Online First

With the emergence and popularity of social media, female online users (especially the younger generation) have opportunities to exchange information with other users, express their thoughts, share their perspectives and personal experiences of gender discrimination, and challenge stereotypes against females. For feminist activities in China, social media have become an important channel for discussing women's rights, drawing attention to feminist activities, and avoiding the strict censorship of feminism from the government. Feminists in China have also engaged in collective action to criticize media content that reproduces stereotypes and prejudices against females (Mao, 2020; Wang & Driscoll, 2019; Wu & Dong, 2019). Recently, as in Western countries, #MeToo movements were popular on Chinese social media such as WeChat and Weibo. Female online users and feminists have actively engaged in this grassroots movement (Zeng, 2020). After the #MeToo movement, women have had more chances and are more willing to speak up and defend their rights, thereby raising awareness of gender issues in China among the public (Lin & Yang, 2019). Afterwards, various movements against gender discrimination have become popular and widely shared among social media users (e.g., the Naked Chest against Domestic Violence movement and the housewives’ empowerment movement (Hou, 2020; Ling, 2022).

However, feminist activities have often been regarded as a threat to the stability of the social system and national security by government authorities in China (Yang, 2014). As a result, censorship of feminism has been intensified. When social media posts include radical perspectives on gender issues, they have been removed without notice, and sometimes the accounts have become suspended (Mao, 2020). In addition, there has been a backlash against feminist activities in the form of attacks on the principles of feminism and gender equality on social media and female online users being exposed to harassment (Turley & Fisher, 2018).

Specifically, social media have enabled the circulation of sexist and misogynistic narratives and humiliation against women, thus reproducing the power inequalities in society (Boynton, 2012). Female users have been frequently harassed by other users when sharing their viewpoints and experiences related to sexism and injustice online (Turley & Fisher, 2018). Meyer and Cukier (2006) have found that using female usernames significantly increased the chance of getting messages that include threats or sexual harassment, meaning that female online users tend to face online harassment and hatred more than male users.

In the comment section, women have often been subjected to incivility and aggressiveness. For instance, female politicians are targeted more frequently for uncivil online comments than male politicians (Rheault et al., 2019). Moreover, female journalists have reported that they often receive sexist comments when interacting with their audiences online (Chen et al., 2020). As female users are more likely than men to be harassed online and attacked due to their gender, Van Duyn et al. (2021) employed a survey of commenters and comment readers. They found that women are less likely than men to comment on news sites. Proust and Saldaña (2023) discovered that male users tend to leave more uncivil comments than female users. Thus, there exists a gender gap in the comment section. However, there has not been much research on how female users perceive and react to uncivil comments. Thus, we examine how the current situation affects the identity of gender groups and other consequences among female online users. The next section discusses gender identity based on the social identity theoretical framework.

Social identity theory has been developed to examine intergroup relations (Hogg et al., 2004). According to Tajfel (1972), social identity is defined as an individual’s knowledge that he or she is a member of certain social groups, along with the value and emotional significance of that membership. When individuals embrace their identity as members of a social group, they categorize themselves as a member of the group rather than as a unique individual (Turner, 1999). Depending on the importance of a particular group membership in one’s self-concept, the strength of social identity can be different.

Among different types of social identities, this study focuses on gender identity. Different from biological sex, gender is the cultural definition of conduct, which has its own culturally constructed group norms and stereotypes as females or males (Lerner, 1986). When individuals have strong gender identities, they understand themselves as being related to the culturally ascribed feminine and masculine meanings attached to their gender group and act in accordance with these gendered aspects of themselves (Wood & Eagly, 2015). In this context, gender identity as a group identity is a self-conceptualization derived from acknowledging oneself as a member of one’s gender group and all meanings and values associated with that gender group (Maldonado et al., 2003).

Similar to other social identities, when gender identity becomes activated, individuals’ judgments, attitudes, and behaviors all tend to align closely with the stereotypes of their gender group (White & Gardner, 2009). Given that conflict and inequality situations might increase the salience of social identity (Abrams & Hogg, 2010) due to the relative minority position of females in society, females tend to identify more strongly with their gender group. This salient gender identity might lead them to take part in coordinated social action to overcome their group-based disadvantages, including feminist causes (Van Zomeren et al., 2008).

With the emergence and popularity of online media, scholars have examined how individuals’ identification with social groups might change, and how it could affect their participation in collective action to advocate for their social groups in online and social media environments (e.g., Dai & Shi, 2022; Mikal et al., 2016; Triwibowo, 2018). To understand how Chinese online users' collective action changes based on their identification with gender groups, this study employs the SIDE model (Reicher et al., 1995; Spears & Lea, 1994; Spears & Postmes, 2015).

