Adverse Childhood Events, not age of Acquiring Smartphones or Tablets, Predict Mental Health in Young Adults

Abstract

Objective

recent years have seen considerable attention to concerns regarding youth access to screens. Many professional bodies and clinicians have advised parents to limit youth access to smartphones and tablets, particularly at young ages. However, little clear evidence has emerged to support this advice.

Method

a dataset of 80,878 young adults (mean age: 23.3) was provided by Sapienlabs. Data included childhood age of first access to smartphones and tablets, as well as exposure to adverse childhood events. Relationships with current mental health were tested using OLS regression.

Results

results indicated that exposure to adverse childhood events was predictive of mental health difficulties in early adulthood. Age of access to smartphones and tablets was not.

Conclusions

advice to parents to avoid earlier access to smartphones and tablets appears to be unwarranted. Moral panic concerns about technology may distract society from more pressing causes of mental health difficulties such as abuse, neglect, and parental mental health.

The issue of youth technology use has been controversial for decades with repeated concerns about screentime and access to technology. Policy makers, scholars and clinicians have historically taken an abstinence approach. This has led political health organizations such as the World Health Organization (WHO) to advocate for abstinence or near-abstinence positions on screen time, particularly for young children (World Health Organization, 2019). Conversely, such recommendations tend also to put more pressure for increased parental interactions with children. Professional guilds such as the American Academy of Pediatrics (2024) and American Academy of Child and Adolescent Psychiatry (2025) recommend against tablet and smartphone use specifically in childhood, often including dire warnings about mental health and learning outcome harms. To the extent screen time is allowable, this is only with a parent actively involved, likely defeating the main motive of parents using screens to be able to take a break themselves.

At the same time, few parents are able to achieve these guidelines, with many parents expressing resentment about them (Brown & Smolenaers, 2018), and others noting that it causes parental guilt and conflict within families (Wolfers et al., 2025). This is of particular interest given concerns that abstinence related screentime recommendations may not be rooted in clear data. For instance, some studies suggest that content is more important than screentime (Linebarger & Walker, 2005), and that issues such as smartphone or tablet ownership correlate poorly with cognitive, sleep, social or mental health outcomes in youth (Martin et al., 2025; Röhlke, 2025).

Screen guidelines and the precautionary principle

Screen guidelines have been ubiquitous for decades, typically warning parents about the various dangers of too much screen time or specific forms of content such as violence, sex, smoking or strong language. Despite this, the empirical validity of such recommendations are often low. For instance, a comprehensive analysis of screen time guidelines put forward by professional guilds found that they routinely overstated the evidence for harmful effects, ignored null studies, improperly generalized laboratory studies to real-world issues such as violent crime and often proposed recommendations that simply weren’t grounded in empirical science (Elson et al., 2019). The consequence of such guidelines so poorly rooted in empirical science is parents struggling to meet utopian expectations of limiting screens with both parents and children expected to be constantly engaged in developmental experiences, even though these recommendations may be of limited practical value.

This is not to imply that such recommendations are ill-intended. Rather it is likely that various cognitive biases contribute to their misuse. First, negativity bias may influence the committees that propose such guidelines. As is often common for human beings, they may be more in tuned to research that proposes the potential for harm than research that suggests such harms are unlikely. This may be even more true when publication bias and prevalent unreplicable p-hacked studies distort scientific understanding.

The second issue is a continued overreliance on p-values when assessing whether data provide support for a hypothesis. Despite decades of understanding that p-values are a poor metric by which to judge hypothesis support (e.g., Wilkinson & Task Force on Statistical Inference, 1999), p-values nonetheless are too often naively interpreted by scholars when evaluating hypotheses. These issues are worsened in large sample studies and meta-analyses when increased power creates a situation in which many “noise” or “crud” responses become statistically significant despite not representing real effects in the real world (Ferguson & Heene, 2021). Scholars may, in good faith, falsely interpret such weak effects as hypothesis supportive, despite effect sizes being near 0.

