Background

Hashtag activism is a form of digital activism referring to the use of hashtags on social media platforms to raise awareness about a particular cause or issue and to facilitate grassroots organizing around that cause (Yang, 2016). Hashtags have become an increasingly popular form of activism as they allow individuals to easily participate in social and political movements, and they amplify the voices of common citizens through modern technology. Hashtag activism can also support community-building and spark political action within specific online communities (Thompson, 2020). In fact, the internet and social media have become new outlets for political and social participation, especially for women (Norris, 2002, 2020; Norris and Curtice, 2006; Theocharis et al. 2023). Hashtag feminism is a form of digital activism that uses hashtags to draw attention to issues affecting women and gender equity and to protest mistreatment, misogyny, and inequity (Khoja-Moolji, 2015).

#MeToo is one of the most well-known examples of hashtag feminism, highlighting the experiences of sexual harassment and assault among women, primarily in the workplace (Kaufman et al. 2021). Other notable hashtag feminism efforts include: #BeenRapedNeverReported (Keller et al. 2018; Mendes et al. 2018), which provided a space for girls and women to share their stories of sexual violence and explain why they did not report their experiences (Mendes et al. 2018); #WhyIStayed, used to share personal experiences about abuse to combat victim-blaming (Linabary et al. 2020); #NiUnaMenos, used to exhibit and fight against the high rates of femicide in Mexico and later internationally (Puente et al. 2021); and #MahsaAmini, used to raise awareness and garner solidarity to protest violence against women (VAW) perpetuated by Iran’s morality police following the death of Mahsa Amini, who was arrested for allegedly not complying with hijab laws and who later died in custody (Kermani, 2023). The present study explores the hashtag #YesAllWomen as a form of global hashtag feminism.

History of social media activism

Social media activism, whether hashtags, group organizing, or fundraising, has a short but dynamic history. The concept of using technology and the internet for activism and social change dates back to the late 20th century, with the rise of the internet and online communities. Early instances of online activism emerged in the late 1990s, when individuals started using the internet to organize and mobilize for political and social causes. The global protests against the Iraq War in 2003 marked the beginning of a new era in online activism, as activists used the internet to coordinate protests and disseminate information to a wider audience (Bennett et al. 2008; Nah et al. 2006).

The rise of social media platforms such as Facebook and Twitter (now X) in the mid-2000s indicated a major turning point in the history of digital activism (Sandoval-Almazan and Gil-Garcia, 2014). These platforms provided a new space for individuals to connect, share information, and organize for social and political causes. Activists started using these platforms to mobilize people around specific issues, such as human rights, climate change, and politics. For example, in 2009, the use of Twitter was instrumental in the Iranian Green Movement, during which millions of Iranians took to the streets to protest election fraud (Pew Research Center, 2009). Furthermore, the Arab Spring, which began in 2010, is considered one of the most significant examples of digital activism, as the use of social media platforms played a crucial role in the coordination of protests and the dissemination of information (Arafa and Armstrong, 2016; Frangonikolopoulos and Chapsos, 2012).

More recently, the Black Lives Matter movement used social media to fight against routine police brutality and racial profiling inflicted on the Black community in the United States (U.S.) and to fight for the rights, dignity, and justice of all Black people—a group that has been historically oppressed, silenced, and discriminated against in the U.S. As users increasingly interacted with the movement on social media, engagement in on-the-ground protests also increased (Bowman Williams et al. 2021; Choudhury et al. 2021), suggesting a relationship between online and on-the-ground activism. Ever evolving, social media activism continues to grow as a critical tool for social change.

Theoretical framework

Critical mass theory within online activism may provide insight into the success of social media activism as a tool for political and social engagement and change. Critical mass theory refers to the idea of strength in numbers, or change-making occurring when enough people come together to contribute to the same effort. Using a critical mass approach to online activism considers initial conditions leading to the online activist space, any activities that occur to encourage or inhibit the activism (e.g., supportive responses (encourage) or bullying (inhibit) from other users), and resulting conditions that may include social and/or political change (Ghobadi and Clegg, 2015; Oliver, 2022).

