Cyberbullying among medical school students: an examination of the relationship between frequency, sensitivity, and psychological symptoms: a cross-sectional study

Abstract

Background

Cyberbullying is intentional, repeated harm carried out through electronic devices. Victims may develop anxiety, depression, a negative self-concept, and poorer academic performance.

Methods

This cross-sectional study examined the prevalence of cyberbullying victimisation and perpetration, cyberbullying sensitivity, and the associated psychological symptoms among 259 medical students at Ankara University. Data were collected using a 25-item questionnaire, the Revised Cyberbullying Inventory for University Students, the Cyberbullying Sensitivity Scale, and the Brief Symptom Inventory (BSI); analyses were conducted in SPSS 30.

Results

Overall, 28.2% of participants reported being victims of cyberbullying (victimisation) and 9.7% reported perpetrating it; 3.2% of victims reported suicidal thoughts. Victimisation was more common among third-year students, heavy internet users (> 6 h/day), and users of friendship and betting platforms, as well as among students who had themselves perpetrated cyberbullying. Perpetration was more common among males, third-year students, frequent users of gaming, friendship, and betting platforms, and those reporting victimisation at least once every 6–12 months. For psychological symptoms, women had higher depression and somatisation scores. Anxiety, depression, somatisation, and negative self-concept scores were higher among third-year students. BSI scores rose as the frequency of victimisation increased, and students reporting prior perpetration had higher hostility scores. In logistic regression, victimisation was associated with higher odds among third-year students (OR 2.96), those who met new people online (OR 3.09), and those who reported harassing others online (OR 27.15); perpetration was associated with higher odds among males (OR 16.32), betting-website users (OR 10.79), those victimised at least once a year (OR 24.86), and those victimised at least once every six months (OR 39.17).

Conclusions

Cyberbullying victimisation and perpetration were associated with distinct psychosocial factors, online behaviours, and psychological symptoms. Medical schools should establish units that implement holistic prevention strategies incorporating psychosocial support, early identification of at-risk students, and training on safe online behaviour.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12889-026-28421-x.

Introduction

Cyberbullying can be defined as “intentional and repeated harm inflicted through the use of computers, mobile phones, and other electronic devices”. This type of bullying can be seen on social media, messaging, and gaming platforms [1]. Cyberbullying can occur in different ways, including online harassment, posting offensive or abusive messages or images, privacy violations, and exclusion from online platforms or groups [2]. Examples include sharing nude photographs of someone without consent; falsely accusing someone online and encouraging others to cyberbully them; encouraging self-harm or suicide; bullying individuals because they belong to the LGBTQ community; revealing personal information following disagreements; and creating fake identity profiles to mislead and exploit people [3]. A related behaviour often discussed alongside cyberbullying is trolling, which refers to posting provocative statements, posts, or images on an online portal to deliberately provoke anger and extreme reactions; unlike the behaviours above, trolling is not necessarily directed at a specific individual [4].

There are individual and environmental risk factors associated with cyberbullying. One of the main causes is insufficient communication within the family and limited interaction between parents. A lack of empathy, an inability to consider that one’s behaviour may hurt another person, and an inability to foresee the consequences of one’s actions are also contributing factors. A person may not realise how damaging an action that would not be carried out face-to-face can be when performed in a digital environment. As a result, an individual who does not consider this may engage in bullying for amusement or to pass the time, without realising or accepting that what they are doing constitutes bullying. Other contributing factors include easy access to the internet, spending long periods on social media, and accepting friend requests from strangers [5].

Researchers have found that cyberbullying is associated with various negative consequences for both victims and perpetrators. For victims, these include higher levels of anxiety, depression, and loneliness, increased suicidal thoughts, lower self-esteem, and poorer academic performance. Victims may experience difficulties in their relationships with peers and become socially isolated. They may avoid or skip school because they do not feel safe there, and may be more prone to crime and risky behaviour. They may also experience somatic symptoms such as difficulty falling and staying asleep, headaches, and stomach aches, as well as eating disorders and chronic fatigue [6, 7].

Cyberbullying provokes negative emotions such as sadness, anger, disappointment, shame, and fear; the susceptibility of those exposed to it to social anxiety and depression therefore warrants investigation. The sensitivity and attitudes of individuals towards cyberbullying are another important issue. Given the high prevalence of cyberbullying and its impact on individuals, implementing preventive measures is critically important, and the first step in such efforts is to identify the risk and protective factors associated with both perpetrating cyberbullying and being subjected to it [8].

The present study adopted the broad conceptualisation, defining cyberbullying as any intentional and repeated behaviour carried out through digital technologies—social media, messaging and gaming platforms, and mobile phones—that is intended to frighten, anger, or embarrass others; examples include spreading lies or posting embarrassing photographs of someone on social media, sending hurtful or threatening messages, and sending malicious messages to others while impersonating someone. Consistent with this broad view, provocative behaviours such as trolling which are intended to provoke or distress others were also treated as cyberbullying.

Most research on cyberbullying has focused on children and adolescents in primary and secondary education, whereas far less is known about university students, even though they are intensive users of digital media and face distinct academic and social pressures. This relative scarcity of evidence in higher-education settings is an important gap, particularly for medical students, who face high academic demands that may shape both their involvement in cyberbullying and its consequences. Examining cyberbullying in this understudied population is therefore both timely and necessary.

