OBJECTIVE. The effect of violent television programming on preschoolers’ behaviors is poorly understood. The objective of this study was to test the hypothesis that exposure to violent television viewing when children are 2 to 5 years of age would be associated with antisocial behavior at ages 7 to 10.
METHODS. Data were derived from the Panel Study of Income Dynamics. Our primary outcome was being in the 88th percentile of the Behavioral Problem Index antisocial subdomain. Our primary predictor was exposure to violent screen content.
RESULTS. Data were available for 184 boys and 146 girls at both time periods. Adjusting for baseline Behavioral Problem Index scores and age, parental education, maternal depression, and cognitive and emotional support, violent television programming was associated with an increased risk for antisocial behavior for boys but not for girls. Neither educational nor nonviolent programming was associated with increased risk for boys or girls.
CONCLUSIONS. Viewing of violent programming by preschool boys is associated with subsequent aggressive behavior. Modifying the content that is viewed by young children may be warranted.
Children who are younger than 6 years are reported to experience an average of ∼2 hours of screen time each day, with approximately half of that watching television and the remainder divided between DVD/videos, computer use, and video games.1,2 Many argue that more important than how much they watch is what they watch.3–5 Unfortunately, 95% of children watch programs that are not specifically produced for young audiences6; however, even commercial television programming that is designed specifically for children can still represent a substantial risk to young children, especially with respect to violence and aggression.7 In fact, the level of violence in commercially aired programs that are intended for children exceeds the level in adult programs, including comedies, dramas, and even music videos and reality shows.7 G-rated films for children are no better: every single G-rated film that was released to theaters in the United States up to 1999 contains violence, and half show at least 1 character rejoicing in violence by cheering or laughing.8
Cross-sectional and quasi-experimental studies of television viewing among school-aged children and adolescents have found it to be associated with aggression.9–11 Experimental designs have confirmed that reducing television can reduce aggression among 9-year-olds.12,13 Considerably less attention has been given to the effects of television on preschool children.11 Although a few longitudinal studies of television viewing before age 5 have shown it to be a potential risk factor for the subsequent development of bullying and aggression measured in early elementary school,14,15 these studies did not distinguish between types of programs but rather used overall television viewing as their exposure of interest. Among media researchers, the evolving consensus is that content is a critical mediator of the effects of television on children, but data in support of this for young children are sparse.16,17 We therefore conducted a study to test the hypothesis that exposure to violent programming during the preschool period would be associated with subsequent aggressive behavior.
We conducted a longitudinal study using data from the Panel Study of Income Dynamics (PSID). The study protocol was approved by the Children's Hospital and Regional Medical Center Institutional Review Board.
The PSID is a longitudinal study that was begun with 4800 families in 1968 with a variety of funding sources and overseen by the National Science Foundation. In 1997, a Child Development Supplement (CDS), which is a questionnaire administered to the primary caregivers of 3563 children aged 0 to 12, was added. The recruiting, eligibility, and attrition of the PSID and CDS have been described elsewhere.18,19
The questionnaire, funded by the National Institute of Child Health and Human Development, included detailed demographic data, psychological and behavioral assessment of parents and children, and time use diary data from 1 randomly chosen weekday and 1 randomly chosen weekend day during a school year (September through May). These diaries include primary and secondary activities during a 24-hour period. Such time diaries have been used extensively in research and have excellent validity when compared with direct observation of activities.20,21 Among eligible households in the core PSID sample, 1997 CDS data were obtained for 88%. In 2002, the respondents of the first CDS were followed up with a second, similar CDS (CDS-II). The follow-up rate from the 1997 CDS to the 2002 CDS was 91%.
We included all children who were between the ages of 24 and 60 months in 1997 and who had follow-up data 5 years later (ie, when the were 7–9 years of age).