One recent meta-analysis highlighted this issue (Teague et al., 2026). The authors conducted a large study of screen time including 153 longitudinal studies. The study was highlighted, including in news media, as evidence that screen time was associated with a variety of harmful mental health and educational outcomes. However, a close examination of the study revealed differently. It appeared that many of the included studies lacked basic control variables such as stress or dysfunction in the family or even the time 1 outcome variable. Publication bias was identified for some outcomes, suggesting effect sizes might be inflated. Nonetheless, all effect sizes were very small, mostly near zero. Even taken at face value, far from providing strong evidence for the negative impact of screen time, this meta-analysis found trivial to null effects, well within that expected due to methodological noise or crud. It should never have been interpreted as supporting hypotheses linking screen time to negative outcomes. Yet, the authors ignored many warning signs to do just that.

This may be due to many scholars acting on the precautionary principle. This idea suggests that, in effect, even if evidence is weak or inconsistent it is better to take action to “help children” than not to. However, it’s not always clear this is the case. First, such actions may invite government regulation that can interfere with the free speech rights of both adults and minors. Second, policy efforts may have unintended negative consequences that are poorly understood. Third, policy efforts around screens may distract societal attention away from more important issues such as dysfunction in families, failing schools or even, in the case of violent video games, gun control. Fourth, the precautionary principle may stoke moral panics and create stress and even conflict between parents and youth where it is not helpful. Fifth, policies based on the precautionary principle, at least in regard to media effects, have a long history of failure (Ferguson, 2013).

Smartphones and tablets

One issue that frequently comes up is whether giving children either smartphones or tablets should be delayed for as long as possible. Both instruments are highly desired by many youths. But are they harmful or harmless?

In recent years, professional groups such as the American Academy of Pediatrics (AAP) have backed away from rigid screen guidelines or advocating complete bans (e.g., American Academy of Pediatrics, 2025). However, some advocates are stricter recommending no smartphones before age 14 (e.g., Haidt, 2024). At the same time other scholars have criticized this digital abstinence movement as moral panic rather than good evidence (e.g., Odgers, 2024).

As is often the case, the evidence base can be maddeningly inconclusive. Some studies do find statistically significant relationships between earlier smartphone use and later mental health (e.g., Barzilay et al., 2026). However, effect sizes were so small (approximately r = .07) as to be potentially better explained as methodological noise or a false positive than strong empirical support for an association. By contrast, other studies have found no association between age of smartphone acquisition and later mental health (Vaterlaus et al., 2021).

Evidence for tablet use is, likewise, inconsistent. For instance, one study found that early tablet use was related to later anger (Fitzpatrick et al., 2024) though effect sizes were very small (about d = .22). Other studies have found little relationship between early screen use and later mental health problems with maternal mental health being a much stronger predictor of youth mental health (Sim et al., 2025). This brings us to the issue of adverse childhood events.

Adverse health childhood events and youth mental

In contrast to the data on screens, the evidence that adverse childhood events (ACEs; potentially stress or trauma producing events in childhood) are related to youth mental health is far more consistent, with far larger effect sizes (Flaherty et al., 2013; Meeker et al., 2021). Effects tend to be cumulative across adverse childhood events, though not all adverse childhood events are equal. Exposure to sexual abuse, emotional abuse or neglect, family violence and parental mental health issues appear to have the most impact.

These results suggest that ACEs are far more critical in the production of mental health problems in youth than is the age of smartphone or tablet acquisition. Further, ACEs may serve as an important control variable when examining potential screen effects. It may be the case that youth experiencing more ACEs may turn to screens for information, community support, or pleasant distractions from difficult real-life experiences.

Nonetheless, more data is welcome, particularly given some narratives have focused on age of adoption of smartphones and tablets as an issue of concern. As such, in the current study, age of adoption of smartphones and tablets in childhood is considered as a predictor of mental health issues in young adulthood. These predictors are examined when controlling for a well-known predictor of mental health issues, adverse childhood experiences such as abuse, parental suicide, etc. (Merrick et al., 2017).

Methods

Dataset

All data come from the Brainbase/Global Mind Project dataset provided by Sapienlabs (2025). This is a large online dataset consisting of several hundred thousand adult participants across multiple waves of data collection from 2019-2025. Specific data collected have varied by wave, including initial validation studies in earlier waves. The current analysis uses data from the 2025 wave of data collection, assessed 8/12/25. The option for clean data, eliminating participants whose responses indicate inattention, was used.

