Eric McCloySchool of Technology and ComputingCity University of Seat

bstract—As teenagers increasingly engage with digital

technology, cybersecurity vulnerabilities present significant

risks to their online safety and privacy. Adolescents who lack

awareness of secure online practices are particularly

vulnerable to malicious actors seeking to exploit them. This

paper is an empirical study investigating the relationship

between real-world online behavior of teenagers,

cybersecurity risks, and device interactions. The primary data

set used for this analysis is teenage online behavior and

cybersecurity risks. First, we consider demographic

information: age, education, time spent online. We correlate

this with online behaviors: use of a VPN, type of equipment

(computer, mobile), use of public internet, engagement with

risky websites. Finally, we analyze the data set using a

combination of demographic and behavioral patterns to

search for high-risk, negative outcomes. For our research, we

analyze teenage online behavior patterns to identify key risk

factors, develop predictive models for cybersecurity

vulnerabilities, and produce actionable visualizations that

illustrate the relationship between digital literacy and online

safety. Our findings utilize data and business analytics to

provide evidence-based recommendations for parents,

educators, and policymakers to enhance teenage

cybersecurity awareness and protective strategies.

Keywords—online behavior, teenagers, adolescent, privacy, cybersecurity risk,

data analysis

I. INTRODUCTION

The internet has permeated nearly every aspect of human

life, including the daily experiences of children and

adolescents. While it offers numerous benefits—such as

enhanced communication, access to information,

entertainment, and educational opportunities—it also

introduces significant cybersecurity risks, particularly for

younger users. Due to their limited experience and cognitive

development, teenagers may not immediately recognize

security threats and may fall victim before realizing the danger

[7]. Cybersecurity awareness practices aim to equip users with

the knowledge and skills to identify and mitigate such risks.

However, while these practices are generally considered

effective, research suggests that their implementation may

not be sufficient. Ondrušková and Pospíšil [5] argue that a

single training session does not meaningfully improve

cybersecurity awareness in children. Similarly, Mwagwabi and

Jiow [4] found that even when teenagers suspected their

computers might be compromised, this suspicion did not

influence their password choices—indicating that passive

guidelines alone are inadequate unless enforced through

stronger authentication mechanisms. Moreover, Mwagwabi

and Jiow [4] highlight a notable lack of theory-based studies

on teenage cybersecurity behavior, suggesting a need for

deeper exploration in this area.

This paper seeks to address this gap by analyzing the

Teenage Online Behavior and Cybersecurity Risks dataset. By

examining variables such as age, time spent online, device

usage, and specific online behaviors (e.g., VPN use, risky

website engagement, public network access), the study

identifies key risk factors associated with cybersecurity

vulnerabilities in teenagers. The results are used to develop

correlations and actionable visualizations that illustrate the

relationship between digital behavior and online safety. These

insights aim to provide evidence-based recommendations for

parents, educators, and policymakers to enhance

cybersecurity awareness and protection strategies for

adolescents.

II. LITERATURE REVIEW

An examination of studies related to adolescent online

behaviors and exposures to negative outcomes revealed a

complex interplay of factors, including some counterintuitive

findings. For example, teens who had a higher “digital literacy”

(technical understanding and skillset, including being able to

recognize online threats) were still exposed to a greater

number of negative online experiences because they tended

to spend significantly more time online [12]. Another study

revealed that strict parental control of social media usage

resulted in a small but significant increase in the likelihood of

teens sharing private information online [3] Considering these

observations, it is important to take a careful and nuanced

look at the literature in order to challenge potentially

erroneous preconceptions.

Adorjan and Ricciardelli [1] explored the issue of “online

addiction” as a sensitizing concept. By this, they meant that

2026 Journal of The Colloquium for Information Systems Security Education, Volume 13, No. 1, Spring 2026

©2026 The Author(s) | Creative Commons CC BY 4.0 2 www.cisse.info

instead of trying to confirm previous studies findings about

the extent of addiction or trying to define it precisely

themselves, they used a focus group with a discussion-based

methodology to understand teens’ self-perception of online

addiction. During their study, they avoided using the term

“online addiction” until it was brought up by a focus group

participant. In all 35 independent focus groups they ran during

the study, participants brought up the term “online addiction”

to describe themselves without prompting or previous

mention of the topic. The study found that the primary focus

of online activity tended to be school-based networks of online

friends implying that perhaps the addiction isn’t to technology

itself but to the social interaction that the technology enables.

