Television Exposure as a Risk Factor for Aggressive Behavior Among 3-Year-Old Children

Objective  To examine associations of child television (TV) exposure and household TV use with aggressive behavior among 3-year-old children while controlling for demographic characteristics and risk and protective factors for aggression.

Design  The Fragile Families and Child Wellbeing Study, a prospective cohort study.

Setting  Data collected at home and by telephone from parents of children born from 1998 to 2000 from 20 cities.

Participants  Mothers who completed a 36-month in-home survey and met inclusion criteria (n = 3128).

Main Exposure  Direct child TV exposure and household TV use were the primary explanatory variables. Additional risk factors included neighborhood disorder and maternal factors like depression.

Outcome Measures  Childhood aggression was assessed with the Child Behavior Checklist/2-3. Multivariate linear regression models were used to examine associations between TV measures, additional risk factors, and childhood aggression.

Results  Children who were spanked in the past month (β = 1.24, P < .001), lived in a disorderly neighborhood (β = 2.07, P < .001), and had a mother reporting depression (β = 0.92, P < .001) and parenting stress (β = 0.16, P < .001) were significantly more likely to exhibit aggressive behavior. Direct child TV exposure (β = 0.16, P < .001) and household TV use (β = 0.09, P < .001) were also significantly associated with childhood aggression, even when controlling for other factors.

Conclusions  Three-year-old children exposed to more TV, both directly and indirectly, are at increased risk for exhibiting aggressive behavior. Further research is essential to determine whether pediatric recommendations concerning TV and children should include limits for general household TV use.

Early childhood aggression can be problematic for parents, teachers, and childhood peers and sometimes is predictive of more serious behavior problems to come, such as juvenile delinquency, adulthood violence, and criminal behavior.^1-3 Predictors of childhood aggression include child sex, parenting factors like discipline style, and broader community factors like neighborhood safety.^4,5 Exposure to media is another factor that has been studied as a predictor of childhood aggression. After music, television (TV) is the medium children aged 0 to 3 years are exposed to the most.⁶ The American Academy of Pediatrics recommends no more than 2 hours of media time per day for children aged 3 years and older and no media for children aged 0 to 2 years.^7-9 Despite these recommendations, studies find consistent use of TV in children aged 2 years and younger.^6,10

Although many studies have found associations between media exposure and aggressive behavior in older children,¹¹ research conducted on preschool-aged children is less common.^12,13 Two recent studies found a significant association between child TV use and childhood aggression among preschoolers, one using TV time¹³ and the other incorporating TV content¹⁴; however, the latter study did not control for additional factors that may contribute to aggressive behavior,¹⁴ and the former study only controlled for a small number of factors, including maternal demographics and depression, child sex and health status, household income, and parental involvement in child activities.¹³ Other studies, including experiments, have also shown associations of TV use or viewing of aggressive programs at preschool age with aggressive behavior at both preschool age^15-18 and in grade school.^19-21 Television exposure may act directly to increase aggression by providing models for aggressive behavior²² or normalizing the behavior.²³ However, it is possible that TV exposure also influences behavior indirectly by affecting other child development issues, including problems with attention^24-26 and self-regulation,¹² irregular nap and bedtime schedules,²⁷ and language development.^28-30

Studies of TV use for preschool-aged children have all mainly focused on children's direct TV exposure with little attention to general household TV use; yet, this may be an important risk factor to consider given that nearly 40% of children live in homes characterized by heavy TV viewing.³¹ Household TV use may result in children's unregulated exposure to violent content. In accordance with displacement theory,^32,33 it is also possible that greater household TV use suggests that more family time is focused on TV rather than on other activities that could foster healthy child development, such as reading or playing. Thus, household TV use may have implications for child development and aggressive behavior.

Given the potential importance of household TV use and the need to account for multiple nonmedia factors that can influence aggression, the current study was designed to assess the unique contribution of household TV use to childhood aggression among 3-year-old children, while controlling for other key risk and protective factors that have not been examined previously in studies of media and aggression in this age group, such as spanking, maternal exposure to violence, and neighborhood disorder.