Against this background, the primary aim of this study was to estimate the prevalence of cyberbullying victimisation and perpetration among medical students at Ankara University and to identify the associated risk and protective factors. As secondary aims, the study examined the psychological symptoms (anxiety, depression, somatisation, negative self-concept, and hostility) associated with involvement in cyberbullying, as well as students’ sensitivity to cyberbullying. By clarifying these associations, the study also seeks to raise awareness of cyberbullying in higher education and to inform future preventive and supportive interventions.

Methods

Participants

This was a cross-sectional study of second-, third-, and sixth-year students enrolled at Ankara University Medical School during the 2024–2025 academic year. Using G*Power 3.1, a minimum sample of 79 students per year of study (237 in total) was required to detect a correlation of 0.20 with 95% power at a 5% type I error rate. A total of 259 students participated: 85 s-year, 88 third-year, and 86 sixth-year students. The mean age was 21.4 ± 2.0 years (median 21; range 17–27), and the mean age at first internet use was 10.4 ± 3.1 years (median 10). The full sociodemographic profile of the sample is summarised in Table 1.

Table 1.

Cyberbullying victimisation by participants’ descriptive characteristics

			Cyberbullying victimisation		Effect size	p*
			No, n (%)	Yes, n (%)
Gender	Female Male		97 (71.9) 89 (71.8)	38 (28.1) 35 (28.2)	0.00 (φ)	0.989
Class	2nd year 3rd year 6th year		70 (82.4) 52 (59.1) 64 (74.4)	15 (17.6)a 36 (40.9)a 22 (25.6)	0.22 (Cramér’s V)	0.002
Grade point average	2.0–3.0 3.0-3.5 3.5-4.0		60 (75.9) 95 (71.4) 31 (66.0)	19 (24.1) 38 (28.6) 16 (34.0)	0.08 (Cramér’s V)	0.479
Presence of chronic/psychiatric disease	No Yes		166 (72.2) 20 (69.0)	64 (27.8) 9 (31.0)	0.02 (φ)	0.717
Daily internet usage time	1–2 h 3–4 h 5–6 h More than 6 h		17 (77.3) 89 (79.5) 53 (67.9) 27 (57.4)	5 (22.7) 23 (20.5)a 25 (32.1) 20 (42.6)a	0.19 (Cramér’s V)	0.030
Meeting new people online	No Yes		52 (86.7) 134 (67.3)	8 (13.3) 65 (32.7)	0.18 (φ)	0.004
Purpose of internet usage	Playing games	No Yes	99 (76.7) 87 (66.9)	30 (23.3) 43 (33.1)	0.11 (φ)	0.079
	Social media	No Yes	5 (83.3) 181 (71.5)	1 (16.7) 72 (28.5)	0.04 (φ)	1.000
	Education, homework, research	No Yes	16 (76.2) 170 (71.4)	5 (23.8) 68 (28.6)	0.03 (φ)	0.642
	Chat	No Yes	56 (78.9) 130 (69.1)	15 (21.1) 58 (30.9)	0.10 (φ)	0.121
	Friendship websites	No Yes	183 (72.9) 3 (37.5)	68 (27.1) 5 (62.5)	0.14 (φ)	0.042
	Shopping websites	No Yes	53 (77.9) 133 (69.6)	15 (22.1) 58 (30.4)	0.08 (φ)	0.191
	Newspaper, reading the news	No Yes	78 (65.0) 108 (77.7)	42 (35.0) 31 (22.3)	0.14 (φ)	0.024
	TV, video, music	No Yes	32 (71.1) 154 (72.0)	13 (28.9) 60 (28.0)	0.01 (φ)	0.908
	Betting websites	No Yes	179 (73.4) 7 (46.7)	65 (26.6) 8 (53.3)	0.14 (φ)	0.037
Cyberbullying perpetration	No Yes		183 (78.2) 3 (12.0)	51 (21.8) 22 (88.0)	0.43 (φ)	< 0.001
Know the person who bullied them	No Yes		75 (55.1) 4 (25.0)	61 (44.9) 12 (75.0)	0.19 (φ)	0.022
Hiding name / identity when using internet	No Yes		53 (74.6) 133 (70.7)	18 (25.4) 55 (29.3)	0.04 (φ)	0.533

n frequency, %: row percentage *chi-square tests, φ = phi; V = Cramér’s V, Bold = statistically significant (p < 0.05)

†The manuscript reported Fisher’s exact p for rows with small expected counts; the effect size is computed from the same table and is unaffected

Before the lecture, students were informed about the study and sent a Google Forms link, and data were collected under supervision from those who agreed to participate. Of the 278 students invited (90 in year 3, 92 in year 2, and 96 in year 6), 259 completed the questionnaire, giving a response rate of 93.2%.

Procedure and consent

Permission to use the scales was obtained from the scale developers by e-mail, and ethical approval was granted by the Ankara University Faculty of Medicine Ethics Committee (date: 03.04.2024; decision no: I03-246-24). Participation was voluntary, and verbal and written informed consent was obtained from all participants.