Hashtag feminism can also be considered from the feminist standpoint theory perspective. Feminist standpoint theory affirms the idea that knowledge generated by oppressed groups contributes meaningful and necessary information to mainstream knowledge that does not have the insight of lived experience (Black Feminisms, 2021; Bowell, 2011; Crasnow, 2014; Harding, 2004). For example, women generating and sharing knowledge about women is necessary to create a full and true understanding of the experience of women. Coming from this perspective, the act of storytelling can be incorporated fluidly into feminist standpoint theory as one way of sharing and creating knowledge. Storytelling may be the first step to raising awareness about an issue that can then lead to political activism. When considering hashtag feminism, a single person’s post contributes to a larger quantity of individual stories across generations and geography, leading to a collective story that holds meaning and can act as the catalyst for social and/or political change (Thompson, 2020). Taken together, critical mass theory and feminist standpoint theory can help inform understanding of the rise of a recent hashtag, #YesAllWomen, as a form of online activism focused on women’s experiences with violence.

#YesAllWomen

The hashtag #YesAllWomen has been used for many years to raise awareness of issues affecting women. Its origins come from an incident that occurred in May 2014 in California, when 22-year-old Elliot Rodger murdered six women and injured seven more, after which he died by suicide. According to a note he wrote, he took these actions because women rejected his sexual advances throughout his life (Pendergrass, 2015; Phillips and Milner, 2017; Rodino-Colocino, 2014; Weiss, 2014). The atrocity raised public concern about how men view and treat women. Following the coverage of the murders, an outpouring on social media used an existing hashtag, #NotAllMen, to insist that not all men had the “Rodger mentality” or desire to punish women who refused them.

Social media users responded with #YesAllWomen to demonstrate that nearly all women live in fear of men to some degree as they move throughout the world (e.g., walking alone and taking public transport), even though not all men have the proclivity for VAW (Baer, 2016; Jackson et al. 2019; Phillips and Milner, 2017; Rodino-Colocino, 2014; Thrift, 2014). #YesAllWomen became a medium for women, particularly those in the U.S. and the United Kingdom, to discuss their experiences with violence, harassment, and abuse, refocusing the conversation on the structural and social inequalities leading to negative experiences for girls and women (Baer, 2016; Jackson et al. 2019; Pendergrass, 2015). #YesAllWomen also provided a space for society to engage in conversations that transcended digital spaces, resulting in a tool for enhancing female solidarity on the widespread issue of VAW (Jackson et al. 2019; Rodino-Colocino, 2014; Solnit, 2014; Thrift, 2014).

#YesAllWomen was used outside of the U.S. as well (Fink et al. 2016; Jackson et al. 2020). For example, on New Year’s Day 2017, a mass molestation was reported in Bangalore, India, with some government officials attributing the violence to women’s clothing and behavior (Najar and Raj, 2017). As a rebuttal, Indian women took to social media, revitalizing #YesAllWomen and encouraging others to share their stories of violence and abuse (Banerjee and Kankaria, 2022; Hitchings-Hales, 2017). Women have adopted this hashtag worldwide, using social media to raise awareness about VAW and to speak openly about topics such as misogyny and gender-based violence (GBV).

In this study, we explore how #YesAllWomen has been used to raise awareness and generate solidarity among women and victims of VAW. The study objective was to assess Twitter content using the #YesAllWomen hashtag for evidence that users were using the hashtag to speak up about the pervasiveness of VAW, create calls for collective action, and use the Twitter platform to demand policy and social change. Our hypothesis was that the use of the hashtag would create an online coalition of people (particularly women) voicing the need to address VAW through this new form of political and social participation. Using natural language processing (NLP) techniques, a subfield of artificial intelligence that allows computers to understand human language and to communicate with humans, we reviewed publicly posted Twitter content using #YesAllWomen to better understand users’ experiences and the ways the hashtag is being used to promote social change.

Methods

Data mining

With access to a 1% sample of publicly available Twitter data collected through the application programming interface (API; an interface or software that facilitates a certain list of functions inside another software application), we sampled tweets from May 2014 (the date of the first usage of #YesAllWomen) through the end of 2021. This end date was chosen to avoid changes in the platform—such as the process for verified accounts and changes in flagged and deleted content—during the acquisition initiated by Elon Musk in early 2022 (and subsequent change of the platform name to X). We filtered tweets using the language field of the metadata, where we only kept tweets that were written in English. We determined there were approximately 38,000 documents (tweets) using #YesAllWomen across 32 countries using the geographic tag of the tweets (geo-tagged), where geo-tagging was noted.

