Uncovering Gender Bias within Journalist-Politician Interaction in Indian Twitter: Abstract & Intro

This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. Authors: (1) Brisha Jain, Independent researcher India and brishajain02@gmail.com; (2) Mainack Mondal, IIT Kharagpur India and mainack@cse.iitkgp.ac.in. Table of Links Abstract and Intro Related Work Data Collection Analysis Methodology Results Limitations Concluding Discussion References ABSTRACT Gender bias in political discourse is a significant problem in today’s social media. Previous studies found that the gender of politicians indeed influences the content directed towards them by the general public. However, these works are particularly focused on the global north, which represents individualistic culture. Furthermore, they did not address whether there is gender bias even within the interaction between popular journalists and politicians in the global south. These understudied journalist-politician interactions are important (more so in collectivistic cultures like the global south) as they can significantly affect public sentiment and help set gender-biased social norms. In this work, using large-scale data from Indian Twitter we address this research gap. Specifically, we curated a gender-balanced set of the 100 most-followed Indian journalists on Twitter and the 100 most-followed politicians. Then we collected 21,188 unique tweets posted by these journalists that mentioned these politicians. Our analysis revealed that there is a significant gender bias—the frequency with which journalists mention male politicians vs. how frequently they mention female politicians is statistically significantly different (푝 << 0.05). In fact, median tweets from female journalists mentioning female politicians received ten times fewer likes than median tweets from female journalists mentioning male politicians. However, when we analyzed tweet content, our emotion score analysis and topic modeling analysis did not reveal any significant difference between the journalists’ tweets mentioning male politicians and those mentioning female politicians. Finally, we found a potential reason for the significant gender bias: the number of popular male Indian politicians is almost twice as large as the number of popular female Indian politicians, which might have resulted in the male-biased popularity of tweets (and even the frequency of receiving tweets). We conclude by discussing the implications of this work for the need for gender diversity in the political discourse of Indian social media and the future development of recommender systems for social media that can address this need. 1. INTRODUCTION “Since Smiriti Irani is displaying her acting talent in the emotional /angry category, she may ‘forget ’ she isn’t on set & break into a dance”[1] The tweet above, posted by a male spokesperson of a national political party against a female cabinet minister of India offers a glimpse into the endemic nature of gender bias in political conversations on Indian social media sites like Twitter [2]. In fact, according to an Amnesty study, one in seven tweets directed towards female politicians in India are abusive in nature [22]. Recent studies have examined gender bias in political discourse on social media [1, 2, 18]. These studies have shown that the gender of politicians does not influence their own social media posts, but their gender does influence the content directed towards them. However, these studies have two serious shortcomings. First, many of these studies exclusively looked into global north and discounted global south. However, this is particularly concerning on several counts. One, Southeast Asia has roughly 527 million active social media users, larger than any other contiguous geography. Of these, roughly 470 million are in India. Also, according to political science literature, southeast Asia has not yet matched the Western Hemisphere in gender equality. Therefore, gender bias in political discourse on social media has not yet been studied for a large group of users active in a less progressive social setting. In this study, we focus on the interactions with politicians from a large country in the global south—India. In India the fact that female politicians face handicaps in their ability to participate in the democratic process has been widely recognised. In India, legislators have sought to address it by passing the Women’s Reservation bill which allocates 33% of representation in central and state legislative bodies for women. While this bill will provide representation to women, it remains unclear if it will provide women with the voice to initiate and shape political discourse in India. This is because gender bias in social media interactions in less progressive democracies like India is problematic at several levels. It dissuades female politicians from being able to harness the power of social media for political purposes and therefore puts them at a disadvantage compared to their male peers. This can undermine their effectiveness and career prospects. It also marginalizes their views from the political discourse on social media thereby reducing the diversity of opinions available to neutral observers. In our work, we aim to uncover the potential of gender bias in the specific context of digital conversations towards Indian politicians. Second, previous studies often focused on the interactions between politicians and the general public. Although this understanding is quite valuable, we noted that there is not much work on specific interaction between politicians and more influential social media users. These interactions are important as they can significantly affect the public sentiment. In fact, recent work by Shekhawat et al. has shown that use of Twitter by Indian politicians to engage with newspapers and influencers via social media has been growing [23]. The fact that politicians have been keen to engage with influencers seems to suggest that these influencers play a significant role in shaping political discourse on Twitter in India. It is therefore important to examine if even these journalist-politician interactions are suffering from gender bias. However, no previous study has checked if even the interaction between the well-known journalists and popular politicians hints at existing gender bias. RQ1: Is there gender bias in interaction frequency and popularity of journalist-politician interactions? RQ2: Is there gender bias within the content of journalist-politician tweets? The study investigates these questions by curating a genderbalanced dataset of Twitter accounts (using programmatic data collection from Twitter and gender detection) comprising