Video-Sharing Platform Viewing Among Preschool-Aged Children: Differences by Child Characteristics and Contextual Factors

Abstract

Free video-sharing platforms such as YouTube are highly popular among young children but may contain low-quality and highly commercialized content. This study aimed to describe the prevalence, duration, and timing of objectively measured mobile YouTube viewing in preschool-aged children and test hypotheses about associations with child individual differences and contextual factors. We analyzed mobile sampling data from 349 English-speaking children aged 3–4.99 years whose parents completed surveys about child, parent, and household characteristics. We assessed whether the child ever viewed YouTube during the sampling week and calculated average daily duration in a subsample of 121 participants with their own mobile devices. We built multivariable logistic regression models to test correlates of mobile YouTube viewership and duration. Children were 3.82 years (SD 0.53), 74.6 percent white non-Hispanic; parents were mostly mothers (93.7 percent), 34.0 (SD 4.6) years, and 37.9 percent had less than a college degree. Mobile YouTube viewing (37.0 percent of children) was more likely in children who used Android devices, shared mobile devices with family members, were older, attended home-based childcare, or had parents with lower educational attainment. Median YouTube duration was 61.2 min/day, with longer durations in children whose parents had lower educational attainment. These results demonstrate that many young children use free video-sharing platforms on mobile devices for long durations, and this practice may be disproportionately higher in children from lower socioeconomic status. Longitudinal research is needed on video-sharing platform viewing and child outcomes.

Introduction

YouTube is a free video-sharing platforms widely used by both children and adults.^1–3 The Pew Research Center found that 81 percent of parents with children younger than 12 years of age allow them to watch YouTube, with 34 percent reporting that their child does so regularly.² Children 8 years of age and younger now spend more time per day on video platforms such as YouTube than they do watching television (TV).³ YouTube content has grown exponentially over the past decade; about 1 billion hours of content are viewed each day, about two-thirds on mobile devices.⁴

Although YouTube was not intended for child viewers, videos geared toward children are highly popular on the main site and a 2019 content analysis found that videos featuring children averaged three times more views than those who did not.⁵ Despite increasing child access to YouTube via smartphones and tablets,^6,7 little has been published about how young children use this platform.

Compared to TV and subscription-based video streaming services, content on YouTube has important differences. YouTube contains a variety of user-generated content that includes how-to arts and crafts tutorials, science and nature content, music videos, nursery rhyme cartoons, branded content (e.g., unboxing videos), family video logs, and “Let's Play” video gaming videos. On the main YouTube site, children also have access to adult-directed, violent, or developmentally inappropriate content. In recent years, YouTube has been criticized for insufficient content moderation,⁸ algorithms that promote prolonged viewing,^2,9 child data collection,¹⁰ and child-directed influencer marketing.^11,12

Supporting these concerns, a recent Common Sense Media analysis of the YouTube viewing histories of 183 children aged 8 and younger found a high prevalence of advertising (85 percent of videos) and commercial content (45 percent of videos), with age-inappropriate features such as violence in about one-quarter of videos.¹³ However, that study did not measure the duration of viewing and did not examine whether children who watch YouTube differ from their peers. As more children access video-sharing platforms that were originally designed for adult viewers, it is important to identify which children use such platforms, their usage patterns, and whether disparities in usage exist.

The socioecological model of child media use^14,15 posits that children form unique media habits based on their individual differences, family context, and broader sociocultural factors. At the child level, child self-regulation difficulties (e.g., intense emotional reactions, self-soothing challenges) and externalizing behavior (e.g., aggression, tantrums) have been linked with longer daily media use,¹⁶ age-inappropriate content,¹⁷ use of media for calming purposes,¹⁸ and longer use of mobile apps.¹⁹

No published studies have examined these factors with regard to YouTube use. One prior survey of parents of 18–36-month-olds found that YouTube viewing was positively correlated with use of screens to occupy the child and at bedtime,²⁰ which are common practices with other forms of media.^21,22 In addition, as children grow older they form stronger media preferences that shape usage habits,²³ which may apply to YouTube as well.