The Fragile Families and Child Wellbeing Study is a population-based study that includes data from parents of children born from 1998 to 2000 from 20 large US cities (baseline, N = 4898) and that oversampled unmarried parents. At the time of paper submission, public use data were available for interviews conducted at the time of the index child's birth and at 1 and 3 years. Further details on the sampling and design of the Fragile Families and Child Wellbeing Study are described elsewhere.³⁴

Mothers who completed the 36-month In-Home Longitudinal Study of Pre-School Aged Children survey (n = 3288) were eligible for inclusion in the current study. We excluded participants who (1) reported that their child did not live with them most of the time (n = 64), (2) missed more than 3 items on the externalizing (aggressive) behavior scale (n = 41), and (3) did not provide information about direct child TV exposure (n = 14) or household TV use (n = 41). Thus, the final sample size was 3128 mothers. All of the presented data were assessed during wave 3 of the survey (unless otherwise noted) and are based on maternal self-report in response to questions asked on both the Mother's Three-Year Follow-up Survey and the In-Home Longitudinal Study of Pre-School Aged Children survey. Child age at the time of the maternal interview ranged from 30 to 50 months (mean, 36 months). Child questions refer to the index child only.

Early childhood aggression was assessed using a 15-item aggressive subscale from the Child Behavior Checklist/2-3, which was designed to assess externalizing behaviors for 2- and 3-year-old children.^35,36 There were 3 possible response options for each behavior: “not true” (0), “somewhat or sometimes true” (1), or “very true or often true (2).” An aggression score was created by summing all item responses (α = 0.86), with a possible maximum of 30; this raw sum was used for the analysis.

Two questions were asked to assess direct child TV exposure: (1) “Think for a moment about a typical weekday for your family. How much time would you say [the index child] spends watching TV or watching videos on TV, either in your home or somewhere else?” (2) The same question was asked for a typical weekend day (Saturday or Sunday). The average number of hours (whole numbers) the child spent watching TV on a typical day was calculated using a method used in prior research²⁴: [(number of hours child watched TV on a typical weekday × 5) + (number of hours child watched TV on a typical weekend day × 2)] ÷ 7.

One question was asked about total household TV use: “About how many hours is a TV on in your home during a typical day?” Participants were instructed to include time the TV was on even when no one was watching it. This question was not asked separately for weekdays and weekend days. To avoid counting direct child TV exposure twice, we subtracted direct child TV exposure from the response to this question. For some cases (n = 179), this difference was a negative number, possibly owing to differences in question wording. We explored 3 options for handling these data (dropping these cases, leaving these data as negative numbers, or setting to 0). The results were the same in all cases, and we decided to set negative values to 0.

Covariates were chosen for inclusion based on prior empirical evidence suggesting that they were associated with childhood aggression and their inclusion in the Fragile Families and Child Wellbeing Study maternal interviews.

Maternal illness during pregnancy and maternal depression both raise the risk of childhood aggression.^37-42 Maternal health was assessed by asking, “In general, how is your health?” and responses were analyzed as excellent, very good, or good (0) or fair or poor (1). Maternal depression was assessed using section A of the short form of the Composite International Diagnostic Interview⁴³ and scored and coded based on criteria for major depression from the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (detailed scoring methods are described elsewhere)⁴⁴: no major depression (0) or major depression (1).

Hostile attributions and negative attitudes toward the child; low parental knowledge and monitoring^37,45,46; and hostile, overreactive, or too lax parenting and discipline styles^37,41,45-51 raise the risk for childhood aggression. Reading frequency and having a regular bedtime were included as proxies for positive maternal attitudes and involvement with the child. Reading frequency was assessed and recoded as number of days per week the parent read stories to the child: 3 or fewer days per week (0) or 4 to 7 days per week (1). Respondents were also asked if the child had a regular bedtime: no (0) and yes (1).