Instruments

Data were collected online via Google Forms. A link was sent to all eligible students, and only those who provided consent were included. Each participant completed a sociodemographic and contextual questionnaire followed by three validated instruments: the Revised Cyberbullying Inventory for University Students, the Cyberbullying Sensitivity Scale, and the Brief Symptom Inventory (BSI).

Sociodemographic and contextual questionnaire

This investigator-developed form was based on a review of the literature and comprised 25 items capturing only sociodemographic and contextual information: age, sex, and other demographic characteristics; the frequency of internet use via mobile phone or computer; exposure to and witnessing of cyberbullying; the reactions of those who had been exposed; and the settings in which cyberbullying occurred. These 25 items did not include the cyberbullying, sensitivity, or symptom scales, which were administered separately; the total number of items completed by each participant was therefore considerably greater than 25.

At the beginning of the questionnaire, before responding to any items, all participants were presented with the following definition of cyberbullying so that their responses would be based on a shared understanding of the construct (translated from the Turkish original administered to participants): “Cyberbullying is bullying carried out using digital technologies. This type of bullying can occur on social media, messaging platforms, gaming platforms, and mobile phones. It is a repeated behaviour intended to frighten, anger, or embarrass a targeted person. Examples include spreading lies or posting embarrassing photographs about a person on social media, sending hurtful messages or threats through messaging platforms, and sending malicious messages to others while impersonating someone else.”

Revised cyberbullying inventory for university students

Cyberbullying victimisation and perpetration were assessed using the Revised Cyberbullying Inventory for University Students. This instrument has a layered development history: the original Cyber Bullying Inventory (CBI) was developed by Erdur-Baker and Kavşut (2007) to measure cyberbullying experiences among adolescents, using two parallel forms for cyberbullying and cybervictimisation [9]. Topçu and Erdur-Baker (2010) subsequently revised this inventory and reported the validity and reliability of the Revised Cyberbullying Inventory (RCBI) [10]. Tanrıkulu (2015) then adapted the RCBI for university students, producing the version used in the present study [11]. The inventory comprises 12 items presented in two parallel sections that contain the same items but differ in framing: in the first section participants indicate whether they have engaged in each behaviour (“I did it”), and in the second whether each behaviour has been done to them (“it was done to me”), in both cases over the previous six months. Each item is rated on a four-point scale (1 = never, 2 = once, 3 = two or three times, 4 = more than three times), with higher scores indicating more frequent perpetration or victimisation. In the university-student adaptation, the internal consistency coefficient was 0.80 for the perpetration (cyberbullying) section and 0.73 for the victimisation Sect [11].

Operationalisation and classification of victims and perpetrators

According to the definition given at the beginning of the questionnaire participants answered two direct questions: whether they had been subjected to cyberbullying (“Have you been subjected to cyberbullying online?“) and whether they had engaged in it (“Have you displayed cyberbullying behaviours towards others?“), each with a yes/no response. Those who answered “yes” were additionally asked how frequently this had occurred. Victim and perpetrator status was determined from these self-report questions, not from a cut-off applied to the Revised Cyberbullying Inventory; the inventory was used to determine the difference in scale scores between specific characteristics.

Cyberbullying sensitivity scale

This scale was developed by Tanrıkulu, Kınay, and Arıcak in 2013 [12]. After two rounds of analysis, in which two items were removed to ensure unidimensionality and adequate factor loadings, the scale took its final form of 13 items loading on a single factor. Items are scored as no = 1, sometimes = 2, and yes = 3, with higher total scores indicating greater sensitivity to cyberbullying; internal consistency was 0.87 [12]. Conceptually, cyberbullying sensitivity reflects an individual’s awareness of, and attitudes towards, behaviours that constitute cyberbullying. Greater sensitivity may be associated with a lower likelihood of perpetration and with different appraisals of, and responses to, victimisation. Including this construct therefore allowed the study to examine whether awareness of cyberbullying is related to students’ involvement in it, complementing the prevalence and risk-factor analyses and addressing the role of sensitivity highlighted in the study’s title and aims.

Brief symptom inventory (BSI)

It is a symptom screening scale developed by Derogatis [13]. It consists of 53 items selected from the 90 items of the Symptom Checklist-90-R (SCL-90-R), which was developed to identify psychiatric problems in various medical conditions. The validity and reliability study of the scale for Turkish society was conducted by Şahin and Durak [14]. The BSI consists of 5 subscales. These subscales are somatisation, depression, anxiety, negative self-concept, hostility.

Statistical analysis

Data analysis was performed using the SPSS 30 software package. Descriptive statistics are presented as frequency, percentage, mean, standard deviation, median, minimum, and maximum. The analysis of categorical variables was performed using the Pearson chi-square test and, where necessary, the Fisher’s exact test. The φ coefficient (for 2 × 2 tables) or Cramér’s V (for larger tables) was reported as the effect size for these comparisons.To understand which group the difference originated from in comparisons involving more than two groups, standardised residual values were examined. It has been accepted that the difference is formed by those with a standardised residual value lower or higher than ± 1.96. The normality of continuous variables was examined using visual (histograms and probability plots) and analytical methods (Kolmogorov-Smirnov and Shapiro-Wilk Tests). These variables were compared using Student’s t-test and ANOVA when they followed a normal distribution, and using the Mann-Whitney U test and Kruskal-Wallis test when they did not follow a normal distribution. Cohen’s d and η² (for parametric tests) or the rank-biserial correlation and ε² (for non-parametric tests) were reported as the corresponding effect sizes.

