Sex-gender diversity and cybervictimisation

Discrimination based on gender or sex diversity is a “structural” problem which makes people who diverge from heteronormative standards more vulnerable to bullying (Angoff & Barnhart, 2021; Jonas et al., 2022; UNESCO, 2017). This discrimination also occurs in online contexts (Abreu & Kenny, 2018), leading to invalidation of identity, social exclusion, and even the criminalisation of people from sexual minorities (Döring et al., 2022). This is the case of minors and young people who are lesbian, gay, bisexual, trans, questioning or unsure of their sexual orientation, consider themselves ‘queer’ or do not conform to the binary patterns of sex and/or gender, and any other biological condition, orientation, identity or gender expression, for example, intersexuality (hereinafter, LGBTQ+; Espelage et al., 2019).

The results suggest that LGBTQ+ people are more frequently victims and suffer more than their heterosexual cisgender counterparts, both worldwide (Jonas et al., 2022) and on a national level (Llorent et al., 2016). In fact, previous studies have shown that the prevalence of victimisation depends not only on sexual orientation, but also on other dimensions such as gender identity and gender expression (Kosciw et al., 2018) and that adolescents and young people perceive transgender people and those recognised as gender-incongruent as the most frequently victimised (Gower et al., 2018).

However, despite the fact that cyberbullying has been shown to have more serious consequences than traditional bullying (Kwan et al., 2020), its study in relation to the LGBTQ+ community is still in its infancy. Due to inconsistencies in the use of terminology, the systematic reviews produced to date have found that the prevalence of cybervictimisation among LGBTQ+ youth varies enormously between 10.5% and 71.3% (Abreu & Kenny, 2018). In fact, these figures could be even higher, since the studies tend to analyse either homophobic aggression or general aggression exclusively, and not both types together (Rodríguez-Hidalgo & Hurtado-Mellado, 2019).

In addition, with respect to sex-gender diversity, they usually only deal with sexual orientation (Elipe et al., 2022; Garaigordobil & Larrain, 2020). It is therefore crucial to study cyberbullying in LGBTQ+ by analysing young people’s perception of how these acts of aggression take place and the different dimensions of sexuality (sex, sexual orientation, gender identity and expression), as well as the underlying beliefs.

The role of internalised homonegativity in LGBTQ+ cybervictimisation

LGBTQ+ cyberbullying is a type of bullying based on the stigma attached to beliefs and social ideas that devalue sexual and gender minorities (Earnshaw et al., 2018). Young people, whether LGBTQ+ or not, grow up in environments mainly dominated by the heteronorm or heterosexism, imbued with homonegativity (Russell & Bohan, 2006). Here, the social messages are that heterosexuality is the single most desirable form of sexuality and that it is acceptable to discriminate against any other form of sexual diversity. These environments lead to internalised homonegativity, in other words, the inevitable internalisation of these messages due to the reiteration of generalised and largely unreflective cultural beliefs (Herek, 2007; Russell & Bohan, 2006), and result in feelings of shame and discomfort about belonging to sexual minorities (Puckett et al., 2016).

Internalised homonegativity has been shown to be a risk factor for involvement in face-to-face violence (Berg et al., 2016) and justification of violence (Quirk et al., 2018), as well as having a detrimental impact on mental health (Puckett et al., 2016). However, it is vital to analyse the relationship between this type of prejudice and our awareness about, and identification of virtual phenomena such as cyberbullying and, in particular, cybervictimisation, whether in general, or directed specifically against LGBTQ+.

In recent years, progress has been made in taking action to address cyberbullying, such as the introduction of evidence-based psychoeducational programs (Del-Rey et al., 2018), or the creation of LGBTI government regulations and innovative experiences (Pichardo & Puche-Cabezas, 2019). Despite this, evidence-based psychoeducational programs that focus on this specific type of cyberbullying are still scarce, and it is therefore essential to study the role of internalised homonegativity in this phenomenon, in order to lay the foundations for future educational progress in this field.

The present study

Studying LGBTQ+ cyberbullying requires a comprehensive analysis of cybervictimisation based on sexual-gender diversity as a whole, to obtain an integrated vision that will enable us to develop psychoeducational strategies to prevent and intervene in this form of cyberaggression (Abreu & Kenny, 2018). For this reason, in this research, we used a mixed sequential qualitative and quantitative methodology (Sampieri, 2018) in order to explore perceptions of cybervictimisation in a community sample of adolescents comprising both LGBTQ+ and cisgender heteronormative (hereinafter, CH).

