Who creates trends in online social media: The crowd or opinion leaders?

Leihan Zhang, Jichang Zhao, Ke XuJ. Comput. Mediat. Commun 2014 · OpenReview

TL;DR

This paper investigates the drivers behind the formation of trends on social media platforms like Weibo. Contrary to the prevailing "influentials hypothesis," the study reveals that the collective participation of ordinary users, rather than just opinion leaders, is crucial for a trend to gain widespread traction and become a lasting phenomenon.

What Problem Does This Paper Address?

Online social media platforms have become massive conduits for information, leading to the rapid emergence of trends, from viral slang to news events. A long-standing question in this domain is who is primarily responsible for initiating and propagating these trends: a select group of influential individuals (opinion leaders) or the broader user base (the crowd). Existing research often focuses on the role of influentials, assuming their outsized impact. However, this paper questions this assumption by examining the temporal dynamics of trend formation and the distinct roles of different user groups. The authors aim to uncover the underlying mechanisms that lead to the widespread adoption of new ideas and language on social media.

Methods

The study employs a quantitative approach to analyze the diffusion of Internet slang words on Weibo. They identify two distinct peaks in the popularity of these slang words over their lifecycle. These peaks are interpreted as representing two different diffusion events: a smaller-scale, less successful diffusion (p1) and a larger-scale, more successful diffusion (p2).

To understand the drivers of these diffusions, the researchers analyze the characteristics of users involved in each stage, focusing on:

  • User Influence: Measured by the number of followers (#Followers). Users with many followers are considered opinion leaders, while those with few are considered part of the crowd.
  • User Verification Status: Differentiating between verified (often public figures or organizations) and non-verified users.
  • Retweet Activity: Analyzing the number of times a user retweets content, distinguishing between one-time and repeat retweeters.
  • Geographical Distribution: Examining the spatial spread of adopters.

By comparing the composition of early adopters during the p1 and p2 phases, the study aims to pinpoint the factors that differentiate a fleeting trend from a widespread phenomenon.

Datasets

The study utilizes a large dataset of tweets collected from Weibo throughout the year 2013. This dataset includes approximately 173 million tweets and over 9 million unique users after spam filtering. The specific Internet slang words analyzed (42 in total) were identified from the Chinese encyclopedia website baike.com.

Key Insights and Ideas

  • Two Peaks of Popularity: Internet slang words exhibit a surprising two-peak popularity curve, suggesting that a trend can have multiple opportunities to gain traction, with the first often failing to achieve widespread adoption.
  • The Crowd's Early Role is Crucial: The successful diffusion (p2) is characterized by a significantly higher proportion of ordinary users (low #Followers) participating in the early stages compared to the less successful diffusion (p1). This suggests that broad participation from the crowd is essential for a trend to go viral.
  • Opinion Leaders' Role is Limited: While opinion leaders (high #Followers) might initiate a diffusion, their dominance in the early stages of p1 leads to limited coverage. They are more effective at starting a local spread rather than a global trend.
  • Critical Mass of Ordinary Users: The study identifies a critical number of followers (around 232) for ordinary users. When a sufficient proportion of users below this threshold are involved early on, the diffusion is more likely to succeed. This number is noted to be close to Dunbar's number, suggesting a link to human cognitive limits in social interactions.
  • Verified Users are Less Critical for Mass Adoption: In the successful diffusion (p2), the proportion of verified users is lower than in the less successful diffusion (p1). This implies that ordinary, non-verified users play a more decisive role in attracting collective attention for a trend to become popular.
  • Repeat Engagement Matters: Users who engage in multiple retweets (two or more times) are more prevalent in the early stages of successful diffusions (p2), indicating that sustained engagement from the crowd amplifies the spread.
  • Uniform Geographical Spread: Successful trends (p2) show a more uniform geographical distribution of early adopters compared to less successful ones (p1), which tend to be concentrated in specific regions.

Key Results

  • The study found that in the early stages of diffusion, the crowd (users with < 200 followers) is significantly more represented in successful diffusions (p2) than in failed ones (p1).
  • A critical threshold of approximately 232 followers was identified for ordinary users, beyond which their participation significantly deviates between successful and failed diffusions.
  • The ratio of verified to non-verified users in early stages was approximately 3:7 for the small-scale diffusion (p1) and less than 2:8 for the large-scale diffusion (p2).
  • Information entropy for geographical distribution was higher for successful diffusions (p2) (4.46) compared to less successful ones (p1) (4.26) in the early adoption stage, indicating a more uniform spread.

Limitations and Future Work

The authors acknowledge that their study primarily focused on the occupation of the crowd and opinion leaders. They suggest that further research is needed to investigate the intrinsic factors that drive the crowd's participation in adopting slang words. These factors could include the nature of the slang word itself, external events, or even the users' living environment. Understanding these drivers would be crucial for designing effective strategies to attract crowd adoption in real-world scenarios. The study also notes that while opinion leaders are important for initiating a spread, their focus alone does not guarantee broad coverage.

Summary updated

Abstract

Trends in online social media always reflect the collective attention of a vast number of individuals across the network. For example, Internet slang words can be ubiquitous because of social memes and online contagions in an extremely short period. From Weibo, a Twitter-like service in China, we find that the adoption of popular Internet slang words experiences two peaks in its temporal evolution, in which the former is relatively much lower than the latter. This interesting phenomenon in fact provides a decent window to disclose essential factors that drive the massive diffusion underlying trends in online social media. Specifically, the in-depth comparison between diffusions represented by different peaks suggests that more attention from the crowd at early stage of the propagation produces large-scale coverage, while the dominant participation of opinion leaders at the early stage just leads to popularity of small scope. Our results quantificationally challenge the conventional hypothesis of influentials. And the implications of these novel findings for marketing practice and influence maximization in social networks are also discussed.
BibTeX
@article{zhang2014who,
  author = {Leihan Zhang and Jichang Zhao and Ke Xu},
  title = {{Who creates trends in online social media: The crowd or opinion leaders?}},
  journal = {J. Comput. Mediat. Commun.},
  year = {2014},
  url = {https://openreview.net/forum?id=Dbk3Kn-lffu},
}

Loading related content...

Figure