small number of studies, we were not able to categorize the participants in a more fine-grained manner and the age range of participants that lumped in the children category was from 6 to 17 years. More research studies with participants in a narrower age range group are needed to consider the important developmental differences. Besides the moderation analysis, future studies could also examine the potential mediating factors between different types of AVGs and the EE outcomes. Specifically, psychological factors such as enjoyment or intrinsic motivations might explain the difference in effect sizes of physiological outcomes. Several methodological limitations of the current research in AVGs need to be highlighted. First, the total participants included in the meta-analysis were only 354. However, as all the studies in the meta-analysis had small sample size, no single study was given more weight because of the sample size. Second, the Q and I 2 statistics of heterogeneity were very large. Moderator analyses were conducted and revealed significant differences of different types of AVGs and AVGs for different populations. Therefore, the summary effect size of each type of AVG is a better indicator than the summary effect size of all AVGs. Third, all the included studies employed within-subjects design. However, not all the studies randomized the order of the multiple treatment conditions. Even though the participants rested for a while before moving to the next treatment condition, order effects cannot be completely eliminated. Finally, as the studies included in the meta-analysis were all published in peer-reviewed journals, the results may be influenced by publication bias or the bias in social scientific research to publish those studies with statistically significant findings. Endnotes a Energy expenditure is a typical form to quantify physical activity, defined as any bodily movement produced by the contraction of skeletal muscles, resulting in energy expenditure. 38 For review of the energy expenditure concept, see refs. 39,40 b These three outcome variables are the most commonly used in the kinesiology field to assess physical activity. 39,40 These are also the most reported outcomes in the studies in this meta-analysis. c These are the raw data of the total number of hits from these databases. There are many overlapping articles from these databases. d One study 31 did not report resting data in the article, but the data were provided by the authors. e Indirect calorimetry is the best choice when assessing physical activity in a laboratory setting in a short period of time. 42 f Q test examines whether the studies share heterogeneous effect sizes, and I 2 further indicates the proportion of true heterogeneity to total variance of observed effects. If significant heterogeneity exists, moderator analysis or metaregression is recommended to explain the heterogeneity. 23 g One study 11 was excluded in the analysis following Lipsey and Wilson 32 because the effect reported in the study (g = 11.6) exceeded the mean effect size for remaining studies in the sample by > 7 standard deviations. h Graf et al. 11 was excluded as an outlier. i Graf et al. 11 was excluded as an outlier. j Graf et al. 11 was excluded as an outlier. Acknowledgment This study is partially supported by Robert Wood Johnson Foundation Health Games Research. Disclosure Statement No competing financial interests exist.