The relationship between variables was assessed using Pearson and Spearman correlation tests. The predictive power of independent variables for experiencing and perpetrating cyberbullying was assessed using logistic regression analysis. Prior to the logistic regression analysis, the model assumptions were assessed. It was found that the assumption of independent observations was met; multicollinearity among the variables was examined, and no significant multicollinearity was detected. The logit linearity of the continuous variables was assessed, and it was determined that the model assumptions were met. Model fit was evaluated using the omnibus likelihood-ratio test and the Hosmer–Lemeshow goodness-of-fit test, and the explained variance was summarised using the Cox & Snell and Nagelkerke R² statistics.For all analyses, a significance level of p < 0.05 was accepted.

Results

This study included 85 s-year, 88 third-year, and 86 sixth-year students enrolled in the medical school. Overall, 28.2% of participants reported cyberbullying victimisation and 9.7% reported perpetration. Victim and perpetrator status was based on participants’ yes/no responses to the two direct questions asked immediately after the cyberbullying definition (see Methods), rather than on a cut-off applied to the Revised Cyberbullying Inventory. Because the definition explicitly described cyberbullying as a repeated behaviour, these self-classifications reflect a repetition-based understanding of the construct.

Table 1 compares cyberbullying victimisation across participants’ descriptive characteristics. Victimisation differed significantly by year of study (p < 0.05), and the post-hoc comparison showed that the difference was between second- and third-year students, with third-year students being more frequently victimised (40.9% vs. 17.6%); the sixth-year group did not differ significantly from either. Victimisation also differed by daily internet use (p < 0.05), where the difference was driven by the comparison between those using the internet for 3–4 h and those using it for more than 6 h per day, being higher among the heaviest users (42.6% vs. 20.5%). Victimisation was also significantly more common among participants who met new people online, those who used friendship websites, those who used betting websites, those who had themselves perpetrated cyberbullying, and those who knew the person who had bullied them (all p < 0.05). In contrast, victimisation was significantly less common among those who used the internet to read newspapers/news (p < 0.05), suggesting a protective association. No significant differences were found for the remaining variables (p > 0.05).

Table 2 compares cyberbullying perpetration across participants’ characteristics. Perpetration differed significantly by sex, being more common among men than women (p < 0.05). It also differed by year of study (p < 0.05); the post-hoc comparison showed that the difference was between third- and sixth-year students, with third-year students reporting more perpetration (15.9% vs. 5.8%), while the second-year group did not differ significantly from either. Perpetration also differed by the frequency of victimisation (p < 0.05); the post-hoc comparison showed that participants who had never been victimised reported significantly less perpetration (2.3%) than both those victimised at least once a year (18.6%) and those victimised at least once every six months (35.7%), while the two victimised groups did not differ significantly from each other. In addition, perpetration was significantly more common among participants who used the internet for gaming, those who used friendship websites, and those who used betting websites (all p < 0.05). No significant differences were found for the remaining variables (p > 0.05).

Table 2.

Cyberbullying perpetration by participants’ descriptive characteristics

			Cyberbullying perpetration		Effect size	p
			No, n (%)	Yes, n (%)
Gender	Female Male		133 (98.5) 101 (81.5)	2 (1.5) 23 (18.5)	0.29 (φ)	< 0.001
Class	2nd year 3rd year 6th year		79 (92.9) 74 (84.1) 81 (94.2)	6 (7.1) 14 (15.9)a 5 (5.8)a	0.15 (Cramér’s V)	0.048
The place where he/she lives	Home With family/relatives Dormitory		58 (87.9) 71 (85.5) 105 (95.5)	8 (12.1) 12 (14.5) 5 (4.5)	0.15 (Cramér’s V)	0.051
Grade point average	2,0–3,0 3,0–3,5 3,5 − 4,0		71 (89.9) 121 (91.0) 42 (89.4)	8 (10.1) 12 (9.0) 5 (10.6)	0.02 (Cramér’s V)	0.936
Presence of chronic/psychiatric disease	No Yes		210 (91.3) 24 (82.8)	20 (9.7) 5 (17.2)	0.09 (φ)	0.174
Daily internet usage time	1–2 h 3–4 h 5–6 h More than 6 h		21 (95.5) 104 (92.9) 70 (89.7) 39 (83.0)	1 (4.5) 8 (7.1) 8 (10.3) 8 (17.0)	0.13 (Cramér’s V)	0.219
Meeting new people online	No Yes		58 (96.7) 176 (88.4)	2 (3.3) 23 (11.6)	0.12 (φ)	0.059
Purpose of internet usage	Playing games	No Yes	126 (97.7) 108 (83.1)	3 (2.3) 22 (16.9)	0.25 (φ)	< 0.001
	Social media	No Yes	6 (100.0) 228 (90.1)	0 (0.0) 25 (9.9)	0.05 (φ)	1.000
	Education, homework, research	No Yes	19 (90.5) 215 (90.3)	2 (9.5) 23 (9.7)	0.00 (φ)	1.000
	Chat	No Yes	67 (94.4) 167 (88.8)	4 (5.6) 21 (11.2)	0.08 (φ)	0.178
	Friendship websites	No Yes	230 (91.6) 4 (50.0)	21 (8.4) 4 (50.0)	0.24 (φ)	< 0.001
	Shopping websites	No Yes	64 (94.1) 170 (89.0)	4 (5.9) 21 (11.0)	0.08 (φ)	0.220
	Newspaper, reading the news	No Yes	110 (91.7) 124 (89.2)	10 (8.3) 15 (10.8)	0.04 (φ)	0.504
	TV, video, music	No Yes	43 (95.6) 191 (89.3)	2 (4.4) 23 (10.7)	0.08 (φ)	0.270
	Betting websites	No Yes	227 (93.0) 7 (46.7)	17 (7.0) 8 (53.3)	0.37 (φ)	< 0.001
Frequency of victimisation	At least once every 6 months At least once a year Never		18 (64.3) 48 (81.4) 168 (97.7)	10 (35.7)a 11 (18.6)b 4 (2.3)ab	0.38 (Cramér’s V)	< 0.001
Know the person who bullied them	No Yes		113 (83.1) 15 (93.8)	23 (16.9) 1 (6.2)	0.09 (φ)	0 0.470
Hiding name/identity when using internet	No Yes		169 (89.9) 65 (91.5)	19 (10.1) 6 (8.5)	0.03 (φ)	0.687