We preprocessed the collected tweets, where we first removed any URLs, handles, and “RT” (retweets; when someone shares another user’s tweet), then separated sentences into individual words (tokenized text) using TweetTokenizer (Bird et al. 2009; DeLucia, 2020). Our preprocessing pipeline included removing stopwords (Appendix A) using Natural Language Toolkit (NLTK; a group of programs and libraries designed for statistical and symbolic analysis of natural language) Python open-source module (Bleier, 2010; NLTK Project, 2023), where we added task-related stopwords such as YesAllWomen and emojis. Our frequency analysis of the top 100 frequent words showed “men” and “women” to be the most frequent (Fig. 1). As a result, and given the hashtag’s background, we also added these words to our stopwords, as their prevalence was too high, and they would subsequently appear in all topics later in the topic modeling procedure.

Fig. 1
figure 1

Top 100 words used in #YesAllWomen tweets.

Full size image

We also looked at the term-document frequency (tdf; the number of documents (in this case unique tweets) that contain a specific word) of the corpus to focus on more significant words; if the term did not appear frequently in most of the tweets, then the word is less likely to be an important word. We assessed tdf = [2, 5] by removing terms from the corpus when they appeared in less than 2 (or more than 5) tweets. When we used tdf = 5, the vocabulary size decreased from 9386 words to 2712 words, while tdf = 2 retained 4444 words. The frequency analysis showed that tdf = 5 was radically eliminating relevant vocabulary (e.g., races [“african”, “asian”], “brutality”, “bystander”, “confrontation”, “dehumanization”, “horrendous”, “violently”), while tdf = 2 removed less important terms such as: 10th, ab, acc, adding, ag, den, farm, hutton, iii, mango, sb, rife, zoo. Due to the short-sentence nature of Twitter, we wanted to capture as much information as possible; hence, we decided to use tdf = 2, removing fewer potentially important terms. The data preprocessing and cleaning pipeline resulted in 18,749 tweets to be used for analysis.

Topic modeling and selection

We used latent Dirichlet allocation (LDA), an unsupervised machine learning technique that helps identify the underlying topics present in a collection of text documents (Blei et al. 2003). In our work, the collection of tweets would represent the corpus of text documents where we aim to cluster the tweets into coherent topic groups. The MALLET topic model package wrapper implemented in Gensim Python version 3.8 (McCallum, 2002) was used to train LDA models on the cleaned Twitter data.

First, we ran a multiple empirical model training to find the number of topics (n) that could capture unique topics without repetition or nontangential grouping of tweets. N = {5, 10, 20, 25, 40, 50, 60, 100} were used to generate word clouds for each number of topics. The initial analysis concluded that n = 5 and 10 resulted in grouping non-cohesive tweets together, where each topic did not have a common theme. On the other hand, topic models with a large n caused the topics to be repetitive (i.e., there were multiple word clouds that included the same set of words).

Next, we tuned four hyperparameters of topic model (Appendix B): number of topics, burn-in, alpha, and beta.Footnote 1 Three models with topic_n = {20, 25, 40} were trained and clustered the data into the corresponding number of groups. Research team members reviewed word clouds for each group (n = {20, 25, 40}), noting which words and themes within each group were relevant and meaningful to the study aims. We first reviewed independently and then reviewed together as a team, comparing and rating the models by checking if the grouped topics were similar or if the grouped tweets within a topic were irrelevant or incoherent with each other (Fig. B1). After careful analysis and iterations, we opted to use n = 20, as these topics created a set of themes that were most in line with the study objectives. Then we conducted hyperparameter tuning on burn-in = {30, 50, 100, 200, 500, 1000}; alpha = {0.5, 1.0, 3.0, 5.0}; and beta = {0.01, 0.1, 1, 10}. The topic models were diagnosed using the following measures: document entropy, coherence, uniform distribution, corpus distribution, and exclusivity (Fig. B2). The qualitative researchers then reviewed 20 tweets for each of the 20 topics, and the team chose the topic with the most relevance to our study objectives.