Participants

Analyses were limited to young adults (ages 18-29.5) as these age categories are most adjacent to adolescent and childhood experiences. The sample was heavily female (63%), with a median age of 22.5 (SD = 4.78). Country of origin was international and very diverse with high representation from countries outside the developed world. Highest representation by country was for India (16.0%), followed by Iraq (5.8%), Egypt (5.6%), Philippines (5.6%), Mexico (5.5%), Venezuela (4.3%), Algeria (3.8%), Jordan (3.8%), Columbia (3.6%). By contrast, representation from the United States was 1.9%, with France and the United Kingdom each at 0.4%.

Measures

A copy of the datafile with transformed ACE scores was sent to Sapienlabs to check for any errors. Sapienlabs graciously confirmed the data and analyses were accurate.

Mental health quotient

The main outcome measure was the Mental Health Quotient (MHQ), a self-report measure compiled from other mental health inventories and, via validity testing, compressed to 47 items. The MHQ has demonstrated good psychometrics and construct validity during its developmental process (Newson et al., 2024).

Predictors and covariates

The main predictors for the current study were self-reported age of first use of tablets and smartphones. Respondents also answered questions related to a list of 20 adverse childhood experiences (ACEs) ranging from experiencing divorce to parental suicide to experiencing war, as well as parental abuse, death of a parent, parental imprisonment and suicide, etc. In the publicly available dataset, ACEs were provided as narratives in a single string data column making quantification along a continuous variable difficult. As such, this variable was considered as a binary variable of whether respondents experienced ACEs or not. Creating a binary variable inevitably reduces variance and will truncate effect sizes in most circumstances, though was the best solution given the presentation of ACEs as single string narrative rather than specific, individual datapoints in the dataset. Age of the participant was also included as a covariate.

Data analysis

Data was analyzed using Ordinary Least Squares (OLS) regression. The main outcome variable, MHQ, was normally distributed with acceptable kurtosis and skew, making this an ideal candidate for the use of OLS regression. Analyses were conducted with jamovi software. All VIFs were below 1.9, suggesting an absence of multicollinearity issues.

One unfortunately aspect of large datasets is that methodological noise can become “statistically significant” leading to false positive results. The recommendation to avoid this is to set a smallest effect size of interest (SESOI) of a minimum of r (or beta) of .10, or even .20 to be more robust (Ferguson & Heene, 2021).

Results

The overall regression model for the MHQ outcome was significant [F(4, 42,508) = 939, p < .001, R = .285, R² = .081]. In this model ACEs were related to reduced MHQ scores (β = -.246) meaning that exposure to ACEs is related to reduced mental health. This value was above the higher SESOI and, as such, likely to represent a clinically significant predictor. Age was also modestly related to improved MHQ scores (β = .120). Although this was above the lower SESOI of r = .10, it was below the upper SESOI of r = .20, indicating it may be of limited clinical value. By contrast, neither smartphone nor table first age of ownership were related to MHQ scores. Though they were technically “statistically significant” this is often a poor indicator of hypothesis support in large sample studies such as this. Both standardized coefficients were well below the minimum SESOI, with effect sizes in opposing directions. Rerunning the analyses with only females or males did not change the results suggesting gender is not a moderator. These results are presented together as table 1.

Table 1. Outcomes for the OLS regression of MHQ scores.

Model Coefficients – Total Mental Health Questionnaire Scores

Predictor	Estimate	SE	t	p	Stand. Estimate
Intercept	21.332	1.8027	11.83	<.001
age	1.823	0.0825	22.10	<.001	0.1203
Adverse Childhood Exp.	-36.407	0.6905	-52.73	<.001	-0.2455
Age of Tablet Ownership	0.811	0.0907	8.94	<.001	0.0602
Age of Smartphone Own.	-0.812	0.1085	-7.49	<.001	-0.0492

Note: Higher MHQ scores equal better mental health.

Some concerns regarding technology use and mental health have focused specifically on biological females. As such, the regression was rerun with the sample of only biological women. This did not substantially change any of the data outcomes.

Discussion

Concerns about youth access to technology have led to some calls to postpone youth access to smartphones and tablets for as long as possible. Despite this, it has remained unclear whether such advice or policies were effective in promoting mental health moving into young adulthood. The current study found no evidence that age of first use of smartphones or tablets was associated with later mental health outcomes in young adulthood. By contrast, experiencing ACEs was clearly associated with decline in mental health at clinically significant levels. This raises the concern that much of the social narrative on mental health has focused on the wrong predictors to the detriment of youth wellness.