Many of the focus group participants also noted their parents’

behavior modeled online addiction for them.

The question is, does online addiction have negative

consequences? There were several direct, broad categories of

online risks identified in these studies, as well as negative

well-being outcomes associated with extensive internet usage

and the exposure to these risks. The major types of direct risks

to teens can be categorized as Content, Contact, and Criminal

Risks.

1. Content Risks are related to potentially harmful content,

such as sexual images, violent images, or hate speech [8].

2. Contact Risks are unwanted interactions with people

online. This includes bullying, being drawn into unwanted

conversations and arguments, attempted real-life contact

by strangers, recruitment by hate groups, and sexual

solicitation [1] [8]

3. Criminal Risk includes offers to sell alcohol or drugs,

gambling, and blackmail [8][9].

In addition to direct risks, the studies identified a number

of negative well-being or mental health outcomes. These

included: low self-esteem, depression, anxiety, sleep

deprivation, loneliness, and feelings of inadequacy [1][12].

Specifically, the “Fear of Missing Out” (FOMO) was cited as a

major reason for constant addictive behavior on social media

sites and was positively correlated with “boredom, loneliness,

depression, and feelings of inadequacy and anxiety, as well as

diminished well-being, overall mood, and life satisfaction” [1]

(p. 51).

Poor academic performance may also be correlated with

addictive online behaviors. However, this relationship is

complex. Use of online tools can be correlated with positive

performance, but high internet use at school and home has

been linked with low academic performance, specifically in

declining mathematics grades. Online addiction is also

associated with a decreased motivation to study, poor

cognitive behavior control, and deteriorating relationships

with teachers and schoolmates. All these issues correlated

with degraded academic performance [6].

Parental intervention is a particularly complex aspect of

teen online addiction. Álvarez-García et al. [2] found a small

but statistically significant negative relationship. With

increased parental restriction, there was an increase in

adolescent risky behaviors online. Kang et al. [3], found that

the type of parental intervention mattered. They categorized

parental intervention as “restrictive” when a tight set of rules

were expected to be followed and “active” when the parents'

helped teens develop an understanding of the issues and

involved the teens in creating boundaries. The study showed

that teens with restrictive parents were less likely to use tight

privacy controls on Douyin (a Tik Tok type of app popular in

China) than teens with active parents. The researchers

concluded that teens who were actively included in the

process were better able to make considered judgements

regarding protecting themselves online.

Finally, the issue of digital literacy was examined in the

literature. Vissenberg et al. [12] define digital literacy as “the

skills, knowledge and attitudes that make learners able to use

digital media in a critical, responsible and creative manner” (p.

77). However, as they examined the issue more closely, they

came to the conclusion that, even though digitally literate

teenagers are better able to avoid risks, they still encountered

direct risks more often, because they spend more time online.

When the researchers questioned teens who had risky online

experiences, they also discovered that teens with higher levels

of digital literacy also demonstrated higher levels of “Online

Resilience”, that is, they were better able to process and cope

with the negative experience without a long-term effect on

their well-being.

III. BACKGROUND

The data used in our analysis was collected from network

activity logs and e-safety monitoring systems across various

educational institutions and households in Texas and

California over a 7-year period (2017-2024). The dataset

comprises 67,921 observations and 30 columns including 19

numerical and 11 categorical variables.

The quality of the dataset is demonstrated by its use in the

literature. Most notably, Xu, et al. used the dataset to train an

AI to create a framework for teen learning related to

cybersecurity [12].