Data analysis

We reviewed a set of 500 tweets from the chosen topic to create a codebook that described the different ways people used the hashtag. Two team members (ETOK, OF) explored the first set of 500 tweets to describe themes that arose within them. Making multiple passes through the data and regularly discussing insights and discoveries with the research team, we then categorized the different themes into codes that were more prominent within this initial set. We used Atlas.ti (v.22) for organizing and coding the data. Two team members (ETOK, OF) conducted a manual content coding of 5000 tweets from the model, during which each tweet was coded independently. By tweet 1500, only about 10% of the tweets remained relevant to our study aim. We coded an additional 200 tweets to confirm saturation. Coder agreement and reliability were confirmed by recoding/double coding, with each coder coding the same portion of tweets and then comparing results. Throughout the process, team members met regularly to discuss coding outcomes and to make decisions about any discrepancies.

This phenomenological study used thematic content analysis (Creswell and Creswell Baez, 2020; Leavy, 2022; Ravitch and Carl, 2020) to explore common themes within the coded data. Groups of codes with interconnected relevance were reviewed together and side-by-side to establish meaning, and reading through the data multiple times allowed for more clarity on the most prominent themes. Thematic frequency and connectedness provided the impetus for understanding relationships between and among the themes.

Results

Out of the 38,542 tweets used for the analysis, 393 contained geo-tagged country information (see Table 1 for details.) Despite the low number of tweets indicating a country location, there was broad geographic variability.

Table 1 Countries of origin for geo-tagged tweets (N = 393).
Full size table

Additionally, 35,809 tweets in the final data set indicated the user’s primary language was English. Most of the tweets written in English did not indicate their country (users are able to set their country information to private).

In the qualitative analysis, #YesAllWomen was found to provide a space on Twitter for people to share their experiences, ranging from feeling unsafe in certain spaces to enduring sexual assault. We identified four key themes in our analysis: trepidation in personal disclosure; backlash, hatefulness and trolling; unity and support; and hashtag activism. See Table 2 for definitions and salient quotes related to each theme.

Table 2 Themes and selected quotations.
Full size table

Trepidation in personal disclosure

Although many people felt safe and supported sharing their stories on #YesAllWomen, others struggled to share their experiences through the hashtag due to fear. Codes contributing to this theme included: abuse/harassment and the need for self-defense (n = 119), emotional reactions (n = 86), and fear and anxiety (n = 71). This fear manifested in various ways, including: fear of judgment (“After decades I still feel too intimidated to share my story for fear of being doubted and judged.”); fear of perpetrators (“Because a friend decided to delete her #YesAllWomen tweet for fear it would antagonize her abusive ex.”); fear of losing employment or other opportunities (“Passionate about #YesAllWomen but afraid of posting personal details that could affect jobs & my life. Mad props to those who do it anyway!”); fear of intimidation (“The fact that I even feared making a comment on this hashtag from fear of criticism or verbal abuse. #YesAllWomen”); fear of facing hate, threats, or sexual harassment both in real life and on social media (“I’m too afraid to even participate in this hashtag because of the hate/threats/sexual tweets I’ll wake up to tomorrow. #YesAllWomen”); and fear of shame and embarrassment (“Being anxious to contribute stories to #Yesallwomen because it’s still so ingrained that I should be embarrassed by my own harassment.”).

Other tweets provided no context or reason for the fear, only describing they were afraid to share their stories: “#YesAllWomen because I am still too afraid to post about my experiences.” Some users reported they could not share their experiences in other offline contexts due to shame, fear, embarrassment, privacy breaches, or a lack of closure: “I kept secrets out of shame. I kept secrets because it was normal. I kept secrets because no one would listen.” Twitter seemed to provide an outlet where these women could maintain anonymity and still have their experiences be heard and validated by those with a shared lived experience.

The emotions expressed in these tweets underscore the long-lasting effects of traumatic experiences and the need for proper resources and support for victims and survivors. For example, users explored their hope for a supportive and safe environment for survivors to come forward and share their stories without fear of judgment or retribution: “As a survivor of physical & sexual abuse, I ask all girls to never keep quiet from fear of blame/disbelief. Tell! Talk! Share! #YesAllWomen;” “Sometimes deciding to #RaiseYourVox is scary. But I encourage you to do it anyways. #YesAllWomen;” “#yesallwomen because we are stronger than our negative experiences. There are better men out there, and those are the ones we #love.”

