Cyberbullying and Non-consensual Sexting

Cyberbullying behaviours are intentional and repeated hostile behaviours carried out via electronic media [6]. These can range from account hacking, impersonation, and sending viruses, to making and distributing offensive or threatening emails, texts, comments or images [52]. Cyberbullying is worryingly common and growing in prevalence, particularly amongst adolescents and young adults [48]. Estimates suggest lifetime prevalence for cyberbullying victimization of between 10 and 40%, and cyberbullying perpetration of between 3 and 24% [30, 72]. There is ample evidence that cyberbullying victimization is associated with a range of negative psychosocial outcomes [6, 54], such as impacting on mental health and even suicidal ideation. Moreover, there is evidence that these associations are to some extent independent from, and additional to, those of traditional forms of bullying [54].

Behaviours that have been closely linked to cyberbullying include forms of non-consensual sexting. Sexting refers to the sharing online of texts or images of a suggestive or sexually explicit nature [35]. Although sending and receiving ‘sexts’ between consenting adults can constitute a legitimate form of sexual expression [19], the use of deceptive or coercive methods to obtain sexts, and/or the sending or sharing of these sexts without consent, constitute particularly damaging forms of online aggression, also referred to as image-based sexual abuse [71]. Non-consensual sexting behaviours, particularly the use of sexts to threaten or manipulate others, typically involves a perpetrator who is known to the victim. When online methods such as non-consensual sexting are used to harass, control, threaten, or humiliate one’s current partner, it is classified as cyber dating abuse [73]. Sexting, non-consensual sexting, and the use of sexts in cyber dating abuse, are increasingly prevalent amongst young adults [35, 67]. Reported rates of receiving intimate images range from 41.5 to 86.4%, rates of sending images to others range from 38.3 to 65.2%, and rates of forwarding or disseminating images to others is approximately 20% [71]. Because the sexual material contained in sexts is often highly sensitive and personally damaging to the victim, non-consensual sexting behaviours can result in depression, anxiety and stress, and decreased self-esteem [34], along with humiliation, shame, panic, and reputational and professional damage [51, 22].

In the present paper we argue that cyberbullying and non-consensual sexting differ in several potentially important ways that warrant their inclusion and separate evaluation in research. Most obviously, non-consensual sexting has a narrower focus on sexual content, and the contexts within which it occurs are more likely to involve the victim’s close personal networks [35], their intimate personal relationships and sexual activities [53]. Sexting behaviours are also more normalised and the attitudes that underpin them, including those related to non-consensual sexting, are less uniformly negative, more subjective, and often more divergent between perpetrator and victim [e.g., [64]. Although the two behaviours are conveyed across the same online medium and have overlapping negative mental health effects, sexting is more likely to be associated with additional risks relating to sexual activities such as participation in unsafe and unprotected sex [66]. Despite this, sexting victimization generally has less severe mental health effects than cyberbullying victimization [e.g., 11]. These differences argue for the need to include measures of both cyberbullying and sexting, focus on non-consensual rather than consensual forms of sexting, and measure attitudes/beliefs from the perspective of both victim and perpetrator. There is also evidence that rates of perpetration and victimization differ substantially between genders.

Gender Differences in Online Behaviours and Underlying Attitudes and Beliefs

Cyberbullying and non-consensual sexting demonstrate significant gender differences, with ciswomen more likely to be victimized online and cismen more likely to be perpetrators. Ciswomen are more likely to receive unwelcome sexts [34], more likely to experience coercion to send sexts [e.g., 24], and more likely to have sexts of themselves used to cause reputational damage [62, 49]. To understand why, researchers have turned to gender differences in attitudes and beliefs concerning the acceptability and perceived consequences of inappropriate and harmful acts perpetrated online. This includes the view that cyberbullying is an acceptable form of revenge, that is relative harmless and potentially entertaining relative to offline forms of traditional bullying [8]. Similar positive beliefs about sexting and sext dissemination – that it is acceptable, relative innocuous, entertaining, and justifiable – also account for sexting perpetration behaviours [28, 15]. As with the behaviours themselves, positive and normalising attitudes and beliefs about cyberbullying and sext dissemination are more strongly held by cismen than ciswomen [9, 7].