The SIDE model proposes that, in (particularly anonymous) computer-mediated communication environments, individuals have a tendency to identify themselves as a group member, while generalizing others as representations of a specific social group. As personal details are unidentifiable in online environments, increasing levels of anonymity would minimize individuals’ self-awareness and increase their awareness of the salient norms of a social group. In this process, individuals not only have a high level of identification with their social groups but also see others as social group members instead of unique individuals due to the blurring of individual differences (Spears, 2017). By going through this process, individuals have a strong shared social identity in online and social media environments, and view themselves as a group member, which refers to the “depersonalization” process in the SIDE model (Mikal et al., 2016; Spears, 2017). Previous studies have consistently shown positive relationships between anonymity and group identification (Postmes et al., 2001). The SIDE model also explains that online users show a tendency to rely heavily on a few cues that are available, considering that there are not many social cues they can use in an online environment.

Concerning the strategic dimensions of the SIDE model, scholars have argued that online users would consciously manage and strategically choose to express or conceal their identity (i.e., identity performance), depending on context and power dynamics (Klein et al., 2007; Spears et al., 2002, 2011). For example, when the discussion topic was relevant to males, it was dominated by males, and vice versa (Postmes & Spears, 2002). In addition, women showed a tendency to hide their gender identity when the discussion topic was related to male expertise (Flanagin et al., 2002; Spears et al., 2011).

Thus far, scholars have applied the SIDE model to understand how online users identify themselves with social groups and behave in certain ways based on the norms of social groups (e.g., Chung, 2019; Rösner & Krämer, 2016; Seiter & Brophy, 2022). For example, Seiter and Brophy (2022) conducted a content analysis of posts uploaded on Reddit, YouTube, and Facebook and found that Reddit users showed less aggressive and more supportive communicative behaviors. They explained that Reddit users identif y themselves as representatives of the platform, with a higher level of anonymity than other platforms; thus, they are more likely to follow the norms of the platform. Similarly, based on an experiment by Rösner and Krämer (2016), they found that online users showed more aggressive wording in the comment section when other users’ comments were aggressive. The anonymity did not have a direct effect on aggressiveness here; rather, they found that users showed a tendency to conform to aggressive social norms of commenting in the anonymous condition. Chung (2019) also revealed that when comment readers identified themselves with those who posted comments, the comment readers showed a strong tendency to adjust their attitudes and perceptions following those of the commenters. Thus, these three studies show that online users would identify themselves as users when using the same platform and would follow other users’ behaviors.

Additionally, several studies have used the SIDE model to explain how comments or posts targeting individuals’ in-groups would affect their group identification as well as their emotional and behavioral reactions (e.g., Lee, 2021; Rösner & Krämer, 2016). Individuals’ identification with their social groups, and their emotional and behavioral responses based on identification would be enhanced when conflict and inequality occur (Abrams & Hogg, 2010). For instance, Postmes and Spears (2002) showed that identification with one’s gender group was stronger when a gender-related stereotype was provided, and when group members were depersonalized (i.e., anonymous). Lee (2021) conducted an experiment and found that exposure to hate messages on social media targeting White people triggers negative emotions among White participants. Interestingly, the effect was reduced when personal information regarding the source was provided.

Based on the social identity and SIDE models, people would identify themselves with their social groups in (especially anonymous) online environments. Particularly when people interact with their in- or out-group members, or with other situational signals (e.g., stereotype information or hate speech to their in-group), their group identity becomes activated. Given that female online users are oftentimes subjected to abuse in online comment sections due to incivility and aggressiveness against their gender group, gender identity might be activated in online and social media environments, especially when female users read or become exposed to aggressive comments against females. Thus, this study proposes the following hypothesis:

• H1. When women are exposed to aggressive online comments against their gender group, their gender identity becomes salient.

The next section discusses the effect of being exposed to aggressive online comments on female users’ emotional reactions and behaviors.

Having negative emotional reactions, such as anger or fear, has been one of the most common reactions when encountering aggressiveness (Lee, 2021; Phillips & Smith, 2004). In the online environment, previous studies have found that being exposed to aggressive online comments tends to evoke negative emotions among social media users (e.g., Kim & Kim, 2019; Lee, 2021; Nadal et al., 2013).

Emotion has been regarded as crucial for understanding intergroup relations (Mackie & Smith, 2018). Previous studies have found that individuals’ emotions in an intergroup situation would be related to a group with which they identify, and the intensity of identification with the group would be positively related to their group-based emotions (Kim & Kim, 2019; Lee, 2021; Mackie & Smith, 2018). For example, Kim and Kim (2019) conducted an online experiment and found that participants who were exposed to uncivil disagreeing comments reported higher levels of negative emotions, compared to those who were exposed to civil disagreeing comments. Lee (2021) also discovered that seeing hateful tweets directed at White people on Twitter caused unpleasant feelings such as anger and fear.