n frequency, %: row percentage φ = phi; V = Cramér’s V, Bold = statistically significant (p < 0.05)

Average scores for the Revised Cyberbullying Inventory for University Students; 13.2 ± 2.9 for the Cyberbullying subscale (median: 12, min: 12, max: 40), 13.8 ± 3.8 (median: 12, min: 12, max: 42) for Cyber Victimization subscale, while the average score for the Cyberbullying Sensitivity scale is 32.2 ± 5.7 (median: 33, min: 13, max: 39) (Table S1).

According to a question in the questionnaire, 3.2% of participants subjected to cyberbullying had suicidal thoughts. Table 3 shows the comparison of BSI scores according to descriptive characteristics. Women scored significantly higher than men on depression and somatisation (p < 0.05). Scores differed significantly by year of study on anxiety, depression, somatisation, and negative self-concept (p < 0.05); the post-hoc comparisons showed that, for anxiety, somatisation, and negative self-concept, the difference was between third- and sixth-year students, with third-year students scoring higher, whereas for depression the sixth-year group scored significantly lower than both the second- and third-year groups. Anxiety was significantly higher among participants who reported cyberbullying victimisation (p < 0.05). Scores also differed by the frequency of victimisation (p < 0.05); the post-hoc comparisons showed that, for anxiety, depression, and negative self-concept, participants who had never been victimised scored significantly lower than both victimised groups, for hostility the never-victimised group differed from the most frequently victimised group (at least once every six months), and for somatisation the difference was between the “at least once a year” and “never” groups. Participants who had witnessed cyberbullying online scored significantly higher on all five subscales (p < 0.05), and those who had perpetrated cyberbullying scored significantly higher on hostility (p < 0.05).

Table 3.

Brief Symptom Inventory (BSI) subscale scores by participants’ descriptive characteristics