In terms of explaining the motivations for acting on these attitudes and beliefs, researchers have turned to gender roles and power imbalances that underpin offline interpersonal physical violence between cismen and ciswomen [6, 39]. Research into offline violence suggests that while cismen and ciswomen are equally complicit in what might be referred to as situational couple violence” the sort of violence that does not typically lead to criminal offending or reporting to authorities, cismen are far more likely to use violence in the form of patriarchal oppression and coercive control over ciswomen [32, 38]. Indeed, non-consensual sexting within relationships is also associated with offline dating violence, interpersonal conflict, and abuse by sexting perpetrators [30,31,32]. Extending this interpretation to cyberbullying perpetration and sexting, one might expect to find that cismen are more likely to engage in higher severity oppressive and controlling behaviours that victimise women, behaviours such as pressuring or coercing a woman to provide a sexually explicit sext and/or to distribute such a sext without her consent.

Trans and Gender Diverse Perspectives

Transgender and gender diverse (TGD) people are often drawn to online environments in the belief that they provide safe spaces (relative to offline environments) within which to express their identity authentically, obtain and provide social support, and receive information and advice in relation to their gender identity and expression [33,34,35]. However, there is growing recognition that although the online environment can be less physically threatening to TGD people, it can still be a psychologically hostile environment [36,37,38].

Cyberbullying is commonly experienced by TGD adolescents in the form of verbal harassment, abuse, threats of violence [1], whereas non-consensual sexting is experienced in the form of unwanted sexts, being pressured to provide others with sexts, or the use of sexts to cause reputational damage for adolescents and young adults [69, 37]. The victimization experiences of TGD people appear to be expressions of the same gender-based oppression they experience offline. This includes discrimination, harassment, vilification, actual or threatened violence, negative stereotyping, exclusion/rejection, invalidation or non-confirmation of gender identity [42,43,44,45,46,47,48,49,50,51,52,53,54,55].

Although quantitative research in the area is limited, particularly for adults [69, 4], existing studies show that TGD people are more likely than both ciswomen/cisgirls and cismen/cisboys to experience victimization in the context of online dating [37], and that these experiences are more likely to contribute to negative mood, low self-esteem, and suicidal thoughts and behaviours in TGD people [1, 58, 56]. However, the focus of previous research has been on victimization experiences, and comparatively little is known about online perpetration by TGD adults and the attitudes and beliefs that motivate these behaviours. However, there is some speculation that the factors that motivate online engagement and perpetration by TGD people may differ from those of cisgender perpetrators [5].

Aims and Hypotheses

The above literature review identified the problematic nature of cyberbullying and non-consensual sexting, including their increasing prevalence across ciswomen and cismen. Gender differences in these behaviours were highlighted along with attitudes and beliefs thought to explain this. The review also identified several limitations or omissions in the literature, most notably the limited research into the behaviours, experiences, and attitudes/beliefs of gender minorities who, when included, are often grouped together with cisgender individuals with diverse sexual orientations, In addition, we note that previous research in the area has tended to utilize qualitative rather than quantitative research methods [cf. 49], target adolescent and young adult populations rather than adults [cf. 47], treat cyberbullying and non-consensual sexting behaviours as though they are equivalent in their effects, and focus on the attitudes and beliefs of perpetrators rather than victims. To address these potential omissions and limitations, we administered a survey to a sample of ciswomen, cismen, and TGD adults containing measures of cyberbullying victimization and perpetration, sexting victimization and perpetration, as well as measures of positive attitudes and beliefs in relation to both cyberbullying and sexting. The following hypotheses were tested:

H1: Cismen will report more cyber-perpetration, non-consensual sexting perpetration, and more positive attitudes and beliefs about cyberbullying and sexting, than ciswomen and TGD adults.

H2: TGD adults will report more cyber-victimization and non-consensual sexting victimization than ciswomen and cismen.

H3: Positive attitudes and beliefs will be associated with perpetration behaviours but not victimization experiences in all three gender groups.