As part of the PSID survey, parents complete the Behavioral Problem Index (BPI) which is derived from questions from the Achenbach Child Behavior Checklist. A subdomain of the BPI assesses antisocial behavior, which has been well validated with Cronbach's α of .66.22 The following statements are included in this subdomain: My child cheats, is mean to others, feels no regret, is destructive, is disobedient at school, and has trouble with teachers. Each statement had 3 Likert-type answers ranging from 1 for “not true” to 3 for “often true.” These raw scores were averaged and then dichotomized at 1.4, which classified 12.2% of children as having problems with antisocial behavior. This cutoff was used in an effort to approximate the 90th percentile that has been used in other studies that have used the BPI.23,24 Children at that percentile on the entire BPI have been previously found to be at risk for mental health referrals.25
We used content data for programming that was derived from the time diaries. A coding system was developed previously by others to classify television shows and movies on video (hereafter both referred to as “shows”) by content in several dimensions.26 An educational attribute was assigned when the show had a clear intent to educate, with an explicit cognitive or prosocial component according to the following criteria: (1) the program teaches a lesson with content similar to that found in schools (eg, math skills, reading skills, other school readiness skills), and (2) the program teaches a lesson about appropriate behavior or interpersonal interactions (eg, sharing, friendships, drug education). Violent content was ascribed when “violence was a central and integral part of the plot or of the main characters’ occupations, if the lead characters’ main purpose was to fight or flee from violence, or if there was more violence in the program than would be expected in the everyday life of a child.”27 Although the term “violent” is used here, it should be understood that the definition includes hostile language, threatening behavior, and cartoon violence as well as realistic violence. A minimum of 2 coders evaluated each show for violence and educational content, with a interrater κ score of 0.81.19,27 Differences were resolved by discussion. For this study, we classified shows into 3 categories: educational, nonviolent entertainment (ie, not violent and not educational), and violent entertainment (ie, violent and not educational). No educational shows contained violence.
Some shows could not be coded for violence, either because the name was inadequately reported (eg, “cartoons,” “Channel 13”) or because the researchers could not evaluate the violent content of an uncommon but named video or show. These shows, 20% of the total, were included in the “nonviolent entertainment” category. Examples of selected most popular shows are presented in Table 1. Television viewing was modeled as a continuous variable as number of hours on a typical day.
We adjusted for the following covariates: race and ethnicity, gender, age at survey completion, maternal and paternal education (defined as highest grade completed at the baseline survey); paternal presence in the household; maternal depression as measured by the Center for Epidemiologic Studies Depression Scale; and emotional and cognitive stimulation at the baseline point as measured by the Home Observation for Measurement of the Environment scale,28 which has been used extensively in studies measuring household function. We also adjusted for antisocial behavior at the baseline period using the same subdomain of the BPI that we used as our outcome variable at the follow-up period. Finally, we adjusted for harsh physical punishment and parental coping in the baseline period. Physical punishment was measured as retrospective parental report of whether they spanked their child before age 2 or whether they spanked their child after age 2, with “never spanked” as the reference category. We also included the Parental Coping Problems scale, which included 5 statements:
[Child] seems to be harder to care for than most children.
There are some things that [he/she] does that really bother me a lot.
I find myself giving up more of my life to meet [child's] needs than I ever expected.
I often feel angry with [child].
I would be doing better in my life without [child].
Higher scores indicate poorer coping.
We used logistic regression to determine the independent association between types of programming at baseline and antisocial behavior at follow-up. We formally tested for effect modification of gender on content by testing an interaction term gender × content in our model. Because the P value associated with this term was significant, we stratified our results on the basis of gender. Appropriate sampling weights were used to provide nationally representative samples. We also conducted locally weighted smoothing regression (Lowess) to plot the bivariate association between early violent television exposure and the probability of demonstrating antisocial behavior at follow-up.29,30 All analyses were conducted by using Stata 9.0 (Stata Corp, College Station, TX).