		BSI 1 Anxiety	BSI 2 Depression	BSI 3 Somatization	BSI 4 Negative self-consept	BSI 5 Hostility
Gender
Female	Mean ± SD Median	0.78 ± 0.72 0.62	1.13 ± 0.81 0.92	0.69 ± 0.69 0.56	0.89 ± 0.82 0.75	1,00 ± 0.76 0.86
Male	Mean ± SD Median	0.66 ± 0.69 0.46	0.91 ± 0.77 0.75	0.53 ± 0.59 0.33	0.71 ± 0.75 0.50	0.91 ± 0.74 0.86
Effect Sizes (rank-biserial r)		0.12	0.17	0.17	0.14	0.07
p		0.104	0.021	0.019	0.051	0.337
Class
2nd year	Mean ± SD Median	0.75 ± 0.75 0.46	1.10 ± 0.86 0.83ᵇ	0.62 ± 0.66 0.44	0.85 ± 0.86 0.58	0.97 ± 0.74 0.86
3rd year	Mean ± SD Median	0.90 ± 0.73 0.77ᵃ	1.19 ± 0.80 1.00ᵃ	0.75 ± 0.69 0.61ᵃ	0.95 ± 0.80 0.75ᵃ	1.07 ± 0.79 1.00
6th year	Mean ± SD Median	0.53 ± 0.58 0.38ᵃ	0.78 ± 0.67 0.67ᵃᵇ	0.47 ± 0.55 0.28ᵃ	0.60 ± 0.67 0.33ᵃ	0.83 ± 0.71 0.57
Effect Sizes (ε²)		0.057	0.051	0.036	0.045	0.020
p		< 0.001	0.001	0.009	0.003	0.071
Cyberbullying victimisation
No	Mean ± SD Median	0.66 ± 0.64 0.54	0.96 ± 0.75 0.83	0.57 ± 0.62 0.39	0.74 ± 0.72 0.50	0.89 ± 0.67 0.86
Yes	Mean ± SD Median	0.88 ± 0.82 0.69	1.19 ± 0.88 1.00	0.72 ± 0.71 0.56	0.97 ± 0.93 0.67	1.14 ± 0.89 0.86
Effect Sizes (rank-biserial r)		0.16	0.14	0.14	0.15	0.14
p		0.045	0.080	0.072	0.067	0.081
Frequency of victimisation
Every day/at least once every 6 months	Mean ± SD Median	1.05 ± 0.91 0.85ᵃ	1.35 ± 0.93 1.13ᵃ	0.82 ± 0.79 0.56	1.18 ± 1.01 0.92ᵃ	1.41 ± 0.95 1.29ᵃ
At least once a year	Mean ± SD Median	0.88 ± 0.78 0.69ᵇ	1.23 ± 0.88 0.92ᵇ	0.71 ± 0.65 0.56a	1.01 ± 0.90 0.75ᵇ	1.09 ± 0.80 0.86
Never	Mean ± SD Median	0.62 ± 0.61 0.38ᵃᵇ	0.90 ± 0.71 0.75ᵃᵇ	0.54 ± 0.61 0.33a	0.67 ± 0.67 0.50ᵃᵇ	0.83 ± 0.66 0.71ᵃ
Effect Sizes (ε²)		0.043	0.042	0.033	0.049	0.050
p		0.004	0.005	0.014	0.002	0.002
Witnessing cyberbullying online
No	Mean ± SD Median	0.51 ± 0.58 0.31	0.72 ± 0.68 0.67	0.43 ± 0.55 0.17	0.54 ± 0.63 0.25	0.74 ± 0.69 0.57
Yes	Mean ± SD Median	0.84 ± 0.74 0.62	1.17 ± 0.81 1.00	0.71 ± 0.66 0.56	0.93 ± 0.83 0.75	1.07 ± 0.76 0.86
Effect Sizes (rank-biserial r)		0.31	0.34	0.31	0.32	0.28
p		< 0.001	< 0.001	< 0.001	< 0.001	< 0.001
Cyberbullying perpetration
No	Mean ± SD Median	0.71 ± 0.70 0.54	1.00 ± 0.80 0.83	0.62 ± 0.65 0.44	0.78 ± 0.78 0.54	0.91 ± 0.72 0.86
Yes	Mean ± SD Median	0.85 ± 0.72 0.69	1.19 ± 0.80 1.00	0.57 ± 0.58 0.33	0.99 ± 0.87 0.67	1.33 ± 0.95 1.14
Effect Sizes (rank-biserial r)		0.15	0.16	0.01	0.19	0.26
p		0.225	0.181	0.937	0.121	0.030

SD Standard deviation, p significance level, medians sharing the same superscript letter ^{(a, b)} differ significantly from each other in pairwise comparison (p < 0.05), values are rank-biserial r (two-group rows: Mann–Whitney U) or ε² (three-group rows: Kruskal–Wallis)

* statistically significant (p < 0.05)

Table 4 presents the binary logistic regression models examining the factors associated with cyberbullying victimisation and perpetration. In the victimisation model, higher odds of victimisation were independently associated with being a third-year student (compared with second-year students; OR = 2.96), meeting new people online (OR = 3.09), and having perpetrated cyberbullying (OR = 27.15) (all p < 0.05). In contrast, using the internet to read newspapers/news was associated with lower odds of victimisation (OR = 0.30; p < 0.05), indicating a protective association, whereas the use of shopping websites showed only a borderline, non-significant association (OR = 2.24; p = 0.051).

Table 4.

Logistic regression analysis of factors associated with cyberbullying victimisation and perpetration

		OR	95% CI	p
Cyberbullying victimisation	Class 3rd year (ref. 2nd year) 6th year (ref. 2nd year)	2.96 1.93	1.302–6.728 0.819–4.540	0.010 0.133
	Meeting new people online Yes (ref. No)	3.09	1.252–7.617	0.014
	Purpose of internet use: shopping websites Yes (ref. No)	2.24	0.997–5.047	0.051
	Purpose of internet use: reading newspapers/news Yes (ref. No)	0.30	0.147–0.596	< 0.001
	Cyberbullying perpetration Yes (ref. No)	27.15	7.179-102.655	< 0.001
Cyberbullying perpetration	Gender Male (ref. Female)	16.32	3.052–87.305	0.001
	Class 3rd year (ref. 2nd year) 6th year (ref. 2nd year)	0.36 0.13	0.073–1.741 0.018–0.882	0.202 0.037
	The place where he/she lives With family/relatives (ref. Home) Dormitory (ref. Home)	0.96 0.16	0.222–4.175 0.025–1.038	0.959 0.055
	Purpose of internet use: chat Yes (ref. No)	4.39	0.907–21.268	0.066
	Purpose of internet use: betting websites Yes (ref. No)	10.79	1.910-60.898	0.007
	Frequency of victimisation At least once a year (ref. Never) At least once every 6 months (ref. Never)	24.86 39.17	4.832-127.927 7.336-209.146	< 0.001 < 0.001

CI Confidence interval

^aThe latest model has been achieved in 7 steps. Nagelkerke = 0.345, Hosmer-Lemeshow p = 0.900

^bThe latest model has been achieved in 8 steps. Nagelkerke = 0.563, Hosmer-Lemeshow p = 0.998

In the perpetration model, higher odds of perpetration were independently associated with male sex (OR = 16.32), the use of betting websites (OR = 10.79), and the frequency of victimisation — being victimised at least once a year (OR = 24.86) and at least once every six months (OR = 39.17), each compared with never being victimised (all p < 0.05). Being a sixth-year student was associated with lower odds of perpetration than being a second-year student (OR = 0.13; p < 0.05).