Method

Participants

Participants were 1506 adults ranging from 18 to 66 years old (M = 23.27, SD = 3.66). A total of 146 (9.7%) identified as TGD (34 transwomen, 28 transmen, 69 nonbinary), 638 (42.4%) identified as ciswomen, and 722 (47.9%) identified as cismen. Their ethnicity was predominantly White (N = 803, 53.9%), Asian (N = 422, 28.3%), Hispanic/Latinx (N = 106, 7.1%), Black (N = 88, 5.9%), Jewish (N = 21, 1.4%), Middle Eastern/Arab (N = 13, 0.9%), Pacific Islander (N = 16, 1.1%), mixed race or multi-ethnic (N = 13, 0.9%), and Indigenous/First Nations (N = 9, 0.6%). Their countries of residence were mainly Australia (N = 414, 27.6%), USA (N = 607, 40.5%), India (N = 206, 13.8%), UK (N = 31, 2.1%), Singapore (N = 30, 2%), Canada (N = 19, 1.3%), Brazil (N = 11, 0.7%), Germany (N = 6, 0.4%), and Italy (N = 5, 0.3%). Cisgender participants primarily self-described as opposite-gender attracted (N = 1079, 79.5%), followed by bisexual, pansexual or with no gender preference (N = 206, 15.2%), same-gender attracted (N = 47, 3.5%), and asexual (N = 25, 1.8%). TGD participants primarily self-described as bisexual, pansexual or with no gender preference (N = 85, 58.2%), followed by opposite-gender attracted (N = 31, 21.2%), same-gender attracted (N = 31, 21.2%), and asexual (N = 19, 13.0%).

Materials

Participants were directed to an online survey created using Qualtrics™ containing demographic information and the following scales presented in the order listed below:

Cyberbullying Perpetration and Victimization were assessed using three items regarding cyberbullying perpetration (using digital technologies to disparage others, spreading rumours, and harassing/making threats) [8], and three items regarding cyberbullying victimization (being disparaged, being the topic of rumours, and being the target of attacks online) [75]. Participants responded never (1) to everyday/almost every day in the past year (6) in response to each statement, with responses averaged across the three statements.

Non-consensual Sexting Perpetration and Victimization were assessed using a list of behaviours developed by Clancy et al. [15] to which participants responded No (0)/Yes (1). Five of the behaviours described common examples of sexting victimization: (1) Receiving an image-based sext without consent; (2) receiving an image-based sext that was unwanted/unwelcome; (3) agreeing to send an image-based sext of oneself that one did not want to send; (4) experiencing negative consequences from sending a sext (e.g., bullying, relationship impact, blackmail); and (5) having a sext of oneself forwarded to others without one’s consent. Five of the behaviours described common examples of actual or attempted non-consensual sexting perpetration: (6) Asking someone to send an image-based sext of themselves?; (7) hassling someone to send you an image-based sext of themselves; (8) Sending someone an image-based sext of oneself with negative intent (e.g., bullying, attention-seeking, seeking to cause trouble for them, revenge); (9) Forwarding a received sext to others without permission and with negative intent); and (10) Visiting websites on social media in search of nude images or videos posted without the knowledge of those in the images/videos (sometimes referred to as “Slut Pages / Exposed Pages / Girls of…”). Responses to victimization and perpetration items were tallied separately.

The Positive Attitudes to Cyberbullying Questionnaire [PACQ; 22] was used by participants to indicate their pro-cyberbullying attitudes. Participants used 5-point Likert scales ranging from strongly disagree (1) to strongly agree (5) to indicate their level of agreement with nine statements such as “Teasing others online, via emails, or text messages is fun” and “It is acceptable to send mean emails to others when they deserve it”. Items were averaged to obtain an overall pro-cyberbullying score (although the authors of the scale suggesting summing scores, we chose averaging in keeping with the non-consensual sexting attitudes/beliefs measure described next). Several items were adapted from the original to refer to contemporary social media sites: Instagram, Snapchat, WhatsApp and Facebook [9].

Sexting Dissemination Attitudes and Subjective Norms were assessed using responses seven statements from a cyberbullying measure [28] adapted to refer explicitly to attitudes specifically about sexting dissemination using statements such as “Forwarding or sharing sexually explicit images of others via text or mobile app can be funny” and “Sharing sexually explicit images via text or mobile app of others can enhance social status” [see 24]. Note that these statements did not explicitly refer to non-consensual forms of sexting, only pro-sexting attitudes and beliefs about sexting in general. Response options ranged from strongly disagree (1) to strongly agree (5), with responses averaged across the seven statements to yield a single non-consensual sexting attitude score.