Prior work has suggested that parent-level factors such as depression and parenting stress influence child media use,^24,25 as do household factors such as socioeconomic status (SES).⁷ In addition, children from minoritized race/ethnicity have been found to consume more screen media in prior studies,^26,27 which may be due to a range of structural factors. We therefore sought to examine how different aspects of children's social context might drive YouTube usage—particularly because this free platform offers content that higher-SES parents might eschew in preference for subscription-based alternatives.

Finally, we sought to examine how device-level affordances might predict mobile YouTube viewing. Specifically, the YouTube app comes preinstalled on Android devices—and prior research supports that child users often follow default settings²⁸—yet, must be downloaded onto iOS devices in a separate step. Children's ownership of devices might also influence YouTube habits, as naturalistic²⁹ and lab-based³⁰ studies suggest that parents have more difficulty monitoring media when children are given access to their own mobile device.

To address these gaps in knowledge, this study leveraged objective mobile device data from a cohort study of young children to (a) describe mobile YouTube viewing prevalence and duration, and (b) test hypotheses about correlates of mobile YouTube viewing:

H1 (child-level): YouTube viewing will be more common in older children and those with behavioral and sleep difficulties; H2 (parent-level): YouTube viewing will be more common in families with higher parent stress and depression symptoms; H3 (contextual factors): YouTube viewing will be more common in children from lower income and lower education households, minoritized race/ethnicity, and with less access to regular childcare; and H4 (technology-level): YouTube viewing will be more common among Android versus Apple users, and among children who have their own mobile device versus share with family members.

Materials and Methods

Study design

We analyzed data from the baseline wave (2018–2019) of the Preschooler Tablet Study, a longitudinal study of early childhood development and mobile media use (NICHD R21HD094051). In this study, caregivers of 3–4-year-olds completed online questionnaires and provided a 1-week sample of their child's mobile device usage data. The study was approved by the University of Michigan Institutional Review Board. Parents received a $40 gift card after completing the first data collection wave.

Participants

The study population and recruitment methods have been described in detail previously.³¹ In brief, we enrolled parents from the community (e.g., university research registry; flyers in pediatric offices; social media advertisements; distribution among Head Start teachers; and early education providers) who were (a) legal guardian of a 3–4.99-year-old child; (b) lived with the child at least 5 days per week; (c) understood English sufficiently to complete questionnaires and provide consent; and (d) owned an Android or Apple tablet or smartphone. Exclusion criteria included presence of child developmental delays or use of psychotropic medication.

After providing online consent for themselves and their child, parents were emailed mobile sampling instructions and online REDCap^32,33 surveys. Of 423 parents who submitted any data, we included 349 children with complete mobile device data in this analysis; characteristics of included versus excluded children were not significantly different.

Assessment of mobile YouTube viewing

Children's mobile device usage was sampled using novel objective methods that harness app usage logs already collected by devices.³⁴ Participants with Android smartphones were instructed on how to install the Chronicle app (OpenLattice, Inc.), which queries the Google app usage statistics API to provide a continuous, timestamped log of app usage. After 9 days, Chronicle data were downloaded and processed using Python and R. For participants with iOS devices, which did not allow app logging, parents were instructed on taking screen shots of the battery feature (under “Settings”), which provides summaries of average daily app usage. Screen shot data were transferred into Excel files for analysis. Our laboratory validated the accuracy of this approach through comparison against handwritten smartphone usage logs.

In the full cohort, we categorized children as mobile YouTube viewers if YouTube appeared in their sampling data (and in the case of shared devices, parent indicated use). To calculate average duration of daily YouTube usage and proportion of overall usage, we restricted the cohort to children who had their own mobile device (n = 121). This was because we presumed that parents or siblings may have also used these apps on shared devices, whose output would therefore provide an overestimate of true usage.