Early technology use and childhood

Early use of technology was not a significant predictor of mental health, regardless of whether the full sample or females or males only was examined. This suggests that narratives that female youth may be particular vulnerable to early technology use are unwarranted. Overall, these results suggest that advice to restrict youth access to technology is unlikely to be useful, at least as it pertains to mental health.

It is worth exploring why such advice persists despite limited evidence for it. One possibility is that older adults may experience nostalgia bias whereas they view their own childhoods retrospectively as idyllic whereas modern childhoods are corrupted by technology. Thus, youth enjoying new technology may be seen to be doing childhood “wrong” though older adults may overestimate their time outside as opposed to watching television or playing Atari 2600 inside, issues that were, likewise, criticized as “screen time” during their day. As such, modern concerns about smartphones and tablets may resemble moral panics of past generations.

There may also be a viewpoint that both children and their parents should be endlessly productive… continually learning and developing as opposed to just relaxing and enjoying downtime. This may explain advice that suggests parents must always interact with kids and their screens, and that “learning” screens may be ok, but entertainment ones are less desirable. Overall, however, effect sizes for screen time of any kind and either mental health or cognitive outcomes appear to be so weak as such advice may be of limited actual practical value.

Unwarranted fear-based advice may cause increasing pressure on parents resulting in stress. Mothers, in particular, may feel pressure to be “perfect”, increasing stress, overprotectiveness and intrusion in children’s lives (Venard et al., 2024). Such advice may create a culture of “intensive motherhood” in particular that puts unrealistic expectations on parents and mothers specifically (Hays, 1996). This may create a “paranoid” approach to parenting that, ironically, undermines children’s health and independence (Furedi, 2008).

Adverse childhood events

By contrast, ACEs were clearly predictive of negative mental health outcomes in early adulthood. This was above the level of clinical significance. This is particularly interesting given that the variable was bivariate and, as such, of reduced variance. It’s possible the actual effect may be even stronger, though some ACEs may tend to have more impact on others so simple quantitative counts may not be the best way of evaluating ACEs.

The issue of ACEs may be unintentionally undermined by the moral panic regarding smartphones and technology. As researcher Mike Males has noted, Centers for Disease Control data suggests the minors are heavily influenced by ACEs in their family, whereas technology use has a neutral or perhaps even protective influence (e.g., Males, 2026). Moral panics often have a distracting influence, causing policy makers to lose sight of complex but important issues, and instead focus on “quick fix” cultural issues that may be momentarily popular even if of limited practical value. This has occurred repeatedly throughout the 20th and into the 21st century from movies to radio to comic books to video games and continues with modern technology. Unfortunately, scholars and even professional guilds often contribute to these moral panics with biased summaries of research (Elson et al., 2019).

Limitations

As with all studies, this one has limitations. All data here are correlational and, as such, no causal assertions can be made. Data is self-report and retrospective which also may reduce reliability to some degree, though arguably this may tend to produce artificially high effect sizes rather than artificially low. Although age of access to technology may be a poor predictor of mental health, it is possible that other aspects of technology may be associated with mental health. This may include issues such as active versus passive use, a proneness to engage in negative self-comparison, oversharing of personal information, etc. That ACEs were used as a binary variable limits both variance and the ability to examine them in a sophisticated manner. Very likely, some ACEs will have more influence on later mental health than others and grouping them together will eliminate that nuance. Further investigation of ACEs analyzed as separate incidents is warranted. Lastly, age of use is only one potential metric of technology use. However, it is a frequently targeted one given interest in government bans on social media use, and various professional organization recommendations for screen use at various ages. Other studies have suggested that frequency of use is a poor predictor of mental health. As such, this study adds age of acquisition to lesser predictors, perhaps unworthy of clinical consideration. Other analyses that consider how youth use technology would likely be of greater importance.

Conclusions

Concerns about young people and technology are unlikely to abate in the foreseeable future and may be a permanent fixture of aging generations. However, repeated patterns of moral panic may undermine rather than help youth welfare. Further, misstatements of evidence or premature advice based on the “precautionary principle” ma undermine people’s faith in science. It is important that beliefs about youth mental health and technology be held to the highest empirical standards. Unfortunately, such hypotheses are arguably held only to standards that are very, very low. The current study does not support efforts to restrict youth access to smartphones or tablets. Instead, if policy makers are serious about the mental health of young people, they should focus away from these issues and more on adverse childhood experiences.