Discussion

In this study, 28.2% of participants reported cyberbullying victimisation. This rate is lower than the 42.8% reported among adolescents in Saudi Arabia [15], but higher than the 23.0% reported in a study of adolescents aged 12–18 in the United States [16]. Compared with a systematic review reporting rates generally ranging from 6.5% to 35.4%, the rate in this study is close to the upper limit of this range [17], and across the literature victimisation rates vary between 4% and 72% among children and adolescents [18]. A cross-sectional study of 836 university students in Qatar reported a comparable cybervictimisation rate of 29.2% [19]. These comparisons should be interpreted with caution, because much of the variation between studies arises from differences in the definitions and cut-off criteria used to classify cyberbullying, in addition to differences in age group, sample characteristics, measurement tools, and cultural context. Moreover, most of these comparison figures come from adolescent samples, whereas directly comparable data from university populations remain limited. This demonstrates that cyberbullying is a multidimensional phenomenon across societies.

In this study, 9.7% of participants reported cyberbullying perpetration. A review of 27 articles showed that perpetration rates ranged from 3.0% to 44.1% (average 18%) [20]. This wide variation may stem from the lack of a common, established definition of cyberbullying, the heterogeneous nature of the study samples, and differences in the methods used.

In this study, although no significant difference was found between males and females in cyberbullying victimisation, perpetration was significantly higher among males than females. Another study similarly reports higher rates of perpetration among males [18], and a study of high-school students found that males perpetrate cyberbullying more frequently than females [21]. Alrajeh et al. similarly reported that male university students were significantly more likely to be classified as cyberbullies and cyberbully–victims than females [19]. The fact that males in Türkiye use the internet more than females (92.2% vs. 85.4%) [22] may be related to their greater engagement, particularly on gaming and betting platforms, and consequently their greater involvement in cyberbullying perpetration.

Cyberbullying victimisation increased significantly as internet use rose, particularly at six hours or more per day. This is consistent with evidence that increased internet use is one of the key risk factors associated with cyberbullying, and studies have shown that daily internet use exceeding five hours increases the tendency toward cyberbullying [23]. Furthermore, expanding children’s access to online opportunities increases their exposure to online risks alongside the benefits offered to them [15–24]. Greater internet use and more intensive online interaction may increase digital contact and, in turn, the likelihood of encountering cyberbullying.

In this study, participants who reported victimisation also had higher perpetration scores on the inventory, and vice versa, indicating substantial overlap between the two roles. Consistent with this, the logistic regression showed that higher odds of perpetration were associated with more frequent victimisation. The literature similarly reports a moderate positive correlation between victimisation and perpetration [25], and Dou et al., in a study of university students, reported a positive correlation between the two and suggested that victimisation can be considered a risk indicator [26]. Because the present study is cross-sectional, the direction of this association cannot be established: victimisation and perpetration may form a mutually reinforcing cycle, but this remains a hypothesis that the current design cannot confirm. Even so, the association suggests that intervention programmes may benefit from addressing victimisation and perpetration together.

This study found that meeting new people online was associated with higher odds of cyberbullying victimisation. This may expose individuals to higher-risk online environments and is largely consistent with the literature. Gámez-Guadix et al. found that meeting strangers online significantly increased the likelihood of perpetration at later stages [27], and another study reported that meeting or dating strangers online is highly correlated with cyberbullying [28]. Young people’s behaviour of meeting new people online may therefore be a useful risk indicator for cyberbullying victimisation or perpetration.

In this study, use of online betting websites was associated with higher odds of perpetration (OR = 10.79). Significant relationships between gambling/betting behaviour and cyberbullying have been reported; for example, a positive correlation between gambling and cyberbullying perpetration has been found in individuals with emotion-regulation difficulties and inadequate coping strategies [29]. Conversely, using the internet to read newspapers/news was associated with lower odds of victimisation. This suggests that the purpose of internet use, and not only its duration, may be related to the risk of cyberbullying. However, this association should be interpreted cautiously: rather than news reading being directly protective, it may reflect other characteristics of these users, and alternative explanations cannot be excluded with cross-sectional data.

In this study, 3.2% of participants who reported victimisation also reported suicidal thoughts. A study in rural South Africa reported that 23% of secondary-school students who were victimised had suicidal thoughts [30]. In a study using the Mental Health Inventory-5, 21.1% of participants reported considering self-harm because of cyberbullying [15], and some studies report that cyberbullying is more strongly associated with suicidal thoughts than traditional bullying [18]. This stronger association has been attributed to victims being targeted continuously online, the anonymity of perpetrators, and a lack of social support, although these mechanisms remain hypotheses. In our study suicidal ideation was assessed with a single self-report item, this finding should be interpreted with caution; nonetheless, it is consistent with evidence that cyberbullying victimisation, and the frequency of exposure, is associated with poorer mental health.