Procedure

We complied with APA ethical standards in the treatment of participants, and our study was approved by the Human Research Ethics Committee of our institution. Most cisgender participants (74.0%) were recruited either through Amazon MTurk (and received a payment of 1USD) or through Prolific (and received a payment of 1GBP). The remainder were recruited through social media sites (Facebook, Instagram and Reddit) and via snowball sampling. Approximately half of TDG participants (50.7%) were recruited via MTurk or Prolific, with the remainder recruited through social media as well as via online TGD support groups and organisations, online forums, trans blogs, and using physical flyers posted in local gender support and health clinics.

Results

Preliminary data screening, variable creation, and descriptive statistics were conducted using IBM SPSS Statistics for Windows, Version 27 (IBM Corp, Armonk, NY, USA). Item responses were screened for random or non-varying responding, incomplete responding, and evidence of bots in the form of repeatedly used IP addresses. Missing items were evident in fewer than 2% of responses and distributed randomly across measures and cases. These were excluded from variable creation. Variables were computed in accordance with instructions detailed in the Materials subsection. Univariate outliers for these variables were defined as values >  ± 3.29 standard deviations from the mean [65]. No cases were identified as multivariate outliers (Mahalanobis’s distance p < 0.001). Summary descriptive statistics and bivariate correlations for all variables used in inferential analyses are provided in Table 1. Cyber-victimization and cyber-perpetration behaviours were found to be positively skewed. Therefore, in inferential analyses throughout the paper, reciprocal-transformed versions of these variables were used (with the resultant transformed variables inverted so as to preserve their original valence). Note also that internal consistency of the Sexting Dissemination Attitudes and Subjective Norms measure was low, particularly for TGD participants (α = 0.58, Ω = 0.63), even with the removal of two items for all participants due to item-total correlations < 0.2 (“Sexually explicit images via text or mobile app usually end up being seen by more than just those to whom they were sent” and “Women have to worry more than men about sexually explicit images of themselves being viewed or distributed via text or mobile app to someone other than they were intended for”).

Table 1 Bivariate correlations and descriptives (ciswomen N = 638; cismen N = 722; trans and gender diverse N = 146)
Full size table

Associations Between Behaviours, Roles, and Beliefs/Attitudes

To evaluate hypotheses about associations between online behaviours (cyberbullying v non−consensual sexting), role (perpetrator v victim), and positive beliefs/attitudes (about cyber−perpetration v sexting), Pearson bivariate correlations were conducted separately for each gender group. These are included in Table 1, and show evidence of (i) associations between online behaviours in the form of strong positive correlations between cyber−victimization and perpetration, and moderate positive correlations between sexting victimization and perpetration, (ii) associations between online behaviours, specifically modest positive correlations between cyber and non−consensual sexting perpetration, and moderate positive correlations between cyber and sexting victimization, and (iii) associations between attitudes/beliefs in the form of moderate positive correlations between cyberbullying and sexting attitudes. These correlations were in the same direction and of similar magnitude for ciswomen, cismen, and TGD adults.

Gender Differences in Cyberbullying and Non-consensual Sexting Behaviours and Experiences

To test hypotheses concerning gender differences in behaviours and experiences, two MANCOVAs were conducted, one on cyber-perpetration and victimization, and the other on non-consensual sexting perpetration and victimization, by gender group and sexual orientation with age included as a covariate. Sexual orientation was included in recognition of the fact that the target of non-consensual sexting perpetration is typically a person of the gender to whom the perpetrator is sexually attracted. Age was included as a covariate in recognition of generational differences in usage of and attitudes to online social interactions, and also for the pragmatic reason that the older a person the more opportunities they will have had to interact online. Cell means and standard errors are plotted in Fig. 1.