Backlash, hatefulness, and trolling

#YesAllWomen was met with backlash and negative sentiment from some users toward the hashtag and its supporters. Codes integrated within this theme included: hatefulness/trolling (n = 206), misogyny (n = 164), backlash (n = 141), ignorance (n = 55), and victim blaming/shaming (n = 31). These tweets were characterized by misogyny, male entitlement, anger, and ignorance regarding the purpose of the hashtag or women’s experiences. In addition, the use of hashtags such as #NotAllMen or #YesAllPeople attempted to invalidate the message of #YesAllWomen and highlighted the idea that other groups outside of women were also oppressed: “Men butt in to women’s conversations with #notallmen ‘are like that’—so women reply ‘but all women #YesAllWomen experience it;’” “#YesAllWomen should really be #YesAllPeople or #YesAllHumans. Was done with this, but I keep getting it in my feed and messages;” “This #YesAllWomen thing is awesome and I support it but there is no support or attention for the men who are also raped…By men and women.”

Angry comments and insults were directed at supporters of #YesAllWomen, with some tweets even threatening people who shared their experiences: “#YesAllWomen because I only just posted some YAW tagged tweets about 30 min ago and already have angry men trying to speak down to me;” “This #YesAllWomen thing is really stupid in my opinion, because, as always, it’s feminists making assumption of all men, not a select few.”

Bullying and trolling (i.e., unsolicited, controversial tweets intended to insult and provoke other users) were also present, with insults directed at those who shared their stories or who supported the hashtag. The trolls showed a lack of understanding and appreciation for the hashtag and its message, often finding humor in serious topics such as rape: “#YesAllWomen because my support of this hashtag got me a barrage of negative tweets. I dared to criticize a joke about rape. How dare I…;” “Rape is funny and women should be forced to suck dick (this may not be used to piss off feminists) #YesAllWomen;” “God I hate these retarded #yesallwomen tweets #stopragingsexuallyfrustratedfeminists.”

Some tweets criticized users for sharing their experiences, calling them attention-seekers or doubting the validity of their stories, sometimes even discounting them based on “scientific research.” These tweets exposed how unrelatable some people felt #YesAllWomen was for them and at the same time provided more evidence of the reality that women’s stories are often disbelieved and discarded: “Oh look, an entire comments section of people talking about how their personal experiences trump scientific research.”

There was also a lack of awareness and understanding as to why #YesAllWomen was trending and why women wanted to share their stories: “Can we stop with the #YesAllWomen now please, I’m tired of reading naive feminist tweets;” “I’ve never met a feminist with good values. They’re all angry insecure people who left a hurt festering in their hearts 2 long #YesAllWomen.”

The backlash and hatefulness seen in response to #YesAllWomen demonstrate both a need for continued education to raise awareness and understanding about the experiences and struggles of women, as well as a need for a safer and more supportive environment for survivors.

Unity and support

Despite some negative feedback from unsupportive users, #YesAllWomen appears to serve as a tool for women to come together and support one another, inspiring others to keep moving forward despite the challenges they face. Codes considered in this theme included: sharing experiences/stories and speaking up (n = 279), insightful/inspirational/eye-opening (n = 134), and unity/support (n = 127). Tweets showed unity and support through words of encouragement, admiration, and reassurance for those who shared their stories in this online setting. Some offered support by retweeting and encouraging others to do the same; others extended words of love to those who shared, creating a positive space for people to comfortably share their difficult experiences; others provided helpful resources: “Your story is disturbing, extremely sad (till the end, anyway) and utterly inspiring. Walk in beauty and grace. #YesAllWomen;” “Thank you for your Slate article #YesAllWomen. I shared with every man I could think of (and a lot of women).”

The stories shared offered a call for the unification of women to bring more attention to these experiences and concerns. While some unified around their common experiences, others expressed solidarity with survivors: “#YesAllWomen gives me hope that ALL women will feel able to share each other’s experiences & gain strength in unity w one another;” “Women unite; turning stories of solidarity into unity. #YesAllWomen;” “#YesAllWomen unites women in pain and grief.”