Child-level variables

Parents reported their child's age and usual sleep patterns, including bedtime, wake up time, and sleep latency, which we dichotomized into <30 minutes versus 30+ minutes. Parents completed the Emotional Reactivity subscale of the Child Behavior Checklist-Preschool,³⁵ a valid measure of child behavior for 18 months to 5 years. This subscale sums nine symptoms (e.g., “rapid shifts between sadness and excitement,” “demands must be met immediately,”) rated on a 3-point scale from 0 (not true) to 2 (very or often true). Parents also reported child's sex, prematurity, and whether they were an only child, which were examined as possible confounders.

Parent-level variables

Parents completed the Centers for Epidemiologic Studies-Depression Scale (CES-D),³⁶ a widely-used and validated measure of depression symptoms that asks parents to rate frequency of items such as “I was bothered by things that usually don't bother me” from 0 (rarely) to 3 (all of the time). They also completed the Parenting Stress Index-Short Form,³⁷ a validated measure of parent stress, in which parents rate agreement with items such as “I feel trapped by my responsibilities as a parent” on a 5-point Likert scale. Parents also reported their own age, sex, marital status, and employment status, which were examined as possible confounders.

Contextual factor variables

As measures of household SES, we used parent-reported educational attainment, household income, and size (from which we calculated income-to-needs ratio). Parents also reported their child's race/ethnicity (collapsed into white non-Hispanic vs. underrepresented minority due to small cell sizes) and indicated whether their child attended center-based childcare/preschool, a home-based program, or no childcare/preschool.

Device-level variables

Device operating system and child device ownership were collected at enrollment; 129 children used Android devices (28.7 percent child's own device and 71.3 percent shared) and 220 used iOS devices (38.2 percent child's own device and 61.8 percent shared). Overall, n = 121 (34.7 percent) children had their own mobile device.

Data analysis

We used chi-square tests and Wilcoxon rank-sum tests to examine associations between predictors, potential covariates, and YouTube viewership. We then built multivariable logistic regression models, estimating the odds of being a mobile YouTube viewer, testing all technology-level, child-level, parent-level, and contextual factors that had shown associations at p < 0.20 in bivariate tests. All hypothesized variables and covariates were included in the same model and we conducted backward elimination to retain only variables with a p value <0.05. Although child race/ethnicity p value was .09, we retained it in the model because its removal led to a large change in the magnitude of other effect estimates.

In the 121 children with their own mobile devices, daily YouTube duration variables were skewed with many zero values; we therefore created a 3-category variable to distinguish children with high use (1+ h/day) from those with moderate use (<1 h/day) versus 0 min/day. In these analyses, due to small cell sizes, the parent education variable was dichotomized into less than a 4-year college degree versus 4-year degree or higher. We then built multivariable logistic regression models, estimating the odds of high and moderate YouTube duration. As above, we included variables associated at p < 0.20 and conducted backward elimination to retain variables with a p value <0.05.

For exploratory descriptions of YouTube usage patterns, we plotted average hourly data from children with their own Android mobile device (n = 37), which provide timestamps of daily usage (unlike iOS output, which only provides summary app usage data). We also calculated how many children used YouTube overnight by identifying any timestamps within the hour of the child's parent-reported bedtime through 5:00 a.m.

Results

Children in the analytic sample were 3.82 years old (SD 0.53), 48.7 percent were female, 74.6 percent were white non-Hispanic, 8.4 percent multiple races, 7.8 percent Hispanic (any race), 4.9 percent lack, non-Hispanic, 2.5 percent Asian or Pacific Islander, and 1.5 percent Native American or Alaskan Native. Most children attended childcare (65.2 percent center-based and 7.7 percent home-based). Parent respondents were mostly mothers (93.7 percent), were on average 34.0 (SD 4.6) years old, married or cohabitating with a partner (90.5 percent); and most had graduated from college (4-year degree 26.7 percent and advanced degree 35.5 percent). Families had a wide range of income-to-needs ratio (Table 1).