The specific objectives of our exploratory research were, in the qualitative stage, (1) to find out the pupils’ perceptions about LGBTQ+ cyberbullying and the type of aggression it involves, which, in turn, provides a basis for the quantitative study. In the quantitative stage, the research objectives were (2) to analyse general cybervictimisation and collective LGBTQ+ cybervictimisation based on affective-sexual, bodily and gender diversity (from now on, ASBGD) both as a whole and in its different dimensions: sexual orientation, identity and gender expression; and (3) to examine the possible moderating role of internalised homonegativity between general cybervictimisation and collective LGBTQ+ cybervictimisation (feeling hurt by cyberbullying directed at LGBTQ+ people), taking into account ASBGD.

Participants

In the qualitative study, 175 adolescents took part, with an age range between 12 and 18 (60% girls, 40% boys; M_age=14.12, SD_age=1.88), from eight secondary state schools in Andalusia (Spain). In total, 27 discussion groups took place: 12 focus groups in 1st year of secondary school (aged 12-13), two in 2nd year (aged 13-14), three in 3rd year (aged 13-14), five in 4th year (aged 15-16); three in 1st year of High School/Baccalaureate (age 16-17) and two in 2nd year of High School/Baccalaureate (age 17-18).

The quantitative study included 1,971 adolescents between the ages of 12 and 18 (M_age=15.01, SD_age=1.70) from 12 secondary state schools in Andalusia (Spain). 14.1% were in 1st year of secondary, 18% in 2nd year, 19.2% in 3rd year, 17.3% in 4th year and 31.5% in High School/Baccalaureate or Professional Training. Table 1 shows the details of the sexual diversity of the participants.

Instruments

Semi-structured script. A discussion guide was used in the focus groups to explore perceptions about LGBTQ+ cyberbullying and types of aggression (see in: https://doi.org/10.6084/m9.figshare.21724160.v4). The guide consisted of 20 open-ended questions divided into four dimensions: sex-gender diversity in schools, LGBTQ+-phobic aggression and bullying, associated factors, and the impact of aggression, and contained questions such as: “Why do you think these situations occur? Why do some people attack others because of their diversity?”.

Cybervictimisation. To assess cybervictimisation suffered over the past two months, we used the cybervictimisation subscale from the European Cyberbullying Intervention Project Questionnaire (ECIP-Q; Del-Rey et al., 2015), which contains 11 Likert-type items (α_{cybervictimisation}=.81) with five response options, ranging from "Never" to "Yes, more than once a week". The statements referred to the frequency of being the victim of online aggression, such as insults, threats, or spreading rumours. Sample question: "Someone has used swear words or insulted me on the Internet, social media or WhatsApp".

Collective LGBTQ+ cybervictimisation. As a result of the information obtained in the first qualitative stage with focus groups, an ad hoc Likert-type item was added to assess collective LGBTQ+ cybervictimisation: "I have felt upset because someone posted a meme on the Internet, social media or WhatsApp making fun of someone or something related to the world of LGBT".

Internalised homonegativity. To assess negative beliefs and feelings regarding LGB orientations, from oneself or from others, we used the Spanish version (Vinces, 2016) of the Lesbian, Gay, and Bisexual Identity Scale (LGBIS; Mohr & Kendra, 2011), which includes five Likert-type items (α=.77) with five response options, ranging from “Strongly disagree” to “Strongly agree”. Sample item: "If it were possible, I would choose to be heterosexual." The research team also added a sixth item: "My life would be easier if I were heterosexual" (α=.78). In addition, the scale instructions were tweaked slightly so they could also be completed by non-LGB people, by asking them to respond through the eyes of how an LGB person would answer, which involves attributing internalised homonegativity.

Sex and gender identity. Following previous studies (Bradlow et al., 2017; Kosciw et al., 2018), a direct item was used to assess sex identity with the response options "male" or "female". For gender identity, a direct item was used: "Do you consider yourself...", with the response options being "boy", "girl", "both" or "none".

Sexual attraction. In line with other researchers (Collier et al., 2013), we followed the recommendations given by Austin et al. (2008) to measure sexual attraction with the item: "Generally, do you feel romantic and/or sexual attraction towards...", with the response options: "boys", "girls", "boys and girls", "people, regardless of their sex or gender", "neither boys nor girls" or "I'm not sure".