To enhance our understanding of the data, we organized

the columns into the following groups: Time and User

Demographics, Device and Network Information, Security

Threats and Incidents, User Behavior and Activities,

Authentication and Access, Safety and monitoring Controls

and Risk Assessment. (Appendix A).

Each group offers a unique perspective for understanding

and examining the relationship between teenage behavior and

cybersecurity risks. In the User Demographics category, we

analyze the timestamp of each incident, age group of the

teenager and the number of hours they were online. The age

groups are categorized as follows: children under 13, teens

between 13 and 16 and teens aged 17-19. Table I shows the

age distribution in the dataset indicating a predominant

representation of middle teens (13-16) that comprises 70.2%

2026 Journal of The Colloquium for Information Systems Security Education, Volume 13, No. 1, Spring 2026

©2026 The Author(s) | Creative Commons CC BY 4.0 3 www.cisse.info

(47,695 records) of all observations followed by 19.9% (13,491

records) for older teens and 9.9% (6,735 records) for pre-

teens.

TABLE I. Age Distribution of Participants

Age Group Count(n) Percent (%)

Under 13 6,735 9.9

13-16 47,695 70.2

17-19 13,491 19.9

Total 67,921 100.0

In the device and network information category, we

examine several key variables including device type

(smartphones, laptops, desktops, and tablets). These device

types will be used to create a derived feature called Device

Security Index, which assigns values based on the relative

security of each device type.

The dataset includes several risk indicators related to

cybersecurity threats, such as malware incidents, phishing

attempts and visits to risky websites. Positive online security

practices such as the use of strong passwords used, whether

a teenager used a VPN and public network usage is also

captured along with the individual’s level of e-safety

awareness.

Fig. 1. Distribution of Device Types

As shown in Figure 1, mobile phones are the most

dominant platform accounting for 70.17% (47,662 records) of

usage. This suggests a mobile first mindset in how teenagers

interact online. Traditional devices (Laptops, Tablets and

Desktops) account for 29.8% (20,259 records) of usage

suggesting these devices are used in more targeted or specific

contexts.

IV. METHODOLOGY

A. Data Acquisition

The data was downloaded from Kaggle – the data science

competition platform at https://www.kaggle.com/datasets/

datasetengineer/teenage-online-behavior-and-cybersecurity-

risks as a single 8.8Mb CSV file. The repository was first

uploaded on October 9th, 2024, and has been downloaded 338

times with 1987 views as of May 23, 2025.

Our initial task was to examine the structure of the dataset

using the pandas commands df.shape and df.info to

understand its dimensions and data types. The full dataset

comprises 67,921 observations with 30 columns.

B. Data Cleaning Strategy

In our analysis of the data, we identified several issues

with consistency and completeness. As such, we crafted a

detailed strategy to clean the data and determine which

elements should be discarded, replaced or otherwise

modified.

1) Replacing Missing Values

The Education_Content_Usage column contained 47595

null values, significantly affecting the completeness of the

dataset. To address this, we removed the column entirely. For

other columns, we standardized any missing values by

replacing them with np.nan.

2) Clean Dates

Visually inspecting the TimeStamp column, we found the

date format to mainly be consistent. However, to ensure

consistency we applied logic to parse the data in the correct

format.

3) Handling other data inconsistencies

Here is the approach we took to clean the data:

• Replace missing values

• Parse dates as noted above

• Convert to numeric and handle non-numeric numbers

• Create derived columns

70.2

10 9.9 9.9

0

10

20

30

40

50

60

70

80

Mobile Tablet Laptop Desktop

Device Type Distribution Results

2026 Journal of The Colloquium for Information Systems Security Education, Volume 13, No. 1, Spring 2026

©2026 The Author(s) | Creative Commons CC BY 4.0 4 www.cisse.info

TABLE II. Derived Features for Data Analysis

C. Summary of cleaned data

Appendix B contains the full summary of cleaned data. Our

initial analysis shows that 91.68% of the 67,921 records were

100% complete; Average completeness for the dataset was

99.84%. Observations in the dataset were made from Jan 1,

2017, to Jan 1, 2024, a period of 6 years.