Table 1.

Participant Demographic Characteristics and Associations with YouTube Viewership

Characteristic	Overall (n = 349), mean (SD) or n (%)	YT viewers (n = 129), mean (SD) or n (%)	YT nonviewers (n = 220), mean (SD) or n (%)	Bivariate test p value	aORa (95% CI) for YT viewership
Hypothesis 1: Child-level factors and potential covariates
Child age (years)	3.82 (0.53)	3.87 (0.56)	3.79 (0.51)	0.185	1.76 (1.07–2.88)
Child sex female	170 (48.7)	61 (47.3)	109 (49.6)	0.683	—
Average sleep duration (hours/night)	10.8 (0.78)	10.8 (0.83)	10.8 (0.74)	0.884	—
Long sleep latency (>30 minutes)	84 (24.1)	27 (20.9)	57 (25.9)	0.303	—
CBCL-P Emotional Reactivity subscale	3.80 (2.93)	3.65 (3.03)	3.88 (2.88)	0.278	—
Only child	63 (18.1)	24 (18.6)	39 (17.7)	0.837	—
Hypothesis 2: Parent-level factors and potential covariates
Parent age (years)	34.0 (4.6)	34.0 (4.9)	34.1 (4.5)	0.722	—
Parent sex female	327 (93.7)	120 (93.0)	207 (94.1)	0.692	—
Parent married/has partner	316 (90.8)	112 (87.5)	204 (92.7)	0.103	NS
CES-D score	9.63 (9.51)	9.5 (8.4)	9.7 (10.1)	0.679	—
PSI-SF score	45.3 (33.1)	46.2 (33.2)	44.7 (33.0)	0.710	—
Parent employment
Not employed/stays at home	98 (28.1)	44 (34.1)	54 (24.6)	0.269	—
Part-time job	69 (19.8)	25 (19.4)	44 (20.0)
Full-time job	160 (45.9)	53 (41.1)	107 (48.6)
Multiple jobs	22 (6.3)	7 (5.4)	15 (6.8)
Hypothesis 3: Contextual factors
Parent education
High school/GED or less	25 (7.2)	19 (14.7)	6 (2.7)	<0.0001	9.12 (2.77–30.0)
Some college/2-year degree	107 (30.7)	49 (38.0)	58 (26.4)		2.02 (1.08–3.78)
4-year college degree	93 (26.7)	27 (20.9)	66 (30.0)		1.03 (0.53–1.99)
Advanced degree	124 (35.5)	34 (26.4)	90 (40.9)		1.00
Household income-to-needs ratio	3.00 (1.72)	2.57 (1.70)	3.25 (1.68)	0.0002	NS
Child race/ethnicity
White non-Hispanic	259 (74.6)	87 (68.0)	172 (78.5)	0.029	1.00
Underrepresented minority	88 (25.4)	41 (32.0)	47 (21.5)		1.61 (0.92–2.81)
Childcare type
Center-based	219 (65.2)	71 (56.4)	148 (70.5)	0.023	1.00
Home-based	26 (7.7)	14 (11.1)	12 (5.7)		3.27 (1.30–8.26)
Stays home with caregiver	91 (27.1)	41 (32.5)	50 (23.8)		1.77 (0.98–3.22)
Hypothesis 4: Device-level factors
Child mobile device ownership
Shares with parent/sibling	228 (65.3)	98 (78.0)	130 (59.1)	0.0014	2.57 (1.45–4.54)
Has own device	121 (34.7)	31 (24.0)	90 (40.9)		1.00
Operating system
Android	129 (34.0)	71 (55.0)	58 (26.4)	<0.0001	2.62 (1.58–4.36)
iOS	220 (63.0)	58 (45.0)	162 (73.6)		1.00

^a aOR are based on the multivariable model, including covariates, for which aOR are shown.

aOR, adjusted odds ratio; CBCL, Child Behavior Checklist; CES-D, Centers for Epidemiologic Studies-Depression scale; GED, general equivalency diploma; ITN, income-to-needs ratio; NS, Not significant (p value >0.10 in multivariable model; not included in final model); PSI-SF, Parenting Stress Index-Short Form; SD, standard deviation; YT, YouTube.