Procedure

This study was approved by the University of Jaén Ethics Committee (DIC.18/1.PRY). Incidental sampling was used. Collaboration was requested from a large number of schools via email and telephone. In total, 19 secondary state schools agreed to participate and, before the data was collected, informed consent was obtained from the families and pupils.

The first qualitative stage (Objective 1) took place in eight schools, with 27 focus groups of four to 10 participants each from April to June 2021 (23 face-to-face and four online), in sessions lasting approximately 50 minutes each. Four out of the eight schools had no previous experience of conducting awareness-raising activities about sexual diversity (in total, 13 focus groups), whereas four did (14 focus groups). At the beginning of each focus group, the pupils were asked to use an alias for the recording, and it was emphasised that the aim was to learn about their perceptions of everyday school life and not their personal experiences. The main focus for exploration in the focus groups, therefore, was the pupils’ general perception about LGBTQ+ cyberbullying and the type of aggression involved. The discussions were recorded, and the audio was later transcribed. Using a "bottom-up" approach, we used the views expressed by the pupils, for example, the allusion to collective LGBTQ+ cybervictimisation, to specify the choice and extension of the instruments used for the quantitative study, which allowed us to contextualize the phenomenon better.

For the second quantitative stage (Objectives 2 & 3), the study had a cross-sectional, prospective, and unified ex post facto design (Montero & León, 2007), with a fixed minimum number of LGBTQ+ pupils, taking into account the percentages found in previous studies (Garaigordobil & Larrain, 2020), as well as the size required to avoid exceeding a sample error of +5.5%, with a confidence interval of 95.44% (Osuna et al., 1991). We used two formats for the data collection: online, via a SurveyMonkey link, or using a printed questionnaire. In both cases, the pupils completed a series of questions during school hours about their involvement in cyberbullying in general, as well as in LGBTQ+ cyberbullying, their experience of LGBTQ+ prejudice, and their appreciation of ASBGD in their school. During the data collection, we stressed the anonymous and voluntary nature of the study, the confidential treatment of the data, and the importance of answering the questions honestly.

Data analysis

In the first stage, we explored the qualitative data through thematic analysis and basic descriptive analysis, using the Nvivo v.1.6.1 software. Thematic analysis is a method for identifying and analysing themes within data, which organizes and describes the data in detail (Braun & Clarke, 2006, 2014). To provide a result for Objective 1, we first identified the most relevant dimensions to include in the quantitative study, which were as follows: a) various types of aggression, from general personalised aggression to LGBTQ+-phobic aggression towards other people; and b) the full spectrum of sex-gender diversity, including sex, sexual orientation, gender identity and expression.

In the second stage, we used SPSS v.26.0 software to analyse the quantitative data. In the quantitative study, to meet Objective 2, we calculated the mean values of cybervictimisation and internalised homonegativity, and then recoded the variables related to sexual orientation, gender identity and expression, and the integral ASBGD variable. Specifically, sexual orientation was calculated by crossing data from the variables of gender identity and sexual attraction to obtain the values for "heterosexual", "lesbian/gay", "bisexual", "pansexual", "asexual" and "questioning". Gender identity was calculated by crossing the data of the variables sex and gender to obtain the values for "cis boys", "cis girls", "transgender boys", "transgender girls", "bigender" and "agender". Gender expression was calculated by crossing the gender identity and expression data, obtaining the values for "concordant", "androgynous" and "discordant". From this data, the integral ASBGD variable was calculated by crossing sexual orientation, gender identity and expression, making a distinction between “LGBTQ+” and “CH”. The first category included people whose orientation, identity, or expression diverged from heteronormative ones, that is, lesbian/gay, bisexual, pansexual, asexual, questioning, trans, and heterosexual cisgender people with discordant gender expression; the latter included heterosexual cisgender people and those whose expression agreed with their gender or which was androgynous. Next, we performed the basic descriptive analyses and comparisons of means, using non-parametric tests, given the non-normality of the data. To obtain this, we used Student's t-test for independent samples (or, where appropriate, Welch's t-test) to analyse general cybervictimisation and collective LGBTQ+ cybervictimisation based on ASBGD. Cohen's d effect sizes were also calculated.