The sharing of experiences was seen as an act of power, with many tweets mentioning the strength and courage that come with sharing stories and the impact it has on others. Tweets expressed pride in the strength and resilience of users who could share their stories, offering support and expressing gratitude for their bravery: “I might spend my entire day handing out favorites to #YesAllWomen tweets. Thanks for being brave, ladies, on behalf of my daughters;” “I love to hear stories of survival and strength;” “so proud of the ladies sharing their stories on #YesAllWomen. Thank you. Thank you. Your courage to speak truth gives all of us strength.”

Even instances of women being shamed or harassed for telling their stories often led to further connection from other women and some men: “#YesAllWomen is inspirational in many ways. In particular, the way in which women are responding to hate with strength, dignity, & reason.”

Some tweets also showed support from men who joined the movement, creating a sense that men can be allies in this fight for change: “I joined the #YesAllWomen tweets because I believe in solidarity with BOTH men and women, not just woman feminists;” “As a man, the best way to contribute to #YesAllWomen is to simply educate yourself by reading all the things that are bravely being shared;” “Yes, all men can learn something by reading the #YesAllWomen hashtag and taking the perspectives & stories & painful realities to heart.” Similarly, some tweets by women highlighted existing support from men: “The men who get this, are vocalizing support, are upset with us: you’re making me love men even more than I did, which was a lot #YesAllWomen;” “Men tweeps who have suggested other men come read #YesAllWomen & process w/o commenting. THOSE are #notallmen. They support women.”

Even some of those who adamantly insisted that #NotAllMen were implicated seemed to understand that #YesAllWomen remains important. Overall, tweets portrayed support for survivors of VAW and created a space for inspiration and hopefulness.

Hashtag activism

The majority of #YesAllWomen tweets emphasized the need for social change to address the daily experiences and issues faced by women. The primary code within this theme was hashtag activism (n = 493). Users expressed anger, disappointment, and frustration with the lack of societal progress and the insufficient consequences for VAW. Many individuals, including both women and men, acknowledged the serious nature of women’s daily experiences and highlighted the need for change. In this way, #YesAllWomen, as a social media campaign, served as a tool for creating awareness and promoting social change: “Yes, hashtags can be trivial and annoying—but discussion threads like #YesAllWomen can also be powerful;” “Everyone in the entire world needs to read [tweets that use] #YesAllWomen. It’s powerful and moving in ways I have never imagined.” Tweets revealed a desire to do more to support the movement, with many indicating that tweeting and retweeting alone were not enough: “Wish I had something more useful to add to #yesallwomen, other than to say I support the women sharing their experiences.” Of note, some male users joined in the activist tone, offering how #YesAllWomen can be used as an education tool: “I try my best to be a good dude. You should check out the #YesAllWomen tweets and try to be better too.”

Ultimately, #YesAllWomen resulted in a call for action, including the need for actions outside of social media, to support the movement. Users called on friends and family to share the tweets, and specifically on men to become more educated and aware of the daily realities women face: “This thread is required reading for males. Let the words/stories sink in, ask yourself, how can I be better tmrw [tomorrow]? #respect #YesAllWomen;” “Men, read #YesAllWomen. Ignore your defensiveness. Take a breath. Keep reading. Let it wake you up. Look for the log in your eye.”

Discussion

Hashtag feminism has become a way for people from different geographical locations to come together to discuss important, pressing feminist issues. The results of this study highlight several important points regarding hashtag feminism related to VAW. #YesAllWomen provides a space for activism and exploring shared experiences, yet trepidation remains, and backlash continues to proliferate hate and fear. Hashtags can be used to create positive change, but the limitations of this method must be acknowledged and addressed.

