Research

At-risk teens may face increased online threats

Research suggests opportunity for algorithm-based, targeted preventative interventions to combat Internet-initiated victimization

Girls who had been sexually abused were more likely to fit a profile characterized by high pornography and social media use, according to a new Penn State study. Credit: Getty Images: Antonio GuillemAll Rights Reserved.

UNIVERSITY PARK, Pa. — While the Internet has many benefits such as increased access to education, it also has risks like access to pornography and the potential for online victimization and exploitation — and new research has found that certain teens may be at higher risk for these threats.

According to a new study where online activities were objectively observed, girls who had been sexually abused had greater odds of being cyberbullied. They were also more likely to fit a profile characterized by high pornography and social media use, which predicted being sexually solicited online and engaging in more sexual activity two years later.

When considered in the same model with all Internet and psychosocial variables, social media use was a non-zero feature in the prediction of being cyberbullied and receiving online sexual solicitations, and pornography was a non-zero feature in the prediction of heightened offline sexual activities at follow-up.

The researchers also found that two years after Internet usage was monitored on the laptops, the participants who had viewed more pornography were more likely to be engaging in risky sexual behaviors. Additionally, those who had spent more time on social media were more likely to experience cyberbullying two years later.

Jennie Noll, professor of human development and family studies and director of the Child Maltreatment Solutions Network at Penn State, said the findings suggest a need for targeted intervention strategies — both online through predictive machine learning and also to augment trauma treatments — that could help at-risk teen girls navigate the Internet.

“These at-risk teens have very distinct Internet usage patterns, suggesting an opportunity to use machine learning techniques that would recognize at-risk kids based on the URLs they visit and then deliver targeted prevention messaged directly to their newsfeeds and the apps they use,” Noll said. “We're smart enough to use such algorithms for target marketing and to sell consumers all kinds of goods, so why not employ these same technologies to use the internet for good and to keep kids safe?"

The findings were recently published in the journal Nature Human Behaviour.

According to the researchers, today’s adolescents are the first generation to grow up with nearly unlimited access to the Internet. This access gives kids an opportunity to socialize, play games and find information on nearly any subject, but it also opens the possibility for them to be cyberbullied or targeted by sexual predators.

Additionally, the researchers added that those who have experienced child sexual abuse are at higher risk for sexual revictimization, exploitation and sex trafficking, all which can originate by meeting strangers online.

Noll said that while there is spirited debate regarding the impact of the Internet on today's youth, previous research has found conflicting results about whether the Internet can impact teen development. Some studies have found little to no harm, and that the opportunity for socializing can even be beneficial. Others, however, have found links to depression, less sleep, and poorer school performance.

“But there are some problems with prior Internet research given that all prior studies relied on kids' self-reported internet activity,” Noll said. “Asking kids about how much time they spend online, or on social media, or if they ever view pornography could provide inaccurate data given that kids may not know exactly how much time they spend in certain activities or may not want to admit that they view adult content. Our team sought to raise the rigor of research by actually observing what kids are doing.”

For the study, the researchers recruited 460 girls between the ages of 12 and 16 — 156 who had experienced substantiated sexual abuse and 304 comparison participants who had not. Comparison participants consisted of two groups. One included girls from the same neighborhoods, income-levels and ethnicities as the girls who had been abused, and the other included girls who were "census-matched" to the sociodemographics of the region where the study took place.

“We wanted to study these three distinct groups because it was important to be able to tease out whether the prior sexual abuse was linked with problematic outcomes over and above other sociodemographic risks factors such as poverty or living in specific areas,” Noll said.

The participants were then given laptops that were specially equipped with software that recorded all URL activities as well as all YouTube and Netflix keywords. Laptops also had modems to allow high-speed internet access, and participants were instructed to use the laptop as their main device for the next four weeks.

Additionally, the team used a sophisticated authentication algorithm based on keystroke patterns to ensure with 97% specificity that the Internet activity was that of the participating adolescent.

Participants were also interviewed to assess psychosocial risks, such as substance abuse, depression, and impulse control, as well as potentially protective factors such as self-esteem, engagement in school and extracurricular activities, and family support. These interviews were repeated annually for two subsequent years.

The researchers analyzed the data using a technique called latent profile analysis, which sorted participants into three main groups.

The first was a low-risk group — girls who had low Internet activity and little substance abuse but high levels of protective factors like good support from families and involvement in extracurricular activities. The second was a moderate risk group consisting of girls who had moderately high levels of Internet usage but low or moderate levels of other risk factors.

Finally, there was a third group that the researchers considered to be the highest-risk group. These participants were high Internet and social media users, but also viewed more pornography, reported higher levels of substance abuse and depression, and experienced low parental support, poor grades and low self-esteem.

Noll said girls who had been sexually abused previously were the ones more likely to fit into this high-risk profile.

“Girls in this profile were the ones who were more likely to view pornography, be cyberbullied, and be sexually solicited online,” Noll said. “Compared to the census-matched group, sexually abused girls were also more likely to meet strangers offline where the encounter ended up in sexual assault or attempted assault."

Noll said the longitudinal nature of the study was important because it helped them demonstrate that they could predict which kids will experience Internet-initiated victimization.

In addition to using this data to help create targeted interventions for at-risk girls, Noll said the findings could also be used to augment trauma treatments for sexual abuse survivors. Internet safety programming, for example, could be included in the safety planning portions of existing evidence-based trauma treatments. 

“There seems to be something lingering and lasting regarding the traumatic sexualization that can accompany childhood sexual abuse that sets the stage for victimization that is unique to the rise of the Internet,” Noll said. “These findings suggest augmentations to treatments that may enhance recovery.”

The Child Maltreatment Solutions Network (CMSN) was established by Penn State in November of 2012 to serves as a national resource dedicated to solving the complex problem of child abuse and neglect, and is a unit of the Social Science Research Institute at Penn State.

The CMSN conducts impactful new science, accelerates science to practice and policy change, trains the next generation of professionals, and increases awareness to mobilize public investment in prevention and treatment. The CMSN is a group of transdisciplinary scientists working in conjunction with advocates, practitioners, and policy makers to create dynamic system-wide solutions.

Ann-Christin Haag, Columbia University Teachers College; Chad E. Shenk, Penn State; Michelle F. Wright, Penn State; Jaclyn E. Barnes, Cincinnati Children's Hospital Medical Center; Mojtaba Kohram, Cincinnati Children's Hospital Medical Center; Matteo Malgaroli, New York University; David J. Foley, Penn State; Michal Kouril, Cincinnati Children's Hospital Medical Center; and George A. Bonanno, Columbia University Teachers College, also participated in this work.

The National Institutes of Health, National Center for Advancing Translational Sciences, and American Foundation for Suicide Prevention helped support this research.

Last Updated October 5, 2021

Contact