Linking Collective Attention Across Platforms: Do More Tweets Beget More Video Views?
Artist: Janusz Jurek
Linking Collective Attention Across Platforms: Do More Tweets Beget More Video Views?
Ertugrul, A, Wu, S, Lee, J, Lin, Y & Xie, L 2020, ‘Linking Collective Attention Across Platforms: Do More Tweets Beget More Video Views?’, Under Review at ACM CSCW 2021.
We present a quantitative study that links collective attention across two social media platforms -- YouTube and Twitter, around videos of controversial political topics. While online political discussions have been the subject of many studies, most obtain data and observations from a single platform. This exposes an open question: will attention garnered in one online space transcend to another, and to attitudes and behaviours offline? In this work, we develop procedures to collect and curate data that link Twitter posts and topical YouTube videos, yielding three new cross-platform datasets covering the controversial topics Abortion, Gun control, and Black Lives Matter over 16 months. We propose a set of metrics to measure and compare the activities and attention by different political groups. We find that left-leaning videos are more viewed, more engaging, but less tweeted than right-leaning videos. Upon examining the early adopter groups of each video, we find those for right-leaning videos are discentralised, while those for the left-leaning videos have hub-and-spoke structures. Furthermore, attention unfolds quickly on left-leaning videos, but spans a longer period of time for right-leaning videos. We set up regression tasks to correlate cross-platform metrics to a set of offline data such as public support, abortion rate, gun ownership, fatality, and protests. We observe consistent and significant improvement over baselines without social media data. Our findings in the attention dynamics of tweeted videos contrast current understandings of political polarisation within one platform, in that more coordinated interactions in right-leaning groups does not translate to more attention and engagement with videos. This work opens many new questions about the nature and mechanism of influence across social platforms.