The most predominant age group was the 13–16-year-olds

who represent 70.2% followed by older teens (17-19) and

children under 13. We observed an average daily usage of 2

hours when teens went online. In terms of devices, Mobile

phones represent the largest device platform with a total of

70.2%. Traditional devices (Laptops, Tablets and Desktops)

account for 29.8%. We see a clear trend in the data where

70.2% of 13–16-year-olds use mobile devices indicating the

core teenage demographic is driving the mobile-first behavior.

V. DATA ANALYSIS

This section describes data analysis of Teenage online

behavior and cybersecurity risks data analysis dataset from

Kaggle by Sik, (2024). The data were cleaned as described in

the previous section. To explore the relationships among key

behavioral and security-related features, a subset of derived

features was selected for correlation analysis presented in

Table II. This subset included both numerical metrics

(e.g., Digital_Consumption_Index, Threat_Severity_Index) and

encoded categorical constructs (e.g., Usage_Pattern_

Category, Security_Posture_Category). Prior to the analysis,

selected categorical features were ordinally encoded to allow

for correlation computation.

A Spearman correlation matrix was computed to assess

monotonic relationships between the features. According to

Hauke & Kossowski [11], this method is well-suited for ordinal

and non-linear associations, which are expected in behavioral

data. The correlation matrix was visualized using a heatmap

to highlight strong associations (Figure 2).

Feature Name Type Formula

Usage_Pattern_Category category Based on Hours_Online: Light (<=2), Moderate (<=5), Heavy (>5)

Time_Period_Risk category Hour-based risk: Low (7–15), Medium (15–22), High (22–7)

Device_Security_Index float Mapped from Device_Type with fixed scores

Connection_Safety_Score float Composite score: VPN (0.4), Public_Network (0.3 inverse), Network_Type (0.3)

Total_Threat_Exposure int Sum of Malware_Detection, Phishing_Attempts, Data_Breach_Notifications

Threat_Severity_Index float Weighted threat score: Malware (1), Phishing (2), Breach (3)

Defensive_Posture_Score float Firewall_Logs (0.6) + VPN_Usage (0.4) scaled to 10

Risky_Browsing_Ratio float Risky_Website_Visits / Website_Visits

Digital_Consumption_Index float Weighted activity: Website (0.4), Ads (0.3), Cloud (0.3)

Password_Security_Score int Mapped from Password_Strength: weak=2, moderate=6, strong=10

Parental_Oversight_Level category Based on Parental_Control_Alerts: Low (0), Medium (<=3), High (>3)

Supervision_Effectiveness float Prevention / (Threats + Prevention)

Safety_Behavior_Score float Avg. normalized VPN, Connection_Safety, Defensive_Posture

Awareness_Behavior_Gap float Numeric_Awareness - Safety_Behavior_Score

Overall_Security_Posture float Weighted average of multiple security scores

Security_Posture_Category category Categorized from Overall_Security_Posture (Vulnerable, Adequate, Secure)

Behavior_vs_Protection_Gap float Normalized risk - avg. protection scores

Protection_Match_Category category Categorized from Behavior_vs_Protection_Gap (Overprotected,Balanced, Underprotected)

2026 Journal of The Colloquium for Information Systems Security Education, Volume 13, No. 1, Spring 2026

©2026 The Author(s) | Creative Commons CC BY 4.0 5 www.cisse.info

Fig. 2. Heatmap based on a Spearman correlation matrix to identify correlations between derived feature

Two sets of feature pairs were extracted from a Spearman

correlation matrix:

• Positively correlated pairs, Spearman coefficient ≥ 0.5

• Negatively correlated pairs, Spearman coefficient ≤ -0.5

To ensure statistical rigor, each of these pairs was tested

for significance using hypothesis testing. All extracted pairs

demonstrated from moderate to strong correlation as shown

in Table III. This process helped identify key patterns in the

dataset, such as how certain online behaviors may co-vary

with cybersecurity posture, awareness, or parental oversight.