Fig. 1
figure 1

Mean scores for cyberbullying and sexting victimization experiences and perpetration behaviours, and positive attitudes and beliefs about these behaviours. Notes: Ciswomen N = 638; cismen N = 722; trans and gender diverse N = 146; ± 1 standard error bars are included

Full size image

In terms of cyberbullying behaviours and experiences, a significant but small multivariate main effect of cyberbullying was obtained for gender F(4,2940) = 9.29, p < 0.001, η2partial = 0.01, but not sexual orientation F(6,2940) = 1.47, p = 0.185 ns, η2partial = 0.00, and a significant but small multivariate interaction between gender and sexual orientation was also obtained, F(12,2940) = 2.90, p < 0.001, η2partial = 0.01. Significant between-subjects main effects of gender were obtained for both cyber-victimization and perpetration, F(2,1480) = 4.17 and 7.26, p < 0.05 and 0.001, η2partial = 0.01 and 0.01, respectively. Pairwise orthogonal comparisons conducted against a modified Bonferroni correction [α /n = 0.05/3 = 0.017, 52] revealed that TGD adults (M = 1.46, SD = 0.25) reported experiencing higher levels of cyber-victimization than cismen (M = 1.37, SD = 0.27), and that both TGD adults (M = 1.20, SD = 0.24) and ciswomen (M = 1.19, SD = 0.26) reported fewer acts of cyber-perpetration than cismen (M = 1.30, SD = 0.29). A significant between-subjects interaction between gender and sexual orientation was also obtained for victimization but not perpetration, F(6,1470) = 3.32, p < 0.001, η2partial = 0.02, with inspection of Fig. 1 suggesting that this was due primarily to opposite-gender attracted TGD adults experiencing the most victimization (M = 1.65, SD = 0.12).

Effects were typically more pronounced in terms of non-consensual sexting behaviours and experiences, with significant multivariate main effects of gender F(4,2958) = 20.67, p < 0.001, η2partial = 0.03, and sexual orientation, F(6,2958) = 4.52, p < 0.001, η2partial = 0.01, as well as a significant interaction of gender by sexual orientation, F(12,2958) = 3.09, p < 0.001, η2partial = 0.01. Significant between-subjects main effects of gender were obtained for non-consensual sexting perpetration, F(2,1479) = 29.49, p < 0.001, η2partial = 0.04, but not victimization, with pairwise comparisons revealing that cismen (M = 1.73, SD = 1.42) committed more non-consensual sexting perpetration than ciswomen (M = 1.43, SD = 1.18), with TGD adults committing the least perpetration (M = 0.61, SD = 0.72). Between-subjects main effects of sexual orientation were obtained for both victimization and perpetration F(2/3,1479) = 4.20 and 8.15, p < 0.01 and 0.001, η2partial = 0.01 and 0.02, respectively, with pairwise comparisons revealing that asexual adults experienced less sexting victimization (M = 0.59, SD = 0.79) and engaged in less non-consensual sexting perpetration (M = 0.57, SD = 0.97) than opposite-gender attracted (M = 0.76 and 1.52, SD = 0.89 and 1.29), same-gender attracted (M = 1.06 and 1.42, SD = 1.00 and 1.23), or bi/pansexual (M = 1.11 and 1.55, SD = 1.14 and 1.38) adults, respectively. A significant interaction effect of gender by sexual orientation in terms of sexting victimization, F(6,1479) = 5.30, p < 0.001, η2partial = 0.02, but not perpetration, was also obtained, suggesting that while opposite-gender attracted TGD adults experienced the most victimization, opposite-gender attracted cismen experienced the least victimization. The victimization experiences of cismen appeared to be particularly dependent on sexuality, with opposite-gender attracted and asexual men reporting the least sexting victimization, and same-gender attracted and bi/pansexual cismen reporting the most sexting victimization.

Gender differences in cyberbullying and sexting attitudes and beliefs

To test hypotheses concerning gender differences in relation to attitudes and beliefs, a MANCOVA was conducted on cyberbullying and sexting attitudes/beliefs by gender group and sexual orientation with age as a covariate. Significant multivariate main effects of gender, F(4,2950) = 11.37, p < 0.001, η2partial = 0.02, and sexual orientation, F(6,2950) = 2.91, p < 0.01, η2partial = 0.01, were obtained, but there was no interaction between them. Significant between-subjects main effects of gender were obtained for cyberbullying and sexting attitudes, F(2,1475) = 9.95, 21.83, p < 0.001 and 0.001, η2partial = 0.01 and 0.03, respectively. Planned orthogonal comparisons against an adjusted alpha of 0.017 revealed that cismen (M = 2.52 and 2.15, SD = 0.81 and 0.80) held significantly more positive views about cyberbullying and sexting than ciswomen (M = 2.19 and 1.76, SD = 0.92 and 0.74) or TGD adults (M = 1.99 and 1.60, SD = 0.65 and 0.49). A significant between-subjects main effect of sexual orientation was obtained for sexting but not cyberbullying attitudes, F(3,1475) = 5.00, p < 0.01 and 0.001, η2partial = 0.01, with planned comparisons revealing that bi/pansexual adults (M = 1.98, SD = 0.81) held the most positive attitudes and beliefs about sexting. No interactions between gender and sexual orientation were observed.