hundred popular (by number of Twitter followers) Indian politicians and hundred popular Indian journalists. Then this study programmatically collected all the tweets posted by these journalists’ accounts mentioning the accounts of popular politicians in our dataset. we divided our collected tweets four categories according to the gender of the senders/receivers—Male journalist’s tweets mentioning Male Politicians (MJ-MP), Female journalist’s tweets mentioning Male Politicians (FJ-MP), Male journalist’s tweets mentioning Female Politicians (MJ-FP) and Female journalist’s mentioning Female Politicians (FJ-FP). In total we collected 21,188 unique tweets across these four categories. Our analysis revealed a significant gender bias—there are statistically significant differences (p << 0.05) in how frequently Male/Female journalists mention Male politicians vs. how frequently they mention Female politicians. Moreover, we found that the Tweets mentioning Male politicians are more popular. In fact, across all metrics (retweets, likes, replies) the median popularity of the posts mentioning female politicians are consistently lower than posts mentioning male politicians. E.g., the median number of likes received by a post in FJ-FP (i.e., posted by female journalist mentioning female politician) is only 35, whereas the median number of likes received by a post in FJ-MP (i.e., posted by female journalist mentioning male politician) is 398—more than ten times higher. Note that these likes are given by the general public. However, interestingly, the content analysis revealed no bias— the content of actual tweets posted by journalists towards these politicians is not gender-biased in terms of either emotion scores or the topic of tweets. To that end, we identified a more fundamental reason for the low popularity of tweets towards female politicians—the inherent gender bias in Indian Twitter (and perhaps reflective of Indian society), where popular male politicians (by number of Twitter followers) are almost twice as many compared to popular female politicians. We conclude this paper, by discussing the limitations of our study and identifying how our findings hints towards the necessity of countering gender bias. We surmise that our work will help Indian social media platform developers towards making systematic algorithmic changes within their recommender systems for countering the ill effects of gender bias observed in this work. In the rest of the paper we first present the related work in Section 2, describe our data collection strategy (Section 3) and analysis methodology (Section 4). Then we present our results and examine the research questions in Section 5. Then we present the limitations of our study(Section 6) and conclude (Section 7). [1] Smriti Irani is a female Indian politician. The original tweet mentioning her is available at https://twitter.com/tehseenp/status/702491795079364609 [2] At the time of performing this study in 2023, the platform was called Twitter rather than x.com. So, we will call our platform Twitter in this paper This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. Authors: (1) Brisha Jain, Independent researcher India and brishajain02@gmail.com; (2) Mainack Mondal, IIT Kharagpur India and mainack@cse.iitkgp.ac.in. This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. Authors: Authors: (1) Brisha Jain, Independent researcher India and brishajain02@gmail.com; (2) Mainack Mondal, IIT Kharagpur India and mainack@cse.iitkgp.ac.in. Table of Links Abstract and Intro Related Work Data Collection Analysis Methodology Results Limitations Concluding Discussion References Abstract and Intro Abstract and Intro Related Work Related Work Data Collection Data Collection Analysis Methodology Analysis Methodology Results Results Limitations Limitations Concluding Discussion Concluding Discussion References References ABSTRACT Gender bias in political discourse is a significant problem in today’s social media. Previous studies found that the gender of politicians indeed influences the content directed towards them by the general public. However, these works are particularly focused on the global north, which represents individualistic culture. Furthermore, they did not address whether there is gender bias even within the interaction between popular journalists and politicians in the global south. These understudied journalist-politician interactions are important (more so in collectivistic cultures like the global south) as they can significantly affect public sentiment and help set gender-biased social norms. In this work, using large-scale data from Indian Twitter we address this research gap. Specifically, we curated a gender-balanced set of the 100 most-followed Indian journalists on Twitter and the 100 most-followed politicians. Then we collected 21,188 unique tweets posted by these journalists that mentioned these politicians. Our analysis revealed that there is a significant gender bias—the frequency with which journalists mention male politicians vs. how frequently they mention female politicians is statistically significantly different (푝 << 0.05). In fact, median tweets from female journalists mentioning female politicians received ten times fewer likes than median tweets from female journalists mentioning male politicians. However, when we analyzed tweet content, our emotion score analysis and topic modeling analysis did not reveal any significant difference between the journalists’ tweets mentioning male politicians and those mentioning female politicians. Finally, we found a potential reason for the significant gender bias: the number of popular male Indian politicians is almost twice as large as the number of popular female Indian politicians, which might have resulted in the male-biased popularity of tweets (and even the frequency of receiving tweets). We conclude by discussing the implications of this work for the need for gender diversity in the political discourse of Indian social media and the future development of recommender systems for social media that can address this need. 1. INTRODUCTION “Since Smiriti Irani is displaying her acting talent in the emotional /angry category, she may ‘forget ’ she isn’t on set & break into a dance” [1] “Since Smiriti Irani is displaying her acting talent in the emotional /angry category, she may ‘forget ’ she isn’t on set & break into a dance” The tweet above, posted by a male spokesperson of a national political party against a female cabinet minister of India offers a glimpse into the endemic nature of gender bias in political conversations on Indian social media sites like Twitter [2]. In fact, according to an Amnesty study, one in seven