Digital privacy has always been an issue in the Internet age, with the widespread use of digital tools like smartphones, computers and social network platforms, through which many people inadvertently share their private information (Aguiar 2024). Due to a lack of digital literacy, today’s Internet users leave digital footprints across the Internet and traces of their activities (Valevaty & Ilyankou 2024). As much as science and technology have improved the lives of individuals, there are some challenges including those of digital footprints, which cyberbullies use to perpetrate victimisation (Evcimen & Öteleş 2025). The exposure of information communication tools in tertiary education, coupled with the heavy use of social media platforms, has intensified cyberbullying among higher education students (Bussu et al. 2025). While previously most studies focused on schools, overlooking universities, which are more engaged in cyberbullying, this study, therefore, focuses on university students (Shaikh et al. 2021). Most researchers have focused on high school students, leaving a gap in exploring other academic areas (Bussu et al. 2025). The issue faced is that researchers focus more on adolescents and middle school students, leaving a notable gap in data about universities facing cyberbullying (Arif et al. 2024). Most may assume that bullying is left behind in middle and secondary school, and this kind of behaviour is for adolescence and does not get any attention in higher education (Tight 2023). This study focuses on tertiary students.

This study’s objective is to uncover why students fall victim to cyberbullying and common risks of not complying with online privacy and digital footprint awareness.

Primarily, this study aims to equip tertiary students with the skills to stay alert, identify and avoid being cyberbullied. Also, universities are encouraged to start giving free mandatory fun and short courses to raise students’ awareness and skills.

Below, the researchers have created a free online course that can cater for students in the universities to self-study on their own time and must be completed by the end of the year:

National Qualification Framework (NQF) Level 5 – Entry Level:

This study sought to address the following research questions:

Ali and Shahbuddin (2022) note that cyberbullying is linked to traditional bullying because both bullying methods result in harm to the victim. Furthermore, bullying is a deliberate and repeated action of aggressive behaviour imposed on a specific individual (Ali & Shahbuddin 2022). An abusive tone on social media, because of the online disinhibition effect, increases the chances of cyberbullying. Bullying effects threaten both the physical and mental health of young adults (Li et al. 2024). López-Vizcaíno et al. (2021) note that cyberbullying is a dangerous phenomenon in society that severely impacts young adults and is becoming even more harmful because of the rapid growth of social tools. The issue with cyberbullying is that it can be hidden from the public because the victim can seem favourable and unharmed when it starts (Nazarov, Averbukh & Buinacheva 2022). This means that reactive intervention attempts such as cyberbullying moderation may not capture its initial occurrences, which delays remedial actions. According to Lv, Cao and Wang (2025), most students fall victim to hate speech incidents online. This may be a case of misconception or deliberate action intended to cause harm to an individual. Direct chat messages exchanged between two individuals are intended to clearly convey the sender’s meaning; however, due to the symbolic nature of text-based messages, different recipients may interpret the same message in varying ways, leading to an agreement or misunderstanding (Hasyim & Arafah 2023).

According to Khan et al. (2025), online bullying effects include stress, anxiety and low self-esteem in the victims. These negative effects may lead the individual to inflict self-harm and, in the worst cases, suicide (Young 2024). These types of harassment often leave a lifetime scar emotionally (Cohen 2025). Omokhabi et al. (2025) note that bullying effects can also lead to student isolation and poor academic results. Young (2024) states that because of the adoption of online systems and the transition to working fully online, most individuals face cyberbullying victimisation risks. In addition, studies show an increase in mental health risks and risky behaviours among the victims of cyberbullying (Gallegos et al. 2025). According to Sorrentino et al. (2023), one of the biggest risks encountered is not being aware of online risks, such as digital footprints.