Descriptive results: mobile YouTube viewing

Of the 349 participants, 129 (37.0 percent) were YouTube viewers (of note, a smaller number were YouTube Kids viewers, the child-directed version of YouTube, n = 94, 26.9 percent; only 16 children used both platforms). Of the 121 participants with their own mobile devices, 31 were YouTube viewers, whose median daily YouTube viewing duration was 61.2 min/day (interquartile range [IQR] 2.1–165.7, range 0.003–390.1), accounting for 47.3 percent (IQR 7.3–88.3 percent) of their average daily use.

In exploratory analyses of 37 children who had their own Android devices, plots of average hourly use of YouTube are shown in Figure 1, showing peak duration at 11:00 a.m., 6:00 p.m., and 12:00 a.m. Nine (60 percent) YouTube users watched it during the bedtime/overnight hours.

FIG. 1.

Average hourly* viewing of YouTube in 15 children with their own Android mobile devices. *Average minutes/hour calculated from viewing duration only on days when YouTube was used. YT, YouTube.

Correlates of mobile YouTube viewership

H1 (Child-level)

Hypothesis 1 was partially supported (Table 1). In bivariate tests, there was no association between YouTube viewership and CBCL-P emotional reactivity score, sleep duration, or prolonged sleep latency. In multivariable models (Table 1), YouTube viewership was independently associated with older child age (adjusted odds ratio [aOR] 1.76 [1.07–2.88] per year of age).

H2 (Parent-level)

Hypothesis 2 was not supported. YouTube viewership was not associated with parent depressive symptoms or parenting stress in bivariate tests (Table 1).

H3 (Contextual factors)

Hypothesis 3 was fully supported. In multivariable models, YouTube viewership was more common in households with lower parent education (aOR 9.12 [2.77–30.0] for high school degree or less, 2.02 [1.08–3.78] for some college/2-year degree, compared to advanced degree). Although income-to-needs ratio was associated with child YouTube viewership in bivariate tests, it was no longer a significant predictor after adjustment for parent education. Children attending home-based childcare were more likely to watch YouTube compared to those attending center-based childcare (3.27 [1.30–8.26]). YouTube viewership was marginally higher in children from underrepresented minority race/ethnicity (aOR 1.61 [0.92–2.81]; of note, this association was statistically significant before controlling for device operating system used by the child).

H4 (Device-level)

Hypothesis 4 was partially supported. YouTube viewership was more common among Android users than iOS users (aOR 2.62 [95% CI: 1.58–4.36]), but contrary to our hypothesis, children who shared a mobile device with a parent or sibling were more likely to use YouTube than children with their own mobile devices (2.57 [1.45–4.54]).

Correlates of mobile YouTube duration

Among 121 participants with their own mobile devices, the only hypothesized factor associated with duration of YouTube viewing in multivariable models was parental education. Children whose parents had a 4-year college degree or more were significantly less likely to have heavier (1+ h/day) YouTube usage than children from families with low education (OR 0.09 [95% CI 0.02–0.42] compared to no usage and 0.13 [0.02–0.75] compared to moderate usage (>0 to <1 h/day).

Discussion

This study used mobile sampling, an objective assessment of mobile device content, timing, and duration, to examine preschool-aged children's use of mobile YouTube. We found that use of YouTube on children's devices is common, comprises about half of overall viewing, and averages about 60 min/day in children with their own devices.