Three proxies for negative parenting attitudes and behaviors were also included. Aggravation in parenting was assessed with 4 items taken from prior surveys (eg, “Being a parent is harder than I thought it would be”). Response options ranged from “strongly disagree” (1) to “strongly agree” (4), and an average score was created.⁴⁴ Parenting stress was assessed using a shortened version of the Parenting Stress Index.⁵² Eleven items were rated on a scale with responses ranging from “strongly disagree” (0) to “strongly agree” (4) (α = 0.86; range, 0-44). The mother was also asked how often she had spanked the child in the past month for misbehaving or acting up: not at all in the past month (0) or 1 or more times in the past month (1).

Childhood exposure to interparental conflict and violence raises the risk for childhood aggression.^38,40,53,54 Mothers' reports of physical and psychological abuse by their current partner were assessed using 7 items. Three physical abuse items were adapted from the Conflict Tactics Scale for adults.⁵⁵ Four psychological abuse items were adapted from the Spouse Observation Checklist⁵⁶ and studies by Lloyd.⁵⁷ A trichotomous variable was created: no abuse reported (1); at least 1 abuse item occurred sometimes or often (2); or no current partner or missing data (3). Mothers who witnessed violence reported this experience using a subset of items adapted from the My Exposure to Violence Scale.³⁶ Among the 7 measured items, we chose to include in our analyses the 2 items with the greatest variation, which asked “In the past year, about how many times did you see someone else get: (1) hit, slapped, punched, or beaten up by someone? or (2) attacked with a weapon, like a knife or bat, by someone?” Responses were recoded to never (0) or 1 or more times (1).

Children raised in dangerous neighborhood environments are at increased risk for externalizing behaviors⁵⁸ and also more dangerous aggressive behaviors, such as weapon carrying, when they get older.^59,60 Neighborhood disorder was assessed using items adapted from the Neighborhood Environment for Children Ratings Scale.³⁶ The mother was asked to report how often each of 8 activities (eg, loitering, gang activity, and disorderly groups) occurred in the neighborhood. A categorical variable was derived: none of the activities ever occurred (0; average score = 1), some disorderly activity reported (1; average score >1 but ≤2), and more disorderly activity was reported (2; average score >2). If the respondent missed 2 or more of the 8 items, this variable was coded as missing.

Increased risk for childhood aggression is associated with many family demographic characteristics. Most studies find that boys have higher externalizing rates than girls, so we controlled for child sex.^45,47,49,61 Low socioeconomic and educational status also have been associated with early childhood aggression.^42,45,47 Therefore, given that socioeconomic and educational status is often associated with many other demographics, we also controlled for the mother's and father's age and educational level as well as several other maternal characteristics, including nativity, race, work time, and current relationship status. We also controlled for the following household characteristics: mother and father marital status at birth, household income, whether or not the household received food stamps in the past year, and whether or not there were additional adults or children (besides the mother and index child) living in the home.

We conducted bivariate analyses using t tests, analysis of variance, and Pearson correlation coefficient. Linear regressions were run to examine associations between TV variables and childhood aggression. We ran unadjusted models to compare direct child TV exposure and household TV use with aggression. We ran 5 multivariate regression models that controlled for potential explanatory variables, confounders, and covariates. All analyses were conducted using Stata, version 8.2 (Stata Corp, College Station, Texas). This study was approved by the University at Albany's institutional review board.

Despite American Academy of Pediatrics recommendations, about two-thirds (65%) of mothers reported that their index child watched more than 2 hours of TV per day (mean, 3.2 [SD, 2.3]; median, 3; range, 0-24 hours). After subtracting direct child TV exposure, there was an average of 5.2 hours of additional household TV use on a typical day (SD, 5.7; median, 3; range, 0-24 hours). The correlation between direct child TV exposure and additional household TV use was 0.07 (P < .001). The average childhood aggression score was 9.8 (median, 9; range, 0-30 [standard deviation, 5.9]). The correlation of childhood aggression was 0.15 (P < .001) with direct child TV exposure and 0.17 (P < .001) with additional household TV use.

Table 1 presents the demographic characteristics of the sample as well as the means of the 2 TV variables across these characteristics. Most of the characteristics had some level of statistical significance with direct child TV exposure and/or household TV use. Of note, mothers and fathers with less education, mothers not married to the father at birth, and mothers born in the United States reported significantly greater values for both child and household TV time. Table 2 presents the distribution of the risk and protective factors assessed in this study along with their bivariate associations with both TV variables. Several variables were strongly associated (P < .001) with higher levels of TV exposure for the child and for household TV use, including maternal depression, regular bedtime, mother witnessing violence, and neighborhood disorder.