2026 Journal of The Colloquium for Information Systems Security Education, Volume 13, No. 1, Spring 2026

©2026 The Author(s) | Creative Commons CC BY 4.0 6 www.cisse.info

TABLE III. Results of Testing for Significance

Feature X Feature Y Method Corr. Coef P-value

Positively correlated pairs

Connection_Safety_Score Safety_Behavior_Score Spearman 0.84 0

Connection_Safety_Score Overall_Security_Posture Spearman 0.68 0

Total_Threat_Exposure Threat_Severity_Index Spearman 0.99 0

Defensive_Posture_Score Safety_Behavior_Score Spearman 0.77 0

Defensive_Posture_Score Overall_Security_Posture Spearman 0.58 0

Safety_Behavior_Score Overall_Security_Posture Spearman 0.75 0

Safety_Behavior_Score Protection_Match_Category Spearman 0.55 0

Overall_Security_Posture Protection_Match_Category Spearman 0.61 0

Security_Posture_Category Behavior_vs_Protection_Gap Spearman 0.54 0

Negatively correlated pairs

Connection_Safety_Score Behavior_vs_Protection_Gap Spearman -0.58 0

Total_Threat_Exposure Supervision_Effectiveness Spearman -0.97 0

Threat_Severity_Index Supervision_Effectiveness Spearman -0.97 0

Defensive_Posture_Score Behavior_vs_Protection_Gap Spearman -0.51 0

Safety_Behavior_Score Security_Posture_Category Spearman -0.51 0

Safety_Behavior_Score Behavior_vs_Protection_Gap Spearman -0.63 0

Overall_Security_Posture Security_Posture_Category Spearman -0.84 0

Overall_Security_Posture Behavior_vs_Protection_Gap Spearman -0.68 0

Behavior_vs_Protection_Gap Protection_Match_Category Spearman -0.84 0

2026 Journal of The Colloquium for Information Systems Security Education, Volume 13, No. 1, Spring 2026

©2026 The Author(s) | Creative Commons CC BY 4.0 7 www.cisse.info

To uncover distinct patterns in teenage online behavior

and security posture, a K-Means clustering analysis was

conducted using eleven behavioral and security-related

features. These included indices related to digital activity

(e.g., Digital_Consumption_Index, Device_Security_Index),

security practices (e.g., Password_Security_Score,

Defensive_Posture

Score), exposure indicators (TotalThreat_Exposure,

Threat_Severity_Index), and demographic information

(Age_Group_Encoded). All features were standardized prior to

clustering to ensure equal contribution to distance

computations. The clustering of behavioral profile is

illustrated in Figure 3.

Fig. 3. Cluster Behavior Profiles

The heatmap in Figure 3 illustrates average scores of

behavioral and security features across three user clusters

derived via K-Means clustering. Cluster 0 represents risk-

aware users with strong security behaviors. This cluster is

characterized by a high Connection_Safety_Score (8.03) and

Password_Security_Score (5.20), indicating strong adherence

to secure online behaviors. The pattern is further reinforced by

a moderate to high Safety_Behavior_Score (4.74) and an

elevated Overall_Security_Posture (5.32). Additionally, very

low levels of Total_Threat_Exposure (0.36) and Threat_

Severity_Index (0.68) suggest that these defensive practices

are effective in minimizing online risks. This cluster likely

represents cyber-aware, well-supervised, and safety-

conscious users.

Cluster 1 includes low-engagement users with minimal

threat exposure. This cluster demonstrates very low

Total_Threat_Exposure (0.00) and Threat_Severity_Index

(0.00), indicating minimal engagement with online risks. Users

exhibit low to moderate scores in key security behaviors, such

as Password Security (1.24) and Safety Behavior (1.76).

Additionally, Supervision_Effectiveness (1.00) and Defensive_

Posture_Score (1.24) are relatively low, suggesting limited

external guidance or self-initiated precautions. These patterns

imply that users in this cluster may have restricted or minimal

interaction with online environments—possibly due to parental

controls, limited internet access, or inherently low digital

engagement.