Hashtags on social media platforms allow individuals to join a larger conversation quickly and easily, and they provide a platform for all voices to be heard. #YesAllWomen has given women the opportunity to speak out about the ways in which misogynistic behavior has affected them, and to share their stories with a wider audience. Our results show that the majority of #YesAllWomen tweets aim to raise awareness and bring attention to the experiences and issues faced by women, with many tweets calling for an end to harmful behaviors and advocating for change in the treatment of women. In this way, the hashtag has been used as a tool for creating awareness and promoting social change, boosting its potential for positive impact as the number of users interacting with the hashtag increased (Ghobadi and Clegg, 2015). Indeed, many mainstream media outlets, including the New Yorker (Weiss, 2014), the New York Times (Medina, 2014), Time Magazine (Bridges, 2014; Feeney, 2014), NPR (Martin, 2014), and MSNBC (Carmon, 2014; Metzl, 2014), have reported on the hashtag, further expanding its reach and impact beyond social media (Jackson et al. 2019).

Other feminist hashtags have also shown significant impact through hashtag activism. For example, the well-known #MeToo movement invoked a hashtag that supported survivors of sexual harassment and abuse and called perpetrators to account (Murphy, 2019). #MeToo has successfully contributed to the ongoing conversation around believing and supporting women survivors of sexual assault, and it has helped prompt broader society to act, specifically in relation to sexual misconduct allegations (Brown, 2018; Kaufman et al. 2021; Vogelstein and Stone, 2021). Action using this hashtag even reached the level of bringing attention to Justice Brett Kavanaugh during his nomination to the U.S. Supreme Court (Dejmanee et al. 2020). Similar in-depth exploration of on-the-ground impact would improve understanding of how #YesAllWomen, and hashtag feminism in general, can create positive change, culturally and systemically. With a greater understanding of how hashtag feminism can impact social and political realities, online activists can become more intentional with outreach and goals to reach critical mass for persistent and sustainable change. This study found that #YesAllWomen provided a platform for people to share their experiences, including instances of victimization, violence, and sexual assault. Despite the difficulty faced in sharing such experiences, many people bravely came forward, inspiring others and promoting unity and support among users. Our study indicated that overall, users supported the message behind #YesAllWomen. As expected, though, some tweets criticized women, calling them attention-seekers or doubting the validity of their stories, which in turn led to increased support for survivors from other users. Other scholars exploring feminist hashtags found that #YesAllWomen successfully created a space for solidarity and “rhetorical kinship among women” (Jackson et al. 2019), a space for sharing stories and offering support (Phillips and Milner, 2017).

#MeToo created a similar space for survivors of sexual assault and was used to share stories and find strength and solace in solidarity with other survivors and supportive users (Murphy, 2019). Other hashtags have found similar results, such as #BeenRapedNeverReported (Keller et al. 2018; Mendes et al. 2018), #MahsaAmini (Kermani, 2023), #WhyIStayed (Linabary et al. 2020), among others (Kermani and Hooman, 2022). The experience of sharing stories publicly and becoming part of an online community may create a sense of unity, with different genders coming together to support a cause. In addition, by engaging with stories of other survivors, women in this study using #YesAllWomen reported feeling less alone in their experiences and finding validation in the shared struggles of others. Across multiple feminist hashtags, users have expressed similar sentiments around feelings of validation and inspiration due to their experience with the hashtag (Linabary et al. 2020) and the online community. Indeed, Beverly Gooden explained why she created the hashtag #WhyIStayed: “I believe in storytelling. I believe in the power of shared experience. I believe that we find strength in community” (Gooden, 2014). From a feminist standpoint, in addition to offering a safe place for sharing and community building, these online spaces contribute to a more in-depth understanding of knowledge on the experience of women by offering and encouraging storytelling and learning from each other. Through hashtag feminism, women are invited to ask and answer their own questions about their experiences, relationships, culture, and systems. Through this process of questioning and answering through stories, knowledge is created and shared.

#YesAllWomen has not been without its detractors. Our analysis found criticism of the hashtag as being divisive or overly negative, and some users engaged in abuse and harassment against those who supported it. This backlash emphasizes the reality that some users expressed regarding their personal feelings of trepidation around sharing their stories. Other #YesAllWomen commentaries have found similarly disturbing results related to the potential for serious backlash against users, including instances of sexual assault in response to using #YesAllWomen (Jackson and Banaszczyk, 2016; Rodino-Colocino, 2014). This kind of backlash reiterates the purpose of the hashtag itself, proving the ubiquitous experience of VAW and the need for broad social change.

