How Collective Narcissism Spread: Evidence from Twitter on Capital Insurrection

• In Event: The Big Lie on Social Media

Abstract

Theory and Research Question
Collective narcissism describes an ingroup identification tied to an emotional investment in unrealistic belief about an ingroup’s exaggerated achievement or status (Golec de Zavala et al., 2009). Literature has found that collective narcissism is a major predictor of support for Trump’s presidential candidacy (Federico & Golec de Zavala, 2018). Collective narcissism has also been found foment intergroup hostility, beliefs in conspiracy theories about outgroups, and political violence (Cichoka & Cislak, 2020), which threaten the bedrock of democracy.
Collective narcissism has been studied as a psychological trait and measured as scales to examine its relationship with beliefs and behaviors in survey and experiment settings. However, existing scholarship has overlooked the role of collective narcissism in inter-group communication, where this emotionally invested ingroup identification might be highly contagious (Chen et al., 2022) and can thus mobilize ingroup members towards collective expression and actions. Grounded in theories of social identity and protest paradigm, this study examines how collective narcissism is expressed on social media in the case of the #StopTheSteal movement and how users interact with these intergroup expressions.
Social network sites are heaven for narcissists because they invite users to post self-centered content and broadcast their content directly to weak ties or even strangers who share similar group identities, which provide a unique ground for cultivating collective narcissism. Users can actively seek or get fed by algorithms with like-minded ingroup members, reinforcing the ingroup achievement and the unfair treatment they received. Collective narcissism is hence supported, spread, and reinforced within social network sites. This study, thus, develops the following research questions and hypotheses:
RQ1: Who are the “in-groups” engaging in collective narcissistic expression during the #StopTheSteal movement?
RQ2: Among those tweets that use collective narcissism, a) which “in-groups” support election fraud claims? b) Which “in-groups” support the insurrection?
H1: Tweets that use collective narcissism are likely to receive more retweets, comments, and likes.
H2: Tweets that use collective narcissism are likely to receive more replies that also use collective narcissistic expressions.
Data and Measurement
Data description. 110,557 English posts including hashtags (#StopTheSteal and #Electionfraud) that were posted from January 3rd – 7th, 2021 was retrieved by using the Twitter academic API, together with account information and user engagement metrics of a tweet. We also collected all the replies to the 21,800 original posts. 10,228 unique conversation IDs are being retrieved. In this conversation dataset, the text of reply, created time, author ID, in reply to user ID, and engagement metrics are recorded.
Measurement. A dictionary of collective narcissism will be built based on its conceptualization along three dimensions: 1) the salience of group identities (e.g., Republicans, MAGA), 2) exaggerated ingroup achievement or status, 3) the achievement or status not recognized by the outgroups. Each tweet/reply will be coded by integrating manual and computer-assisted content analysis with iterations and cross-validation.
For Support for election fraud and insurrection, we will recruit M-Turk workers for coding a random sample of 5000 tweets. Turkers will be asked whether each tweet support election fraud and insurrection separately. Next, we will train a machine learning classifier using the BERT model to classify the issue stance of the rest posts and replies.
User engagement of tweets are quantified using three indicators: number of likes, replies, and retweets. For a unique original post, the number of retweets will be quantified by the sum of the number of retweets at all stages of the diffusion chain.
Data analysis. Descriptive analyses will be performed to answer which group identities appeared in the election fraud conversation on social media with collective narcissistic expression (RQ1) and which groups supported election fraud claims and insurrection correspondingly (RQ2). To examine H1, we will conduct the negative binomial regression to predict user engagement metrics by the frequency of collective narcissistic expression in a tweet. To test H2, we will regress the frequency of using collective narcissism expressions in replies on the frequency of using collective narcissistic expressions in the original post.
Contribution of the paper. This paper examines collective narcissism and intergroup interaction in an ecologically valid environment: social media, providing empirical evidence of the relationship between collective narcissism, belief in conspiracy, and political violence. We demonstrate how collective narcissistic expression is contagious in the social network and provokes anti-democratic beliefs and behaviors.

Authors

• Liwei Shen

• Kaiping Chen, University of Wisconsin-Madison