All communications on social media platforms create a digital footprint, which can be recorded and analysed (Ophir, Asterhan & Schwarz 2019). Most students face cyberbullying because of digital negligence when on social media platforms, which has caused so much tension because the same data trail left online may be used to bully the same individual (Arafah 2023; Aguiar 2024). Social media has contributed negatively to online behaviour, with social platforms used as a tool to engage in cyberbullying activities (Ray, McDermott & Nicho 2024). The commonly used social media platforms include Instagram, WhatsApp, Facebook and TikTok, which are at hand at any time, allowing content sharing, whether good or bad, to spread widely (El Mezayen et al. 2025). With dependence on emerging technologies to communicate effectively and efficiently with one another, a gap has opened, which is soiled data or, as we call it in this study, online privacy. The increase in data sharing on social media platforms has placed the rights and protection of data at risk, which raises a big question on data security and privacy (Yadav et al. 2022). With the constant evolution of social media platforms and real-time content updates, social networks present security and privacy risks (Paul, Samanta & Koner 2025).

Digital society is growing rapidly because of technological advancements; as technology evolves, more digital identities emerge. Every movement online links to an individual’s identity, finally creating what is known as a digital footprint (Vilar et al. 2024). Vilar et al. (2024) note that a digital footprint is the distinct data trail left by the individual while surfing the Internet or interacting with others on social media. Evcimen and Öteleş (2025) posit that browsing history, location-based services (LBS) data, communication history on social media and point-of-sale (POS) records form a digital footprint that is used to attack victims.

Students access the Internet from university networks or at home, then use online or social media applications to post content and data. Some data are collected by the applications and websites as historical data, and some are live data displayed to everyone, for example, family vacation, cell phone number and address. While individuals post the content online, it is a form of negligence by the victim, which now becomes a data trail called a ‘digital footprint’, and bullies can collect the data as leverage or a bargaining chip to use against the victim (Figure 1).

Tools can be created to prevent cyberbullying, but according to this study, the most effective way of mitigation is the traditional method: educating the learners about digital literacy and how to safely navigate sites and social activities. According to Tetteh, Awaah and Addo (2023), it is crucial to educate the students about digital literacy at an early stage. Similarly, Yosep, Hikmat and Mardhiyah (2023) note that there is a need for e-parenting. This way, students will be cyber aware and surf the Internet more diligently.

This study employed a pragmatic philosophical stance, which aligns with the quantitative methodological approach to examine real-life practices and theories within a practical context. Brister (2024) posits that a philosophical pragmatic method combines theory and practical perspectives, which provides firm guidance for research. It aims to assess ongoing experimentation, which can change over time (Lake & Wills 2020).

A deductive reasoning approach is adopted to quantify the raw data through a scientific method of analysis to allow the researcher to test the hypothesis (Kumar & Ujire 2024). This approach relates to the quantitative method selected for this study, which explores theories and validates the outcome by performing rigorous tests on the numeric data (Masadeh 2024).

This study applied a convenience sampling method because it is an efficient, time-saving, and effective data collection approach at a low cost. According to Sarstedt et al. (2018), convenience sampling method enables a larger group of potential respondents, overcoming the time constraints that are normally faced when doing research.

This study targeted the higher education sector by soliciting tertiary young adults’ perceptions. The study focused on students who used social media regularly, aged between 18 and 45 years.

Figure 2 shows all the sampling methods available and the one this paper will be employing in this study.

The data were collected using an online survey from students in various higher education institutions in South Africa. Online surveys gather data samples from a selected group of participants and provide preliminary analysis of the data in relation to the research objective (Manstein, Shiah & Laikhter 2023). The online questionnaire was based on the Likert-scale style, which is in line with the quantitative methodology applied in the study. According to Nemoto and Beglar (2014), the Likert-scale is used to measure data effectively and allows the researcher to gather a larger number of responses with ease.

The data collected from the higher education respondents were downloaded in an Excel format for closer inspection of missing and incomplete responses before analysis. The Statistical Package for the Social Sciences (SPSS version 29) (IBM Corp., Armonk, NY, US) was used to analyse the data.

Ethical clearance to conduct this study was obtained from the College of Business and Economics (CBE) Research Ethics Committee, Department of Applied Information Systems, University of Johannesburg (Ref. No. 2024AIS065).