Data were available for 184 boys and 146 girls at both time periods, 54% of the children in age range in the total sample during this interval. Those with missing data were most commonly missing the measurement of antisocial behavior at baseline, which was not regularly collected for those who were younger than 3 years at baseline. Accordingly, only 4% of the estimation sample was younger than 36 months, as opposed to 35% of the full sample. The mean age of study participants at the baseline period was 49 months; 44% were female. Complete demographic data are presented in Table 2.
Adjusting for all covariates concurrently, violent television programming was associated with an increased risk for antisocial behavior overall (odds ratio [OR]: 2.20; 95% confidence interval [CI]: 1.35–3.60). In the stratified analysis, however, the association was present for boys (OR: 4.10; 95% CI: 2.09–8.02) but not girls (OR: 0.39; 95% CI: 0.04–3.74). Nonviolent programming was not associated with subsequent antisocial behavior in boys (OR: 1.76; 95% CI: 0.64–4.87) or girls (OR: 1.38; 95% CI: 0.68–2.83). Educational programming also was not associated with increased risk for boys (OR: 0.41; 95% CI: 0.09–1.86) or girls (OR: 0.63; 95% CI: 0.11–3.73). Fully adjusted regression models are presented in Table 3. The adjusted results for all of the television programming types are presented in Figure 1. Removal of the single outlier in Figure 2 who watched 5 hours of violent television daily attenuated but did not eliminate the association between watching violent programming and antisocial behavior (OR: 3.27; 95% CI: 1.45–7.39).
We found that watching violent entertainment programming at age of 2 to 4 years was associated with significantly increased risk for antisocial behavior at ages 7 to 9 for boys but not for girls. Furthermore, we found no significant effect of other types of programming on antisocial behavior for either boys or girls during the same period. Our findings confirm and extend findings of others that exposure to violence on screen can promote aggression in real life. The differential effects of content, particularly that the point estimates for educational programming were in the direction of a protective effect, are important in that they suggest that alternative programming types could offer behavioral benefits to children without necessarily reducing overall viewing time. Others have found that select programming can promote prosocial behaviors in preschool children.31–33
It is also interesting that we found an association for boys and not for girls. This effect modification may be attributable to socialization differences between genders, genetic predispositions toward aggression, or perhaps even to the selection of programs because the shows that they watch are not identical. It is not possible in this data set to tease out the marginal contribution of each of these factors.
These results are important because aggressive behavior in the early childhood years has been repeatedly linked to violence in later youth and adolescence.34–37 This later youth violence does not materialize suddenly from nowhere but rather develops as a continuum with escalating aggression.38–42 Although minor aggressive behaviors are present from infancy, peer-directed aggression begins during the toddler and preschool years, and the onset of real physical fighting and interpersonal violence occurs later during youth and adolescence. The toddler and preschool years constitute the time during which most children learn to use nonaggressive alternatives preferentially over aggressive behavior; when that does not occur, young children can continue on a trajectory of aggression.43
Our findings must be interpreted in light of several limitations. First, the observational nature of our study precludes definitive establishment of a causal relationship. However, that we were able to control for antisocial behavior at the baseline period in our regression analyses and that the hypothesis tested is based on experimentally confirmed data in older children lend support to the possibility that in fact these associations are causal. Second, we used time diaries to determine shows viewed. Although imperfect, such diaries have been shown to be valid.44 Third, our cutoff for antisocial problems at the 88th percentile is somewhat arbitrary; however, it is consistent with the 90th percentile that others have used. We were lacking data on some particular shows when parents merely reported that their child was watching cartoons. Our procedure to classify all such shows as nonviolent entertainment might bias our findings toward findings of aggression effects for such programming, because a great deal of cartoons are in fact violent; however, by not classifying them as violent, we should not have biased our estimate for violent programming. Despite these limitations, our results suggest that modification of the media diet during the preschool years may have long-term effects on aggressive behavior in children.
- Accepted May 30, 2007.