Popular feminist hashtags have consistently seen this kind of backlash or trolling within the hashtags themselves, often through victim-blaming, and by using other, sometimes mirror hashtags to promote misogynistic ideas, such as #HimToo (Dejmanee et al. 2020; Linabary et al. 2020; Mendes et al. 2018). Backlash against women’s movements is not new (Faludi, 1991), but the ease with which a backlash can spread through social media is concerning. This serves as a reminder that social media activism is not always a safe and welcoming space, and that those who engage with it may be subject to abuse and backlash (Cole, 2015). Likewise, the fear of retribution expressed within these tweets is not a new finding. Previous evidence from across the globe underscores the impact that fear of retaliation from violent perpetrators has on underreporting abuse outside of social media as well (Heron et al. 2022; Heron and Eisma, 2021; Planty et al. 2013; Shaheen et al. 2020; Silva et al. 2022). The reach of this fear from daily life through social media further emphasizes the point of #YesAllWomen and clearly indicates a need for immediate change.

Our results did not indicate clear, tangible change within the political realm from #YesAllWomen, but the insight and support expressed by users indicated social growth in terms of knowledge, understanding, and motivation. Continued change—both social and political—can occur through expanding the impact of movements such as #YesAllWomen and other feminist online activist spaces to continue building knowledge on VAW and other forms of abuse of power. Through the use of storytelling, alongside the respect and appreciation for the value within both individual and collective experience, communities can be energized and mobilized for positive change. Overall, our results show that #YesAllWomen successfully provided a space for people to explore their own and others’ experiences of misogyny, persistent fear, and VAW. The hashtag encouraged users to come together, learn from each other, and provide support to the entire community. Although some users expressed hate and ultimately abused the space, the general experience was filled with hope, gratitude, and inspiration. The long-term impact of #YesAllWomen and other feminist hashtags needs to be explored to better understand the power of this expanding movement.

Limitations

There are some limitations to this study, the first being the characteristics of those who tweeted. Those who posted on Twitter using the hashtag may be more motivated to engage in activism, even with trepidations, than others who are not posting. Furthermore, only users whose tweets were public were those from whom we extracted the analyzed text. We sampled English tweets only, forcing us to exclude countries and users where English is not commonly used. We also cannot specify the demographic information of those who tweeted using the hashtag (although gender was noted by users in some tweets and was reported in the results where relevant), and geotagging was only available for some tweets. Finally, there may be cultural and socio-economic differences in access and use of this U.S.-based social media platform.

A second set of potential limitations considers the extracted dataset. In the Twitter data that we used to train our topic models, users expressed their thoughts or shared sentiments on the #YesAllWomen hashtag in short-form texts. This makes it challenging for the topic model to detect a theme among a group of documents, as “short” texts may lack contextual information of the conversation, limiting the scope of analysis when sorting through such large amounts of data. There are also some limitations of the NLP pipeline. During the processing of the raw data, we removed abbreviations, slang, and emojis (“cleaning the data”) as a traditional practice. However, on Twitter, where people deliver their thoughts with comparatively short posts, the removed data may have provided more context to the topic model. We had to choose a balance between keeping noisy irrelevant text as opposed to potentially removing relevant data.

A final technical limitation from topic modeling is overfitting. As illustrated in our study’s method, we ran a set of hyperparameter tuning where we precisely configured topic models to improve the quality of topics based on our manual annotations. This often results in overfitting of the model on the data, making the results hard to generalize to other datasets.

Conclusions

#YesAllWomen has served as a powerful social media activism tool, bringing attention to the fears regularly faced by women. It has provided a platform for sharing experiences and has offered insight, community, and support for users. Despite the challenges that social media activism can pose, including backlash, #YesAllWomen has been a source of inspiration for many. Users found the courage to speak out about their experiences and to advocate for social change. The hashtag has served as a call to action, encouraging others to join the conversation and the broader effort towards creating a safer, inclusive and equitable society.

That being said, #YesAllWomen, alongside other hashtags focused on VAW, provides an enormous narrative of global abuse. Hashtag activism is playing an important role in increasing awareness and creating solidarity among users, but the impact must reach outside of the social media space and into real change within relationships, communities, and governments. Our work feeds into that goal and focuses on analyzing the topics discussed within these hashtags to better understand the sentiment and related concerns to #YesAllWomen.