Our findings build upon prior survey-based research, in which the majority of parents reported that their young children watch the main YouTube site.² Although we assessed only mobile YouTube usage, our duration estimates are higher than the 39 min/day reported by parents of 0–8-year-olds in the 2020 Common Sense Census.³

We found that many children use the YouTube app during bedtime/overnight hours, which may be explained by YouTube's design features such as autoplay and tailored recommendation feeds, which are intended to promote engagement and viewing. Young children may be using YouTube at bedtime to play calming videos, as was suggested in prior work surveying parents of toddlers.²⁰ It was notable that in our sample, more children used the main YouTube app than YouTube Kids, which has been designed for child viewers in terms of more content filters and parent controls.

Several child-level and contextual factors were linked with YouTube viewing in our sample. Older children were more likely to use YouTube, which may be due to preference for the larger or more mature selection of videos on the main platform.

We did not find associations between child behavior regulation, sleep quantity/quality, parent depressive symptoms, or parenting stress with YouTube viewership or duration. YouTube viewing is now so common that it may not be specifically driven by parent or child distress; alternatively, parents may report that their child's behavior or sleep is under control because they use mobile YouTube to help manage the child's attention through daily routines.³⁸ Given how frequently mobile YouTube was used around or after bedtime in our exploratory analyses (60 percent of children with their own Android devices), more research on this area is needed.

Children who attended center-based childcare were less likely to watch mobile YouTube than children in other childcare settings, independent of other socioeconomic variables. Caregivers may be using YouTube as a source of early learning videos, which are common on the platform but many of which are low educational quality and have high advertising load.¹³ This finding also highlights that attending high-quality center-based child care is important for multiple reasons, including regulating children's media intake.

We found that children from minoritized race/ethnicity (partially explained by use of Android devices) and lower education households were more likely to use the YouTube app. The latter characteristic also predicted prolonged mobile YouTube viewing (1+ h/day). These findings contribute to a growing body of research documenting differences in quantity or quality of media use by SES^3,39 and have digital equity implications. Specifically, YouTube content analyses suggest that popular videos have low educational quality, a large amount of advertising and commercial influence,¹² and many have depictions of violence,¹³ disturbing content,⁴⁰ or racial/ethnic stereotypes.⁴¹

This study has several limitations. Participating families had a higher proportion of white non-Hispanic race/ethnicity and college graduation than the general U.S. population, but align with our county's demographic characteristics. Although our methods omitted YouTube viewing on browsers or smart TVs, YouTube reports that most users access its platform through the mobile app.⁴

Data were collected before the COVID-19 pandemic, when child media usage changed significantly. However, YouTube viewing during remote learning was a common habit,⁴² so our results still carry relevance for family experiences during the pandemic. We did not collect data on parental coviewing of these platforms, which is important for young children's capacity to learn from technology and build advertising literacy skills. Because this analysis was cross sectional, directionality of associations is unclear, and further longitudinal research is needed.

Conclusions

Online video-sharing platforms such as YouTube take up a growing proportion of young children's media use.³ This is the first objective evidence that lower-SES children commonly use mobile platforms intended for adult users, such as YouTube, and thus may have disproportionate exposure to age-inappropriate content. Updated industry design codes and government policies are needed to promote more equitable access to positive, child-centered digital content.⁴³

Authors' Contributions

J.S.R. designed the parent study, oversaw data collection, designed the analysis, wrote and critically edited the article. J.L.S. helped clean study data, design the analysis, drafted the initial article, and edited the final version. H.M.W. contributed to data analysis decisions, performed data analysis, and critically edited the article. N.K. contributed to data analysis, helped design the parent study, and critically edited the article. A.L.M. contributed to data analysis, helped design the parent study, and critically edited the article. All authors approved the final article as submitted and agreed to be accountable for all aspects of the work.

Author Disclosure Statement

J.S.R. is a paid consultant for Melissa & Doug Toys LLC, the Worldwide Early Development Movement, and Noggin (CBS/Viacom) and receives research funding from Common Sense Media.

Funding Information

The Preschooler Tablet Study was funded by the Eunice Kennedy Shriver National Institute of Child Health and Development (NICHD 1R21HD094051). REDCap and recruitment support was provided through the Michigan Institute for Clinical & Health Research (CTSA: UL1TR002240).