Note that because of differences in the origin of cisgender and TGD participants (with a greater proportion of cisgender participants recruited via MTurk or Prolific; see Procedure) all analyses of variance described in the Results were repeated with ‘recruitment source’ (dummy coded as: 0 = other; 1 = MTurk or Prolific) included as a covariate. No changes to the statistical significance of main effects or interactions were observed.

Gender differences in the influence of attitudes and beliefs

The influence of attitudes and beliefs was evaluated using two versions of the path model shown in Fig. 2. The model features cyberbullying behaviours (Model I) and non-consensual sexting behaviours (Model II) regressed separately on cyberbullying and sexting attitudes and beliefs, with age included as a predictor of attitudes/beliefs. Data from the three gender groups were included simultaneously with gender group treated as an unconstrained parameter. Note that the models include correlations between the behaviours and between the measures of attitude/belief, and all are statistically significant. The models exclude non-significant paths from age to attitudes/beliefs about cyberbullying. The figure caption lists fit indices, confirming acceptable fit of these models against the following fit criteria: χ2(df) p > 0.05; χ2/df < 5; root mean square error of approximation (RMSEA < 0.08); standardised root mean square residual (SRMR < 0.08), comparative fit index (CFI > 0.95) and Tucker-Lewis Index (TLI > 0.95) [12, 27].

Fig. 2
figure 2

Baseline models used in multigroup path analyses involving (I) cyberbullying behaviours and (II) sexting behaviours (see Models II and II in Table 2). Standardized statistics are shown, and only significant paths (in relation to the gender-combined data) are shown. Model fit: χ2(3) = 19.35 and 3.61, p = .02 and .001; χ2/df = 2.15 and 3.61; RMSEA = .028 and .042; SRMR = .032 and .039, CFI = .99 and .97, and TLI = .98 and .91, for Model I and II respectively. * p < .05 (ns = ’non-significant’); C = correlation; ^Coefficients determined by multigroup path analysis to be significantly different across the gender groups (at p < .05/6 = .008)

Full size image

Model I (the cyberbullying model) confirms that positive attitudes to both cyberbullying and sexting are associated with cyber-perpetration (paths β1 and β3) and, to a lesser extent, cyber-victimization (β2 and β4), whereas Model II (the non-consensual sexting model) reveals weaker paths but with greater specificity in the sense that non-consensual sexting perpetration is associated primarily with sexting attitudes (β3) and not cyberbullying attitudes (β1). Both models show that positive attitudes to cyberbullying and sexting are moderately correlated (C1), and that perpetration and victimization behaviours are moderately to strongly correlated in the context of both cyberbullying and non-consensual sexting (C2).

Inspection of coefficients in Fig. 2 also reveals substantial gender differences in these associations. To evaluate these gender differences, the models with all parameters unconstrained were used as baseline models against which to conduct multigroup path analyses by gender group. These analyses were used to test for significant differences in path coefficients, separately for cyberbullying and non-consensual sexting behaviours. Multigroup analyses were conducted after the method of Byrne [12] using the AMOS Manage Models and Model Comparison functions. Each coefficient was individually constrained across gender groups and the resultant reduction in model fit relative to the baseline model used to determine if significant differences in that coefficient exist between the gender groups. Because these comparisons inflated familywise error, a Bonferroni correction was made to alpha levels of α/n = 0.05/6 = 0.008.

The results, summarized in Table 2 and in the table insert of Fig. 2, show that significant deterioration of fit in both models occurred when C1 was constrained between gender groups. Inspection of standardized coefficients in Fig. 2 reveals that the association between positive attitudes to cyberbullying and non-consensual sexting was slightly lower for TGD adults compared to cisgender adults. The cyberbullying model’s (Model I) fit also deteriorated significantly when β1, β2, or β3 was constrained. As shown in Fig. 2, while positive attitudes to cyberbullying were associated with cyber-perpetration (β1), the relationship was stronger for TGD adults compared to cisgender adults; positive attitudes to cyberbullying were also associated with cyber-victimization, but only in cismen and TGD adults, not ciswomen (β2); and positive attitudes to sexting were also associated with cyber-perpetration, but only for cisgender adults, not for TGD adults (β3). The non-consensual sexting model’s (Model II) fit was unaffected by constraints placed on these paths but did deteriorate significantly when C2 was constrained. Figure 2 indicates that non-consensual sexting perpetration and victimization are more closely related in cisgender adults than they are in TGD adults.