The results in Table 4 show that the majority of respondents agree or strongly agree with the goal of educating students about online privacy. These results indicate that educating students about cyberbullying is favoured by most respondents (84.1%). Notably, 22% of the respondents disagreed with avoiding open-source chat groups, highlighting it as the least supported statement in this section, with the lowest mean score of 3.25. Not sharing immoral images and clips with a stranger or partner and not sharing personal information online have the same mean and standard deviation of around 1.19 and a mean of 4.24 with a high agreement of 79% and 80%, respectively, which shows the importance of not leaving digital footprints online.

To address the reliability and validity of the Likert-scale measures used in this study, Cronbach’s alpha was calculated for each construct, including the online privacy scale reported above and three additional scales collected but not previously reported: social media beliefs, cyberbullying beliefs and mitigation measures. An EFA was then conducted on the combined online privacy, social media and cyberbullying items (Table 5).

All five constructs exceeded the recommended threshold of 0.70 (Nunnally 1978), with online privacy and social media showing excellent internal consistency (α > 0.90). A Kaiser-Meyer-Olkin value of 0.952 indicates the sample is highly adequate for factor analysis, and Bartlett’s test of sphericity confirms the correlation matrix is not an identity matrix (χ²(231) = 4522.67, p < 0.001), justifying the use of EFA (Table 6). Online privacy and cyberbullying beliefs were significantly correlated (r = 0.574, p < 0.001), as were social media beliefs and cyberbullying beliefs (r = 0.733, p < 0.001).

Three factors emerged with eigenvalues greater than 1, together explaining 65.41% of total variance (factor 1 = 49.42%, factor 2 = 10.01%, factor 3 = 5.99%). The rotated solution confirms the intended three-construct structure of the instrument, with online privacy, social media beliefs and cyberbullying beliefs items loading cleanly onto their respective factors, supporting the construct validity of the measurement model. Loadings below 0.30 are suppressed for readability.

This study examined whether online privacy, type of attacks, commonly used applications and social media beliefs predict cyberbullying, using multiple linear regression (chosen because the outcome, cyberbullying, is a continuous scale variable predicted by several continuous predictors; a one-way analysis of variance [ANOVA] is deemed appropriate for comparing group means on a categorical factor) (Table 7). Prior to the final model, 13 multivariate outliers were identified via Mahalanobis Distance and removed, reducing the analytic sample from 289 to 276. Table 8 shows that the overall model was statistically significant, F(4, 271) = 68.797, p < 0.001, explaining 50.4% of the variance in cyberbullying (R² = 0.504, Adjusted R² = 0.497). Social media beliefs were the strongest predictor (β = 0.560, p < 0.001), followed by online privacy, which was a significant negative predictor (β = −0.144, p = 0.009), indicating that greater privacy awareness is associated with lower reported cyberbullying. Commonly used applications was also a significant predictor (β = 0.133, p = 0.004). Type of attacks did not significantly predict cyberbullying (p = 0.611). No multicollinearity was present (all variance inflation factor [VIF] < 2, Tolerance > 0.59).

This study focused not on digital trails (footprints) but on risks associated with these trails falling into the hands of cyberbullying perpetrators. Future studies should focus more on digital footprints as one of the prominent ways that bullies use to corner their victims and use a longitudinal study timeframe.

There is an urgent need for universities and colleges to implement a short and fun course that will educate students and equip them with the skill to identify and avoid cyberbullying. These courses could provide training on managing digital footprints, an important risk factor in cyberbullying because every piece of historical data online links to a certain individual. Most students face intense pressure without any help; this raises concerns about their mental, emotional and physical well-being. Higher education institutions should devise intervention programmes by design to help these students cope with these anxieties and stressors. This will enable mitigation of cyberbullying and ensure compliance with online privacy, making it harder for bullies to attack because they will lack the information that empowers them to bully other students. The results showed strong correlations among the models that influence cyberbullying. More focus should be on the online privacy model because if students are educated about online privacy, then cyberbullying can be mitigated, and students will learn what to share and not to share online, how to stay safe, and how not to leave personal data trails. On the other hand, we need to find a safer and more secure way to protect a large quantity of data on social media, because this interaction system produces so much data per day, which raises concerns about how that data can be secured online.