- Address correspondence to Dimitri A. Christakis, MD, MPH, Child Health Institute, 6200 NE 74th St, Suite 210, Seattle, WA 98115-8160. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Rideout VJ, Vandewater EA, Wartella EA. Zero to Six: Electronic Media in the Lives of Infants, Toddlers, and Preschoolers. Menlo Park, CA: Kaiser Family Foundation; 2003
- ↵American Academy of Pediatrics, Committee on Public Education. Media violence. Pediatrics.2001;108 :1222– 1226
- American Academy of Pediatrics, Committee on Public Education. Children, adolescents, and television. Pediatrics.2001;107 :423– 426
- ↵Christakis DA, Zimmerman FJ. The Elephant in the Living Room: Make Television Work for Your Kids. Emmaus, PA: Rodale; 2006
- ↵Comstock G, Paik H. Television and the American Child. San Diego, CA: Academic Press; 1991
- ↵National Committee on Television Violence. National Television Violence Study. Vol 3. Thousand Oaks, CA: Sage; 1998
- ↵Johnson JG, Cohen P, Smailes EM, Kasen S, Brook JS. Television viewing and aggressive behavior during adolescence and adulthood. Science.2002;295 :2468– 2471
- ↵National Research Council and Institute of Medicine. Studying Media Effects on Children and Youth: Improving Methods and Measures. Washington, DC: National Academy of Sciences; 2006
- ↵Anderson CA, Berkowitz L, Donnerstein E, et al. The influence of media violence on youth. Psychological Science in the Public Interest.2003;4 :81– 110
- ↵Modi M. Panel Study of Income Dynamics, Child Development Supplement: surveys measuring well-being; 2000. Available at: www.nber.org/∼kling/surveys/PSID.html. Accessed May 1, 2007
- ↵Vandewater EA, Bickham DS, Lee JH. Time well spent? Relating television use to children's free-time activities. Pediatrics.2006;117 (2). Available at: www.pediatrics.org/cgi/content/full/117/2/e181
- ↵Zill N. Behavior Problem Scales Developed From the 1981 Child Health Supplement to the National Health Interview Survey. Washington, DC: Child Trends Inc; 1985
- ↵Byrd RS, Weitzman M, Auinger P. Increased behavior problems associated with delayed school entry and delayed school progress. Pediatrics.1997;100 :654– 661
- ↵Byrd RS, Weitzman M, Lanphear NE, Auinger P. Bed-wetting in US children: epidemiology and related behavior problems. Pediatrics.1996;98 :414– 419
- ↵Gortmaker SL, Walker DK, Weitzman M, Sobol AM. Chronic conditions, socioeconomic risks, and behavioral problems in children and adolescents. Pediatrics.1990;85 :267– 276
- ↵Vandewater EA, Lee JH, Shim M. Family conflict and violent electronic media use in school-age children. Media Psychol.2004;7 :73– 86
- ↵Caldwell BM, Bradley RH. Home Observation for Measurement of the Environment. Little Rock, AR; University of Arkansas: 1984
- ↵Fox J. Applied Regression Analysis, Linear Models, and Related Methods. Thousand Oaks, CA: Sage; 1997
- ↵Härdle W, Müller M, Sperlich S, Werwatz A. Nonparametric and Semiparametric Models. New York, NY: Springer; 2004
- ↵Silverman LT, Sprafkin JN. The effects of Sesame Street's prosocial spots on cooperative play between young children. J Broadcast.1980;24 :135– 147
- Fraser MW. Aggressive behavior in childhood and early adolescence: an ecological-developmental perspective on youth violence. Soc Work.1996;41 :347– 361
- ↵Tremblay RE, Nagin DS, Seguin JR, et al. Physical aggression during early childhood: trajectories and predictors. Pediatrics.2004;114 (1). Available at: www.pediatrics.org/cgi/content/full/114/1/e43
- ↵Robinson JP, Godbey G. Time for Life: The Surprising Ways Americans Use Their Time. 2nd ed. University Park, PA: Pennsylvania State University Press; 2000
- Copyright © 2007 by the American Academy of Pediatrics