Unadjusted regression analyses predicting childhood aggression at age 3 years found that when examined individually, direct child TV exposure (β = 0.39, P < .001) and household TV use (β = 0.17, P < .001) were each statistically associated with child aggression; the coefficients remained similar when both were included in the same model (child TV use, β = 0.36, P < .001; household TV use, β = 0.16, P < .001).

Multivariate regression findings are presented in Table 3. Models 1 and 2 included all demographic characteristics presented in Table 1 along with child TV exposure (model 1) and household TV use (model 2). Each TV variable was a strong predictor of aggression along with child sex. In models 3 and 4, we added risk and protective factors from each of the 4 covariate groups (maternal health, maternal parenting, maternal experience with violence, and neighborhood disorder) to models 1 and 2, respectively. Child TV exposure (model 3) and household TV use (model 4) each remained strong predictors of aggression. When child and household TV variables were included simultaneously in model 5, along with all other demographic and covariate variables, both remained statistically significant. This final model shows that, when holding all other measured factors constant, the predicted aggression score increases by 0.16 for each additional hour a child directly watches TV, and by 0.09 for each additional hour the TV is on in the household on a typical day. We found no threshold effects for the relationship between both TV variables and aggression score.

Children in the current study reportedly had more direct TV exposure (an average of about 3 hours total per day) than is recommended for young children, and prior studies have found direct TV exposure to be associated with childhood aggression, a finding confirmed in our study. However, no other studies we found have accounted for additional household TV use or for the full range of potential confounders and other covariates that we included.

In our study, we found the link between direct child TV exposure and childhood aggression to hold after accounting for a multitude of parent (ie, maternal health), family (ie, other children in the home), neighborhood (ie, neighborhood disorder), and demographic characteristics (ie, child sex). Without knowing the content of the programs viewed, we cannot say for certain whether it is violent TV content that causes children to behave more aggressively, as there is a lot of nonviolent educational content available on TV, which may be beneficial to children and is unlikely to lead to aggressive behavior.⁶² Yet even when children are primarily watching such programming, they may be exposed to violence through TV commercials⁶³ or even G-rated movies⁶⁴ according to findings from prior research. The concept of observational learning from social cognitive theory suggests that children learn to model aggressive behavior that they observe on TV and could explain aggressive behavior if the child is exposed to violent programming.²² Other theories suggest that extensive viewing of violent content causes desensitization to violence⁶⁵ and that heavy TV viewing of aggressive behavior can lead children to believe that it is a normal part of everyday life.²³

It is also possible that direct child TV exposure can alter child development in ways unrelated to content. For example, Schmidt and colleagues⁶⁶ found that although children aged 3 years and younger rarely looked at the TV when it was on in the background, they engaged in shorter play time and had limited attention to play activities, which negatively influences development.

In addition to direct child TV exposure, we found additional household TV use to be associated with child aggression even after accounting for direct child TV exposure and all other examined covariates. The low correlation between child TV exposure and household TV use suggests that these measures are assessing unique constructs. Although findings concerning the potential effects of household or background TV on children is limited,^28,67 prior studies do suggest possible mechanisms for this unique finding linking household TV use and child aggression.

One explanation that could link both child and household TV measures with aggression involves the parenting environment.⁶⁸ Given that media habits are pervasive throughout a household,⁶⁹ households with greater TV use have been found to have fewer restrictions on media use for children^18,70 and higher rates of media use among youth in the household.^31,71 Households with heavy TV use are also more likely to have fewer restrictions on other activities, such as having a regular bedtime,⁷⁰ as we found in our sample. Parents reporting high child TV viewing have also admitted to using the TV as a babysitter.¹⁰ Heavy household TV use, then, may be an indicator of general attitudes concerning discipline and rules regarding media habits and child rearing in general.