With respect to the gender group, when individuals faced aggressive behaviors against their gender group, they also showed negative emotional reactions (Nadal et al., 2013). Nadal et al. (2013) conducted focus group interview studies to examine how women reacted to gender microaggressions; they found that women reported feelings of anger, guilt, humiliation, and discomfort.

Based on previous studies on emotional reactions in an intergroup situation, we can assume that when online female users see uncivil comments related to a gender issue, they may feel negative emotions. In addition, this study examines whether this pattern would be stronger when the incivility is targeted at their gender group. Although there is no research comparing the effect of targeting different groups in aggressive comments on individuals’ emotional reactions and behaviors, previous studies have suggested that individuals’ group identity is activated when their in-group is attacked and targeted (e.g., Lee, 2021).Thus, the following hypothesis and research question are proposed:

• H2. When women are exposed to aggressive online comments, they show negative emotional reactions, compared to those who are not exposed to aggressive online comments.
• RQ1. When women are exposed to aggressive online comments against their gender group, do they show negative emotional reactions more, compared to those who are exposed to aggressive comments against males?

With the emergence of social media and the popularity of online feminist activism, gender issues have been actively discussed online. Users not only consume online content and discussion but also actively engage in discussion and different types of collective action (Majumdar et al., 2022). In this process, gender identity has been regarded as an important factor leading to online collective action. By extending the SIDE model, Klein et al. (2007) proposed that social identity performance strengthens group identity and mobilizes group members to engage in supporting certain actions in the pursuit of group goals. Previous research using the SIDE model has already shown that individuals’ group identity could affect their behaviors in supporting and defending their in-group (Mikal et al., 2014; Täuber & Van Zomeren, 2012).

As another theoretical framework, this study uses the Social Identity Model of Collective Action (SIMCA) to examine the relationship between social identity and collective action (Van Zomeren et al., 2008). Collective action, as a socio-psychological phenomenon, refers to any action performed by individuals who identify as group members for the goals of the group, including social change (Van Zomeren, 2016). SIMCA proposes that perceived injustice, perceived efficacy to make changes, and a sense of social identity would contribute to individuals’ engagement in collective action (Van Zomeren et al., 2008). Previous studies have argued that identifying with a disadvantaged group would generate emotional reactions to perceived injustice (e.g., group-based anger), and thus would solidify their commitment to collective action (Adra et al., 2020; Chan, 2016; Hercus, 1999; Reicher, 2001; Smith et al., 2008; Thomas et al., 2012). More importantly, identifying with their social group or with those sharing the same opinion would strengthen their dedication to collective action (Brügger et al., 2020; Eisner et al., 2022; Lee et al., 2022; Simon et al., 1998; Uluğ et al., 2020). For example, Lee et al. (2022) conducted an online survey with Asian Americans; they found that Asian Americans’ group identity increased their perceived injustice, efficacy, and situational motivation and was positively associated with their willingness to engage in online activism (via social media) in regard to antiracism, which also fostered offline activism. Uluğ et al. (2020) conducted a survey with protesters and also discovered that protesters’ identification with women’s rights defenders affected their willingness to engage in collective action.

By examining actions that advocate and show support for one’s in-group, this study focuses particularly on corrective action in the comments section and in online and offline activism. First, corrective action refers to people’s attempts to correct the influence of what is presumedly wrong in the public sphere when they see media as being biased and influential (Barnidge et al., 2020). Although no research has found that identification with a group would be related to their corrective actions, a few scholars have focused on how users put effort into stopping contagious behaviors by focusing on corrective actions taken by online users (Naab et al., 2021; Ziegele et al., 2020).

Specifically, Naab et al. (2021) found that being exposed to uncivil comments encouraged online users to engage in corrective actions (e.g., disagreeing in the reply comments, sharing a video that disagrees with the previous comment, etc.) and show support for restrictive actions by the authorities, such as censorship or regulation. Ziegele et al. (2020) examined factors that influenced online users’ intention to engage in the social movement #ichbinhier as a collective corrective action against uncivil comments. They discovered that feelings of personal responsibility and perceived group efficacy were important factors for engagement in corrective action. Although they did not find the effect of a shared social identity as a member of the #ichbinhier movement, they explained that this might be due to the notion that identifying as a member of social activity might not be as strong as other group categories, such as race or gender. Still, the activation of a social identity would be likely to lead female online users to engage in collective action against threats to their social group (Hogg et al., 2004; Tajfel & Turner, 1979). Thus, when female online users were exposed to aggressive comments (especially those attacking their gender group), they might engage in corrective actions by themselves and support actions by social media platforms or the government. Thus, we post the following hypotheses and research questions:

• H3. When women are exposed to aggressive online comments, their engagement in corrective actions is higher, compared to those who are not exposed to aggressive online comments.
• RQ2. When women are exposed to aggressive online comments against their gender group, is their engagement in corrective actions higher, compared to those who are exposed to aggressive comments against males?
• H4. When women are exposed to aggressive online comments, their support for corrective actions by (a) social media moderators and (b) the government is higher, compared to those who are not exposed to aggressive online comments.
• RQ3. When women are exposed to aggressive online comments against their gender group, is their support for corrective actions by (a) social media moderators and (b) the government higher, compared to those who are exposed to aggressive comments against males?