Likewise, one-factor ANOVA tests (or, where appropriate, Welch's t test) were performed to analyse the differences in means between general cybervictimisation and collective LGBTQ+ cybervictimisation based on sexual orientation, identity, and gender expression. Effect sizes were calculated using eta squared, and post hoc comparisons were made using the Games-Howell test. To give a result for Objective 3, we designed a simple theoretical moderation model (Figure 1), in which we looked at the moderating role of internalised homonegativity (W) in the relationship between general cybervictimisation (independent variable, X) and collective LGBTQ+ cybervictimisation (dependent variable, Y). Model 1 was tested using the PROCESS v.4.0 macro for SPSS (Hayes, 2017). First, to define Model 1, 10,000 bootstrap samples were selected, using a 95% confidence interval. Next, following the recommendations by Davidson & MacKinnon (1993), we included standard error estimators consistent with heteroscedasticity (HC3) to examine the interaction effect on the variables included in the model. To run the model, we decided to focus the continuous variables on a mean of "0" to make it easier to interpret the direct and indirect effects (Hayes, 2017), as well as integrating the Johnson and Neyman technique to examine their region of significance and simple slopes (Carden et al., 2017). Direct and indirect effects were considered significant if p<.05 and only when the lower and upper limits of the confidence interval did not contain the value "0". This theoretical model was tested on the general sample, with ASBGD as the covariate (0=LGBTQ+ and 1=CH), and it was then repeated with both subsamples, LGBTQ+ and CH, respectively, omitting the ASBGD covariate in both cases.

Perceptions of the nature of LGBTQ+ cyberbullying

The information extracted from the focus groups helped to identify the dimensions which, in the pupils’ eyes, are the most relevant for researching LGBTQ+ cyberbullying. The most important of these included those related to the nature of sex-gender diversity, the vulnerable groups of cyberbullying according to ASBGD, the types of aggression, and the reasons for LGBTQ+ cybervictimisation.

When asked about the nature of sex-gender diversity, the participants highlighted gender identity and expression as key aspects, as well as sexual orientation. However, they also showed prejudices, such as the association between gender expression and sexual orientation, and stereotypes about what is masculine and what is feminine. The following examples illustrate some of these perceptions:

• “There are stereotypes and you’re supposed to comply with them. Men are meant to be, like, more masculine and women more feminine, but I don't agree with that. I think all of us are people and it doesn't matter if you like someone or not, or if you act in a more masculine or more feminine way” (Boy, 13 years old).

• “Some people get noticed because of their gestures, but some of these can be misinterpreted. For example, I’ve got two friends who are like this, and they don't have to be lesbians. But there are other people who, even if they don't say so specifically, you can tell by what they’re like, by how they speak...” (Boy, 13 years old).

Regarding the groups most vulnerable to cyberbullying, there were young people in all the focus groups who reported that people from sexual minorities are more likely to be attacked, especially boys. The following examples illustrate some of these perceptions:

• "I think people who show their sexual diversity get attacked more than people who are heterosexual (...) and I think trans people tend to get bullied even more" (Girl, 15 years old).

• “I think that men attack other men more” (Girl, 16 years old).

Among the different types of cyberbullying attacks, they considered the most common to be verbal aggression or social exclusion and pointed out that their heterosexual cisgender peers tended to play down behaviour considered offensive by LGBTQ+ youth through jokes or normalisation. Likewise, in various focus groups, acts of aggression towards collectives were mentioned in which a person witnesses’ situations of cyberbullying towards other people due to their affective-sexual diversity. The following examples illustrate some of these perceptions:

• "If a type of orientation is used as an insult, then clearly, the person who it’s aimed at, or someone who hears it without it being directed at them, thinks, well, this must be bad, I’d better shut up and keep it to myself" (Girl, 15 years).

• “There are a few kids in class who tease each other and call each other ‘puff’ and things like that. I’ve sometimes wanted to tell them not to make those comments, because although they say it as a joke, they’re perfectly capable of saying it to someone on the street because of their clothes and make that person upset. Sometimes, even though they don’t say it to you, and it’s aimed at someone else, it still bothers you” (Boy, 13 years old).

As regards the reasons for cybervictimisation towards LGBTQ+ youth, they highlighted internalised homonegativity and negative beliefs and emotions towards the LGBTQ+ collective. The following examples illustrate some of these perceptions:

• “Apart from the fact that you see your insecurities reflected in the other person, the way you’re brought up is also a big influence. If you grow up in a house where everyone is very traditional and where being gay or lesbian is seen as wrong, you grow up with those ideas” (Girl, 15 years old).