Cluster 2 represents moderately engaged users with some

risk and limited supervision. Cluster 2 includes users with

higher exposure to online threats compared to Cluster 1, as

indicated by a Total_Threat_Exposure score of 1.20 and a

Threat_Severity_Index of 2.27. While their scores for

Connection Safety (4.03), Password Security (3.00), and

Overall Security Posture (2.94) are moderate, the notably low

Supervision_Effectiveness (0.55) and Defensive_Posture_

Score (1.12) raise concerns about their vulnerability. These

children appear to have some awareness of cybersecurity

risks but may lack consistent supervision or comprehensive

protective practices, placing them at an intermediate level of

risk.

Figure 4 demonstrates the demographic distribution of

age groups across the three behavioral clusters, highlighting

how users from different age ranges are represented within

each cluster.

Fig. 4. Age Group Distribution per cluster

The largest segment of users across all clusters belongs

to the 13–16 age group, with a significant concentration in

Cluster 1, which denotes low-engagement or minimally

exposed users. Clusters 0 and 2 have fewer users overall and

display a more balanced distribution of age groups. This

indicates that while age can affect behavioral clustering, it

does not solely determine cluster formation, as similar age

groups are present across different behavioral profiles.

To support interpretation, Principal Component Analysis

(PCA) was applied to reduce the behavioral feature space to

two principal components, enabling 2D visualization of cluster

2026 Journal of The Colloquium for Information Systems Security Education, Volume 13, No. 1, Spring 2026

©2026 The Author(s) | Creative Commons CC BY 4.0 8 www.cisse.info

separability. The resulting scatter plot (Figure 5)

demonstrates visible differentiation among the three clusters,

with some overlap between Cluster 0 and Cluster 2, while

Cluster 1 forms a clearly distinct linear group.

Fig. 5. PCA Behavior Clustering

The source code of the data analysis is in Appendix E.

VI. FUTURE RESEARCH

There are a number of gaps that stand out as opportunities

for further research. The Kang et al. study [3] lays out an

interesting relationship between restrictive and active

parenting styles. However, the categories are broad, and the

outcomes were only measured with reference to a single

social media platform. This study was primarily an empirical

look at a longitudinal data set. Further theoretical research

would be beneficial. A study including additional apps and

more nuanced categories regarding parenting restrictions

could uncover more about teens and risky online behavior. A

longitudinal study relating digital literacy to online resilience

would also be interesting. A larger sample size and time period

would show whether the observed resilience is more than a

short-term coping mechanism.

VII. CONCLUSION

This paper is an analysis of teenage online behavior and

its associated cybersecurity risks, leveraging the Teenage

online behavior and cybersecurity risks dataset [9] to identify

key patterns and vulnerabilities. The data shows that digital

technology is an integral part of life for teenagers and that

there are significant challenges in ensuring their online safety.

After careful data cleaning, a number of calculated fields

were created. Examples include features such as: Time spent

online, age, parental oversight, and threat scoring. With this

data, we performed analyses such as Spearman correlations

and K-Means clustering. This allowed us to define three

distinct behavioral profiles among teenagers.

The profiles are: risk-aware users with strong security

behaviors (Cluster 0), low-engagement users with minimal

threat exposure (Cluster 1), and moderately engaged users

with some risk and limited supervision (Cluster 2).

These profiles gave us insight into how different levels of

digital activity, security practices, and parental oversight

influence a teenager's overall cybersecurity standing.

Our findings support several conclusions:

• Digital literacy is important for threat recognition but is

not sufficient to guarantee online safety. The digitally

literate spend more time online and are exposed to more

risk.

• The literature shows that strict parental measures are

less effective than parents helping teens to understand

and create boundaries to keep themselves safe.

• The different profiles provide a foundation for

personalized training and intervention, instead of the

traditional one-size-fits-all approach.

In conclusion, this analysis provides a framework for

creating effective, personalized strategies for raising

awareness and keeping adolescents safe online. It is hoped

that these insights can lead to informed, resilient teenagers

capable of making decisions and setting boundaries to keep

themselves secure online.