As another behavioral reaction against aggressive online comments, this study examines online and offline activism. Online activism (i.e., collective action that relies on online platforms such as social media) and offline activism (i.e., attempts to resolve controversial problems through communication and collective action in physical locations) can be regarded as forms of collective action (Margetts et al., 2015). Examples of online activism include a range of activities including commenting, passing on the received information, “liking” information, joining an online group, and uploading materials, which are typically done on social media. Due to the ease of sharing information and connecting with others, online and social media platforms enable individuals to galvanize the public (Bennett & Segerberg, 2011). Thus, engaging in online action would help create a sense of collective identity, solidarity, and belonging among participants (Ghobadi & Clegg, 2015). In contrast, offline activism behaviors, such as attending protests or signing a petition, require a higher level of commitment in terms of time and effort (Nekmat et al., 2015).

Prior research on SIMCA has suggested positive relationships between identification and online and offline collective action (Brügger et al., 2020; Eisner et al., 2022; Lee et al., 2022; Simon et al., 1998; Uluğ et al., 2020); yet, little research exists in terms of how exposure to aggressive comments targeting a specific group affects engagement in online and offline activism. However, being exposed to aggressive comments would enhance identification with individuals’ social groups, which would raise their willingness to engage in group-based collective action. Thus, we posit the following hypotheses and research questions:

• H5. When women are exposed to aggressive online comments, their engagement in online activism behaviors is higher, compared to those who are not exposed to aggressive online comments.
• RQ4. When women are exposed to aggressive online comments against their gender group, is their engagement in online activism higher, compared to those who are exposed to aggressive comments against males?
• H6. When women are exposed to aggressive online comments, their engagement in offline activism behaviors is higher, compared to those who are not exposed to aggressive online comments.
• RQ5. When women are exposed to aggressive group, is their engagement in offline activism higher, compared to those who are exposed to aggressive comments against males?

Students at a Sino-foreign cooperative university in China were invited via an online announcement on WeChat Moment (social media feed) to share their opinions about social media. They voluntarily participated in the online experiment through the link or QR code to access the survey in which our stimulus material was embedded. The data were collected from May 2 to May 16 in 2023. A total of 169 students who identified as female completed the survey. The mean age was 20.19 (SD = 1.93). All survey questionnaires were in English and Chinese. The present study was approved by the Institutional Review Board at the Singapore Institute of Technology.

Depending on the experimental condition, participants were randomly assigned to view one of six fictitious Weibo web pages (See Figure 1).

Figure 1. 
Example of Stimuli (Aggressive Comments against Women)

The Weibo pages include one non-aggressive post from a hypothetical news organization about marriage in China, followed by five online comments from different Weibo users. The post from the hypothetical news organization delivered factual information about the decreasing number of people getting married in China, which was adapted from social media posts published by news organizations. Regardless of their condition, participants read the same post from the news organization. Participants then read comments from different Weibo users presented in a social media post format. Depending on the condition to which they were assigned, they received different types of comments. All posts and comments were in Chinese, and the profile pictures and names of all user accounts were gender neutral.

For the manipulation of aggressiveness of comments, those assigned to the non-aggressive conditions were shown comments that did not include any aggressive words, whereas participants assigned to the aggressive conditions were shown comments with a more aggressive and uncivil style, which included words in capital letters to illustrate shouting, attacks toward specific gender groups, offensive language, and sarcasm. This manipulation is based on Rösner et al. (2016). Except for the level of aggressiveness, the non-aggressive and aggressive versions of the comments were equal regarding the contents. We conducted pilot studies twice with 65 students to make sure that each comment was regarded as non-aggressive or aggressive before designing the stimulus.

Regarding the manipulation of the target group mentioned in the online comments, those assigned to the “against women” conditions were shown comments that attacked the female gender group [a pair of three comments against men and two comments with no specific target group], while those assigned to the “against men” conditions were shown comments that attacked the male gender group [a pair of three comments against women and two comments with no specific target group]. Those assigned to “both” conditions received comments that attacked both male and female gender groups [a pair of two comments against women and two comments against men and one comment with no specific target group]. After viewing the post and comments, participants completed the questionnaire measuring their gender identity perceptions and engagement behaviors with regard to the comments.

Unless noted otherwise, all variables were measured on seven-point scales ranging from ‘1 = strongly disagree’ to ‘7 = strongly agree’.