• “Because many people say that if you were born a man and want to be a woman, you have to just put up with it. You have to just be a man, because it's not ‘natural’, in inverted commas” (Girl, 13 years old).

General cybervictimisation and collective LGBTQ+ cybervictimisation, according to ASBGD

Taking ASBGD into account, no significant differences were found in cybervictimisation experienced in the form of general aggression aimed at LGBTQ+ and CH youth (t[574.65]=−1.72, p=.085; M_LGBTQ+=0.23, DT= 0.34; M_CH=0.27, SD=0.38), but differences were found in collective LGBTQ+ cybervictimisation (t[574.65]=9.65, p<.001, d=.49), with a medium effect. LGBTQ+ youth feel significantly more affected by LGBTQ+-related cyberbullying (M_LGBTQ+=1.10, SD=1.35; M_CH=0.45, SD=0.87).

With regard to the dimensions of ASBGD, significant differences were found based on all of them in collective LGBTQ+ cybervictimisation. In particular, regarding gender identity, the greatest differences in collective LGBTQ+ cybervictimisation were found among cis girls, who scored higher than cis boys and transgender girls.

According to sexual orientation, pansexual, lesbian/gay, and bisexual people were the most affected and, regarding gender expression, people with discordant or androgynous gender expression scored highest in this type of cybervictimisation. As for general cybervictimisation, although significant differences were found based only on gender identity and sexual orientation, the effect size was minimal (Table 2).

Internalised homonegativity in the general population

Model 1, from the general sample, was statistically significant, F(4.1882)=57.67, p<.001, accounting for 16.02% of the variability in the data. ASBGD showed a negative effect on collective LGBTQ+ cybervictimisation [β=−.634, t(4.1882)=−9.86, p<.001], indicating that the LGBTQ+ subsample has a higher risk of feeling hurt in this way.

• Direct effects: As can be seen in Table 3, Ordinary Least Squares (OLS) regression analyses revealed that overall cybervictimisation had a positive effect [β=.776, t(4.1882)=9.57, p <.001] on collective LGBTQ+ cybervictimisation, while internalised homonegativity had a negative effect [β=−.139, t(4.1882)=−5.84, p<.001]. In other words, a high level of general cybervictimisation and a low level of internalised homonegativity have a strong relationship with feeling hurt or considering oneself a victim of cyberaggression towards the LGBTQ+ collective.

• Indirect effects: as can be observed in Figure 2, the indirect effect revealed a negative moderation of internalised homonegativity in the association between general cybervictimisation and collective LGBTQ+ cybervictimisation [β=−,188, t(4,1882)=− 2.10, p<.05]. In other words, low scores in internalised homonegativity, or in attribution of internalised homonegativity, moderate the association between being a victim of general cyberaggression or feeling hurt by cyberaggression related to the LGBTQ+ collective.

Internalised homonegativity in relation to ASBGD

Model 2, from the LGBTQ+ subsample, was not statistically significant, F(1,456)=.010, p=.921 (Figure 3), which indicates that, among LGBTQ+ people, feeling harmed by collective LGBTQ+ cybervictimisation is not explained by general cybervictimisation or internalised homonegativity.

Model 3, from the CH subsample, was statistically significant, F(3.1423)=30.68, p<.001, accounting for 9.25% of the variability of the data.

• Direct effects: As shown in Table 3, the Ordinary Least Squares (OLS) regression analyses revealed that overall cybervictimisation had a positive effect [β=.597, t(3.1423)=7.23, p<.001] and internalised homonegativity a negative effect [β=−.146, t(3.1423)=−6.34, p<.001] on collective LGBTQ+ cybervictimisation. In other words, a high level of general cybervictimisation and a low level of attribution of internalised homonegativity have a strong relationship with feeling harmed or considering oneself a victim of cyberaggression towards the LGBTQ+ collective.

• Indirect effects: as can be seen in Figure 4, the indirect effect revealed a negative moderation of internalised homonegativity in the association between general cybervictimisation and collective LGBTQ+ cybervictimisation [β=−.178, t(3.1423)=−1.99, p<.05]. That is, low scores in the attribution of internalised homonegativity, or perception of the experience of the LGBTQ+ collective, moderate the association between being a victim of general cyberaggression and feeling, in addition, harmed by LGBTQ+ cyberaggression.