Table 2 Results of multigroup path analyses (ciswomen N = 638; cismen N = 722; trans and gender diverse N = 146)
Full size table

Discussion

This study aimed to evaluate and compare the attitudes and beliefs underpinning cyberbullying and non-consensual sexting perpetration and victimization in cisgender and TGD adults. Consistent with our first hypothesis, cismen reported higher rates of both cyberbullying and non-consensual sexting perpetration than ciswomen irrespective of sexual orientation (asexual adults were lowest in both perpetration and victimization, however, this was similar across gender groups). This is consistent with previous evidence that cyberbullying [6, 39] and non-consensual sexting [34, 19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49] exhibit substantial gender differences, independently of sexual orientation. Cismen also expressed substantially more positive attitudes and beliefs regarding both cyberbullying and sexting. This points to gender differences in beliefs/attitudes as the reason for the higher perpetration rates exhibited by cismen [62, 74].

In our second hypothesis, we predicted that ciswomen would report more victimization experiences than cismen [34]. However, we found no difference between cisgender adults in terms of cyberbullying or sexting victimization. In this respect, our results reflect the often-conflicting findings reported previously in the literature, with some studies finding women to be more likely than men to be the victims of serious and compromising online images and texts while others find no differences by gender [62, 49]. Examination of the sexting victimization results by sexual orientation suggests a possible explanation for these mixed results, one that reinforces the importance of gender in victimization experiences not only in terms of the gender of the victim but also the gender of the perpetrator. As shown in Fig. 2, adults who were most likely to be exposed to men in the context of intimate relationships – that is, opposite-gender attracted ciswomen and same-gender attracted or bi/pansexual cismen – were more likely to experience sexting victimization than adults who were least likely to be exposed to men – that is, same-gender attracted ciswomen, opposite-gender attracted cismen, and asexual adults.

Given the exclusion of TGD adults from most previous research in the area, and the limited focus on them as victims rather than perpetrators [43], it was difficult to make predictions about the relative standing of TGD adults compared to cisgender adults in relation to perpetration behaviours. Our results clearly show that TGD adults were similar to ciswomen in terms of reporting lower levels of cyberbullying-perpetration than cismen, and were even lower even than ciswomen in relation to non-consensual sexting perpetration. These low levels of perpetration behaviour closely mirrored their underlying attitudes and beliefs, with TGD adults holding fewer positive beliefs about these behaviours than cismen, and equal to ciswomen.

We propose two possible explanations for TGD adults’ low online perpetration rates and relatively prosocial attitudes/beliefs. First, TGD adults are more reliant than cisgender adults on online environments to express their gender identity, access social supports, and obtain health information safely [13, 16, 23]. Perhaps the greater perceived value of online environments makes TGD people less inclined to be reckless, disruptive, and hostile in these environments. Second, there is ample research demonstrating that TGD people are more likely to be victims of online hostility than cisgender people [56, 14], particularly in relation to non-consensual sexting [69, 37]. This was confirmed in the present study, with TGD adults reporting the highest levels of cyber-victimization and sexting victimization (see below). Interestingly, similar anti-trans hostility experienced offline have been shown to stimulate positive change in TGD adults in the form of improved self-awareness, insights, personal growth, empathy for others [45, 57], as well as greater understanding and compassion for other disadvantaged people [61]. This ‘growth from adversity’ may translate into greater sensitivity and less hostility towards others online, including other minority people or vulnerable groups.