In this study, depression scores were significantly higher among participants who reported victimisation at least once in the past year. A study in India found that adolescents who are victimised are 2.07 times more likely to have depressive symptoms [31], and a study of high-school students reported that 21% experienced low self-esteem and 35% reported depression because of cyberbullying [30]. This is consistent with previous literature showing that cyberbullying victimisation is associated with poorer mental health. The direction of these associations cannot be established here: adolescents with depressive symptoms or poor mental health may also use the internet and social media more, which could increase their vulnerability to victimisation and, in turn, the risk of further depression. Participants who reported victimisation in the past year also had significantly higher negative self-concept scores. Wachs et al. reported that low self-esteem increases susceptibility to victimisation [32]; however, this relationship may not be one-directional, as victimisation can also lead to low self-esteem or a negative self-concept.

This study found that participants who reported perpetration had significantly higher hostility scores. When the reasons students engage in cyberbullying were examined in another study, peer pressure, boredom, and revenge were prominent [33]. These findings are consistent with research indicating that cyberbullying is often carried out for amusement or revenge and that hostile feelings predict perpetration [34–35]. They suggest that negative emotions may be more easily expressed online and that anonymity may increase the tendency towards aggressive behaviour.

Strengths and limitations

The relationships between variables were also assessed using advanced analyses (logistic regression) based on the results of univariate analyses. This has reduced the impact of confounding factors. These advanced analyses have enabled a more reliable interpretation of the study results and represent a strong aspect of the study. One of the most significant methodological strengths of this study is the use of various scales. Additionally, the simultaneous analysis of six different psychological states via the BSI to examine their specific relationships with cyberbullying is another key strength.

Participants were recruited from a single medical school using convenience sampling, and participation was voluntary; this may have introduced selection bias and limits the representativeness of the sample, so the results cannot be generalised to all medical or university students. The cross-sectional design means the findings reflect associations rather than causal relationships.

The information obtained from the students’ responses is based on self-reporting, so its absolute accuracy cannot be guaranteed.

A further limitation concerns the perpetration model. Only 25 participants (9.7%) reported perpetrating cyberbullying, and logistic regression models based on so few events can produce unstable parameter estimates, with inflated odds ratios and wide confidence intervals.

Conclusions

This study examined the prevalence of cyberbullying victimisation and perpetration among medical students, along with associated individual and behavioural factors. The findings indicate that cyberbullying remains a significant public health and mental health issue, with the nature of online behaviour patterns and their intended purposes playing an important role in this phenomenon.

The study found that 28.2% of participants reported cyberbullying victimisation, while 9.7% reported perpetration. The fact that victimisation and perpetration are meaningfully related to each other suggests that these two may be part of a mutually reinforcing cycle. Meeting new people online was associated with a higher risk of victimisation, while using betting websites was associated with a higher likelihood of perpetration; in contrast, using the internet to read news and newspapers was associated with lower victimisation. This highlights the need to focus on the nature and purpose of internet use rather than its duration. The significantly higher BSI scores among participants who were more frequently victimised are consistent with an association between cyberbullying and poorer mental health. Furthermore, the fact that some victims reported suicidal thoughts suggests that cyberbullying is not merely a form of online harassment but a serious psychosocial problem that may be associated with suicide risk. This emphasises the need to prioritise the mental health dimension in prevention and intervention programmes.

These findings have several practical implications, and prevention should not focus solely on internet access or usage time. Universities could embed digital-citizenship, digital-literacy, online-communication, emotion-regulation, and empathy components into student orientation and counselling services, paying particular attention to the higher-risk groups identified here, such as male students, heavy internet users, and users of gaming and betting platforms. Because victimisation and perpetration were closely associated, prevention and intervention programmes should address both roles together rather than separately, and students’ risky online behaviours (for example, meeting strangers online or using betting and gaming websites) should be identified early in collaboration with psychological counselling units. Routine, confidential screening within student health services could support the early detection of students experiencing victimisation or distress and allow timely referral to psychological support, particularly given the associations with depression, negative self-concept, and suicidal thoughts. At the policy level, awareness could be raised through seminars, helplines, and online-safety training, supported by guidance materials for parents and teachers, while limiting access to betting platforms and promoting safer online behaviour may further reduce risk. From a public health perspective, such interventions would contribute to a digital-safety culture at both the individual and societal levels.

In conclusion, cyberbullying is a multidimensional phenomenon associated with individual psychosocial characteristics, online behaviour patterns, and cultural factors. Therefore, prevention approaches should be sustained through comprehensive, psychosocial support and education-focused strategies.

Authors’ contributions

Cemal Koçak: conceptualization, formal analysis, investigation, methodology, visualization, writing-original draft, review & editing. Zeynep Irmak: conceptualization, methodology, visualization, writing-original draft, review & editing. Ayşe Gül Güven: supervision, writing-original draft, review & editing. Rola Tokan: conceptualization, investigation, methodology. Çile Büyüktaş: conceptualization, investigation, methodology. Zehra Aycan: conceptualization, investigation, supervision, review & editing.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Data availability

The datasets generated during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

All experiments were performed in accordance with the Declaration of Helsinki guidelines. The study protocol was approved by Ankara University Faculty of Medicine Ethics Evaluation Committee (date:03.04.2024, decision no: I03-246-24). Participation in the study was voluntary, and verbal and written consent was obtained from all participants.

Consent for publication

All authors have seen the manuscript and approved its submission to BMC Public Health.

Competing interests

The authors declare no competing interests.