REFERENCES

[1] M. Adorjan and R. Ricciardelli, “Smartphone and social media

addiction: Exploring the perceptions and experiences of Canadian

teenagers,” Can. Rev. Sociol., vol. 58, no. 1, pp. 45–64, 2021. [Online].

Available: https://doi.org/10.1111/cars.12319

[2] D. Álvarez-García, T. García, and Z. Suárez-García, “The relationship

between Parental Control and high-risk internet behaviours in

adolescence,” Soc. Sci. (Basel), vol. 7, no. 6, p. 87, 2018. [Online].

Available: https://doi.org/10.3390/socsci7060087

[3] H. Kang, W. Shin, and J. Huang, “Teens’ privacy management on

video-sharing social media: the roles of perceived privacy risk and

parental mediation,” Internet Res., vol. 32, no. 1, pp. 312–334, 2022.

[Online]. Available: https://doi.org/10.1108/intr-01-2021-0005

[4] F. Mwagwabi and J. Hee Jiow, “Compliance with security guidelines in

teenagers: The conflicting role of peer influence and personal norms,”

Aust. J. Inf. Syst., vol. 25, 2021. [Online]. Available:

https://doi.org/10.3127/ajis.v25i0.2953

[5] D. Ondrušková and R. Pospíšil, “The good practices for

implementation of cyber security education for school children,”

Contemp. Educ. Technol., vol. 15, no. 3, p. ep435, 2023. [Online].

Available: https://doi.org/10.30935/cedtech/13253

[6] B. Ortega-Ruipérez, A. Castellanos Sánchez, and B. Marcano, “Risks in

adolescent adjustment by Internet exposure: Evidence from PISA,”

Front. Psychol., vol. 12, p. 763759, 2021. [Online]. Available:

https://doi.org/10.3389/fpsyg.2021.763759

[7] F. Quayyum, D. S. Cruzes, and L. Jaccheri, “Cybersecurity awareness

for children: A systematic literature review,” Int. J. Child Comput.

Interact., vol. 30, no. 100343, p. 100343, 2021. [Online]. Available:

https://doi.org/10.1016/j.ijcci.2021.100343

[8] E. Savoia, N. W. Harriman, M. Su, T. Cote, and N. Shortland,

“Adolescents’ exposure to online risks: Gender disparities and

vulnerabilities related to online behaviors,” Int. J. Environ. Res. Public

Health, vol. 18, no. 11, p. 5786, 2021. [Online]. Available:

https://doi.org/10.3390/ijerph18115786

2026 Journal of The Colloquium for Information Systems Security Education, Volume 13, No. 1, Spring 2026

©2026 The Author(s) | Creative Commons CC BY 4.0 9 www.cisse.info

[9] L. Sik, “Teenage online behavior and cybersecurity risks.” Kaggle,

2024. [Online]. Available:

https://doi.org/10.34740/KAGGLE/DSV/9587284

[10] K. Stewart, G. Brodowsky, and D. Sciglimpaglia, “Parental supervision

and control of adolescents’ problematic internet use: understanding

and predicting adoption of parental control software,” Young Consum.

Insight Ideas Responsible Mark., vol. 23, no. 2, pp. 213–232, 2022.

[Online]. Available: https://doi.org/10.1108/yc-04-2021-1307

[11] Hauke, J., & Kossowski, T. (2011). Comparison of values of Pearson's

and Spearman's correlation coefficients on the same sets of data.

Quaestiones geographicae, 30(2), 87-93.

[12] J. Vissenberg, L. d’Haenens, and S. Livingstone, “Digital literacy and

online resilience as facilitators of young people’s well-being?: A

systematic review,” Eur. Psychol., vol. 27, no. 2, pp. 76–85, 2022.

[Online]. Available: https://doi.org/10.1027/1016-9040/a000478

[13] S. Xu, S. Peng, and J. An, “TransDenseInceptionNet: A deep learning

framework for teenage cybersecurity awareness using real-world E-

safety data,” Informatica (Ljubl.), vol. 49, no. 18, 2025.