tweets directed towards female politicians in India are abusive in nature [22]. Recent studies have examined gender bias in political discourse on social media [1, 2, 18]. These studies have shown that the gender of politicians does not influence their own social media posts, but their gender does influence the content directed towards them. However, these studies have two serious shortcomings. First, many of these studies exclusively looked into global north and discounted global south. However, this is particularly concerning on several counts. One, Southeast Asia has roughly 527 million active social media users, larger than any other contiguous geography. Of these, roughly 470 million are in India. Also, according to political science literature, southeast Asia has not yet matched the Western Hemisphere in gender equality. Therefore, gender bias in political discourse on social media has not yet been studied for a large group of users active in a less progressive social setting. In this study, we focus on the interactions with politicians from a large country in the global south—India. In India the fact that female politicians face handicaps in their ability to participate in the democratic process has been widely recognised. In India, legislators have sought to address it by passing the Women’s Reservation bill which allocates 33% of representation in central and state legislative bodies for women. While this bill will provide representation to women, it remains unclear if it will provide women with the voice to initiate and shape political discourse in India. This is because gender bias in social media interactions in less progressive democracies like India is problematic at several levels. It dissuades female politicians from being able to harness the power of social media for political purposes and therefore puts them at a disadvantage compared to their male peers. This can undermine their effectiveness and career prospects. It also marginalizes their views from the political discourse on social media thereby reducing the diversity of opinions available to neutral observers. In our work, we aim to uncover the potential of gender bias in the specific context of digital conversations towards Indian politicians. Second, previous studies often focused on the interactions between politicians and the general public. Although this understanding is quite valuable, we noted that there is not much work on specific interaction between politicians and more influential social media users. These interactions are important as they can significantly affect the public sentiment. In fact, recent work by Shekhawat et al. has shown that use of Twitter by Indian politicians to engage with newspapers and influencers via social media has been growing [23]. The fact that politicians have been keen to engage with influencers seems to suggest that these influencers play a significant role in shaping political discourse on Twitter in India. It is therefore important to examine if even these journalist-politician interactions are suffering from gender bias. However, no previous study has checked if even the interaction between the well-known journalists and popular politicians hints at existing gender bias. RQ1 : Is there gender bias in interaction frequency and popularity of journalist-politician interactions? RQ1 RQ2 : Is there gender bias within the content of journalist-politician tweets? RQ2 The study investigates these questions by curating a genderbalanced dataset of Twitter accounts (using programmatic data collection from Twitter and gender detection) comprising hundred popular (by number of Twitter followers) Indian politicians and hundred popular Indian journalists. Then this study programmatically collected all the tweets posted by these journalists’ accounts mentioning the accounts of popular politicians in our dataset. we divided our collected tweets four categories according to the gender of the senders/receivers—Male journalist’s tweets mentioning Male Politicians ( MJ-MP ), Female journalist’s tweets mentioning Male Politicians ( FJ-MP ), Male journalist’s tweets mentioning Female Politicians ( MJ-FP ) and Female journalist’s mentioning Female Politicians ( FJ-FP ). In total we collected 21,188 unique tweets across these four categories. MJ-MP FJ-MP MJ-FP FJ-FP Our analysis revealed a significant gender bias—there are statistically significant differences ( p << 0.05) in how frequently Male/Female journalists mention Male politicians vs. how frequently they mention Female politicians. Moreover, we found that the Tweets mentioning Male politicians are more popular. In fact, across all metrics (retweets, likes, replies) the median popularity of the posts mentioning female politicians are consistently lower than posts mentioning male politicians. E.g., the median number of likes received by a post in FJ-FP (i.e., posted by female journalist mentioning female politician) is only 35, whereas the median number of likes received by a post in FJ-MP (i.e., posted by female journalist mentioning male politician) is 398—more than ten times higher. Note that these likes are given by the general public. p However, interestingly, the content analysis revealed no bias— the content of actual tweets posted by journalists towards these politicians is not gender-biased in terms of either emotion scores or the topic of tweets. To that end, we identified a more fundamental reason for the low popularity of tweets towards female politicians—the inherent gender bias in Indian Twitter (and perhaps reflective of Indian society), where popular male politicians (by number of Twitter followers) are almost twice as many compared to popular female politicians. We conclude this paper, by discussing the limitations of our study and identifying how our findings hints towards the necessity of countering gender bias. We surmise that our work will help Indian social media platform developers towards making systematic algorithmic changes within their recommender systems for countering the ill effects of gender bias observed in this work. In the rest of the paper we first present the related work in Section 2, describe our data collection strategy (Section 3) and analysis methodology (Section 4). Then we present our results and examine the research questions in Section 5. Then we present the limitations of our study(Section 6) and conclude (Section 7). [1] Smriti Irani is a female Indian politician. The original tweet mentioning her is available at https://twitter.com/tehseenp/status/702491795079364609 [2] At the time of performing this study in 2023, the platform was called Twitter rather than x.com. So, we will call our platform Twitter in this paper