7 META-ANALYSIS OF ACTIVE VIDEO GAME 687 References 1. Strum R. Childhood obesity what we can learn from existing data on societal trends, part 1. Preventing Chronic Diseases 2005; 2:A Sinclair J, Hingston P, Masek M. (2007) Considerations for the design of exergames. 5th International Conference on Computer Graphics and Interactive Techniques. Australia and Southeast Asia. 3. Riley D. (2010) 2009 video game software sales across top global markets experience declines for console and portable platforms. html (accessed Oct. 5, 2010). 4. Epstein LH, Beecher MD, Graf JL, Roemmich JN. Choice of interactive dance and bicycle games in overweight and nonoverweight youth. Annals of Behavioral Medicine 2007; 33: Fawkner SG, Niven A, Thin AG, MacDonald MJ, Oakes JR. Adolescent girls energy expenditure during dance simulation active computer gaming. Journal of Sports Sciences 2010; 28: Lanningham-Foster L, Foster RC, McCrady SK, Jensen TB, Mitre N, Levine JA. Activity-promoting video games and increased energy expenditure. The Journal of Pediatrics 2009;154: Maddison R, Mhurchu CN, Jull A, Jiang Y, Prapavessis H, Rodgers A. Energy expended playing video console games: an opportunity to increase children s physical activity? Pediatric Exercise Science 2007; 19: Straker L, Abbott R. Effect of screen-based media on energy expenditure and heart rate in 9- to 12-year-old children. Pediatric Exercise Science 2007; 19: Miyachi M, Yamamoto K, Ohkawara K, Tanaka S. METs in adults while playing active video games: a metabolic chamber study. Medicine and Science in Sports and Exercise 2010; 42: Barkley JE, Penko A. Physiologic responses, perceived exertion, and hedonics of playing a physical interactive video game relative to a sedentary alternative and treadmill walking in adults. Journal of Exercise Physiology Online 2009; 12: Graf DL, Pratt LV, Hester CN, Short KR. Playing active video games increases energy expenditure in children. Pediatrics 2009; 124: Graves LEF, Ridgers ND, Stratton G. The contribution of upper limb and total body movement to adolescents energy expenditure whilst playing Nintendo Wii. European Journal of Applied Physiology 2008; 104: Lanningham-Foster L, Jensen TB, Foster RC, Redmond AB, Walker BA, Heinz D, et al. Energy expenditure of sedentary screen time compared with active screen time for children. Pediatrics 2006; 118:e1831 e Penko AL, Barkley JE. Motivation and physiologic responses of playing a physically interactive video game relative to a sedentary alternative in children. Annals of Behavioral Medicine 2010; 39: Unnithan VB, Houser W, Fernhall B. Evaluation of the energy cost of playing a dance simulation video game in overweight and non-overweight children and adolescents. International Journal of Sports Medicine 2006; 27: Leatherdale ST, Woodruff SJ, Manske SR. Energy expenditure while playing active and inactive video games. American Journal of Health Behavior 2010; 34: Graves LEF, Stratton G, Ridgers ND, Cable NT. Energy expenditure in adolescents playing new generation computer games. British Journal of Sports Medicine 2008; 42: Graves LEF, Stratton G, Ridgers ND, Cable NT. Comparison of energy expenditure in adolescents when playing new generation and sedentary computer games: cross sectional study. British Medical Journal 2007; 335: Tan B, Aziz AR, Chua K, Teh KC. Aerobic demands of the dance simulation game. International Journal of Sports Medicine 2002; 23: Biddiss E, Irwin J. Active video games to promote physical activity in children and youth: a systematic review. Archives of Pediatrics & Adolescent Medicine 2010; 164: Daley AJ. Can exergaming contribute to improving physical activity levels and health outcomes in children? Pediatrics 2009; 124: Mark R, Rhodes RE, Warburton DE, Bredin SS. Interactive video games and physical activity: a review of the literature and future directions. Health and Fitness Journal of Canada 2008; 1: Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. (2009) Introduction to meta-analysis. The Altrium: John Wiley & Sons. 24. Grissom RJ, Kim JJ. (2005) Effect sizes for research: a broad practical approach. Mahwah, NJ: Erlbaum. 25. Hedges LV, Olkin I. (1985) Statistical methods for meta-analysis. Orlando, FL: Acadmic Press. 26. Noar SM, Pierce LB, Black HG. Can computer-mediated interventions change theoretical mediators of safer sex? A metaanalysis. Human Communication Research 2010; 36: Rains SA, Young V. A meta-analysis of research on formal computer-mediated support groups: examining group characteristics and health outcomes. Human Communication Research 2009; 35: Haddock BL, Brandt AM, Siegel SR, Wilkin LD, Joung-Kyue H. Active video games and energy expenditure among the overweight children. International Journal of Fitness 2008; 4: Haddock BL, Siegel SR, Wikin LD. The addition of a video game to stationary cycling: the impact on energy expenditure in overweight children. Open Sports Science 2009; 2: Siegel SR, Haddock BL, Dubois AM, Wilkin LD. Active video/arcade games (exergaming) and energy expenditure in college students. International Journal of Exercise Science 2009; 2: Willems MET, Bond TS. Comparison of physiological and metabolic responses to playing nintendo wii sports and brisk treadmill walking. Journal of Human Kinetics 2009; 22: Lipsey M, Wilson D. (2001) Practical meta-analysis. Thousand Oaks, CA: Sage Publications. 33. Department of Health and Human Services. (2009) Physical activity guidelines for Americans. Available at (accessed Oct. 1, 2009). 34. Baquet G, Van Praagh E, Berthoin S. Endurance training and aerobic fitness in young people. Sports Medicine 2003; 33: Sell K, Lillie T, Taylor J. Energy expenditure during physically interactive video game playing in male college students with different playing experience. Journal of American College Health 2008; 56: Mellecker RR, McManus AM. Energy expenditure and cardiovascular responses to seated and active gaming in children. Archives of Pediatrics and Adolescent Medicine 2008; 162: Ridley K, Olds T. Video center games: energy cost and children s behaviors. Pediatric Exercise Science 2001; 13:

8 688 PENG ET AL. 38. Casperson CJ, Powell KE, Christenson GM. Physical activity, exercise, and physical fitness: definitions for health related research. Public Health Reports 1985; 100: Valanou E, Bamia C, Trichopoulou A. Methodology of physical-activity and energy-expenditure assessment: a review. Journal of Public Health 2006; 14: Levine JA. Measurement of energy expenditure. Public Health Nutrition 2005; 8: Address correspondence to: Dr. Wei Peng College of Communication Arts and Sciences Michigan State University 430 Communication Arts and Science East Lansing, MI pengwei@msu.edu

9 Copyright of CyberPsychology, Behavior & Social Networking is the property of Mary Ann Liebert, Inc. and its content may not be copied or ed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or articles for individual use.