As partially anticipated by our third hypothesis, and as noted above, cyberbullying-perpetration behaviours were associated with positive beliefs and attitudes about cyberbullying, and (albeit to a lesser extent) non-consensual sexting behaviours were associated with positive beliefs and attitudes about sexting. This is consistent with the results of previous research that have been interpreted as suggesting that problematic online behaviours are motivated and enabled by the belief that inappropriate online behaviours are relatively innocuous, justifiable, and/or entertaining [22,23,24,25,26]. However, substantial gender-group differences in associations between beliefs and behaviours were observed. Most notably, cyberbullying attitudes/beliefs had stronger associations with cyber-perpetration in TGD adults than in cisgender adults, whereas sexting attitudes/beliefs had far weaker associations with cyber-perpetration for TGD adults than for cisgender adults. In fact, for cisgender people, sexting attitudes/beliefs were more closely related to cyber-perpetration than were cyberbullying attitudes/beliefs.

These asymmetric relationships suggest that, unlike cisgender adults, different motivations may explain the online behaviours of TGD adults in generic versus sexual/dating contexts. This interpretation is consistent with our multigroup path analyses, in the sense that associations between attitudes to cyberbullying and attitudes to sexting were significantly greater in cisgender adults than in TGD adults, and associations between perpetration and victimization, both in the context of cyberbullying and non-consensual sexting, were substantially higher in cisgender adults compared to TGD adults. Our results also suggest that the idea that non-consensual sexting perpetration is a merely a manifestation of cyberbullying with sexual content [e.g., 9, 23], is less true for TGD people than it is for cisgender people.

The greater distinction between the sexual versus generic online contexts in TGD people may reflect the special and growing importance of online environments for intimacy in gender and sexual minorities [44, 42]. Online environments allow those seeking romantic and sexual partners to maintain their anonymity and privacy [3], manage their public persona and control how much information about themselves they divulge to others, [60, 46]. Individuals can proactively declare their identity thereby signalling their availability to compatible others [21], and take advantage of geolocation features and dedicated online sites to find compatible others even when they are distributed over a large area [68,69,70]. Use of these strategies online can help to reduce the risk of rejection and hostility [31], help to reduce fear [42], and improve confidence and comfort [43] with seeking relationships online. Unfortunately, there is very little empirical research on online relationships TGD adolescents or youths, and even less research into the attitudes and beliefs that influence relationship behaviours across different online contexts [13, 69, 43].

Limitations

The study’s internal validity was limited by our reliance on self-report measures and use of a cross-sectional design with correlational analyses. External validity was limited by volunteers being recruited via online sites such as MTurk and Prolific, a sample that was not particularly diverse in terms of ethnicity or geography (most were ethnically white and from Western developed countries), and a limited sample size for TGD people which prevented separate analysis of responses from transwomen, transmen, and agender people. It is also worth noting the low internal consistency (α = 0.58; Ω = 0.63) of responses to the Sext Dissemination Attitudes and Subjective Norms items provided particularly by TGD participants (which necessitated the removal of two items from the measure for all participants). Finally, the correlations we obtained between attitudes/beliefs and perpetration behaviours were not particularly strong and point to the additional involvement of factors perhaps unrelated to the online environment. Previous research has found that intrinsic factors, including psychopathic traits such as callousness and unemotionality; personality traits of low agreeableness and high extraversion; internalisation and externalisation difficulties; self-esteem; attachment style; and a tendency toward moral disengagement and disinhibition, are important predictors of antisocial behaviours generally and, more specifically, cyber-perpetration [51]. We suggest that future research into online behaviours with TGD adults consider including an evaluation of these intrinsic factors.

Conclusions

Our survey of problematic online behaviours was notable for its inclusion of two forms of online behaviour (cyberbullying and non-consensual sexting), an evaluation of the attitudes and beliefs that underpin cyberbullying and sexting, and the inclusion of gender-nonconforming participants. We found that cismen report the highest cyber and non-consensual sexting perpetration rates, and this was associated with their holding the most antisocial attitudes and beliefs about these behaviours. We also found that TGD people report the lowest cyberbullying and non-consensual sexting perpetration rates and hold more prosocial attitudes/beliefs in relation to these behaviours. We also observed that the attitudes/beliefs associated with TGD adults’ behaviours in generic online interactions are different from those that are associated with their interactions in the context of sex/dating, whereas for cisgender adults this distinction is less apparent. In terms of victimization, we replicated previous findings that TGD adults experience more cyber-victimization and sexting victimization than their cisgender peers. Together, these results highlight the need for a more gender-inclusive approach to understanding the motivations, behaviours, and experiences of adults in online environments.