Another potential explanation concerns how family interactions may be influenced by household TV use, including daily routines such as eating and communication patterns.⁶⁹ Displacement theory, which suggests that time spent watching TV decreases time spent doing other activities,^32,33,72 might explain how heavy TV use in a household may reduce the amount of educational and interactive activities that occur in the home (ie, toy play).^28,32 For instance, a recent study found that background TV negatively affects the quantity and quality of interactions between parents and children,⁷³ and TV viewing by young children is characterized by less frequent interactions with parents.⁷⁴ Parents who are focused on TV programs may pay less attention to their children, be less engaged in play,⁶⁷ and/or may react negatively toward the children if their TV viewing is interrupted. Although 1 study found that when parents and children viewed TV together, they simultaneously engaged in other activities, such as reading,⁷⁵ it is possible that the quality of these interactions was lower than it would have been without the TV on. In our study, mothers who reported more frequent reading with their children also reported fewer hours of household TV use. By replacing valuable opportunities for communication and other activities, household TV use may limit prosocial development.

It also is possible that increased exposure to unregulated TV content may explain the link between household TV use and child aggression. Coviewing with parents is more typical when children are watching programs parents want to view.⁷⁶ Given the high rates of violent content on TV,^77,78 it is likely that the child would be exposed to violent content while coviewing.

There are several limitations to this study. First, it is limited by the cross-sectional nature of the data. Although a longitudinal analysis of these associations was not yet possible with the publicly available data from the Fragile Families and Child Wellbeing Study at the time our paper was submitted,⁷⁹ our next step will include a longitudinal analysis using wave 5 Child Behavior Checklist data.

Second, the TV variables only accounted for amount of exposure; however, quality and content of TV exposure may vary widely. Many parents believe that some TV can be educational for young children.⁶ Without measures of content, we cannot be sure what messages children were exposed to. Although content measures of media use have been found to be better predictors of aggression than time measures of media use,¹¹ other studies of media effects on children have relied solely on time measures.^{13,24,25,27,28}

Third, measurement bias may exist, as all data in this study were self-reported by mothers. However, parental reports are moderately correlated with other methods of data collection, such as observation.⁸⁰ Also, these findings might reflect a social desirability bias in respondent answers to questions about direct child TV exposure given that American Academy of Pediatrics recommendations for parents to limit media time for children have been highly publicized. In this case, household TV use may simply provide a more accurate measure of child TV exposure than direct questions about child TV viewing. Measurement of TV time is also limited by the simple nature of the questions asked (ie, only number of hours per day). Finally, no other types of mass media exposure were assessed in the Fragile Families and Child Wellbeing Study.

The American Academy of Pediatrics recommendations regarding child media time are based in part on the idea that high quantities of media exposure, including TV, are likely to interfere with parent-child interactions and also on research that suggests media content can have a negative influence on a variety of health behaviors, especially violence.^7,9,29 Current American Academy of Pediatrics recommendations mainly suggest limitations for direct child exposure to TV and other media; however, our findings suggest that additional household TV use may also be an important predictor of negative childhood outcomes, such as early childhood aggression. Future research in this area should consider inclusion of both of these TV variables along with additional parent-child interaction assessments, observational assessments when possible, quality and/or content of TV programs, and longitudinal analyses.

Correspondence: Jennifer A. Manganello, PhD, MPH, School of Public Health, University at Albany, State University of New York, Department of Health Policy, Management & Behavior, One University Place, Rensselaer, NY 12144 (jmanganello@albany.edu).

Accepted for Publication: April 30, 2009.

Author Contributions: Drs Manganello and Taylor both had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Manganello and Taylor. Analysis and interpretation of data: Manganello and Taylor. Drafting of the manuscript: Manganello. Critical revision of the manuscript for important intellectual content: Manganello and Taylor. Statistical analysis: Manganello. Study supervision: Manganello.

Financial Disclosure: None reported.

Additional Contributions: Benjamin Shaw, PhD, provided editorial comments and statistical advice concerning this manuscript and Ransome Eke, MBBS, MPH, FRIPH, assisted in conducting a literature review. We would also like to thank the anonymous reviewers at Archives of Pediatrics & Adolescent Medicine for their thoughtful comments and suggestions.