Tweets that reference or share academic articles are often "mechanical and devoid of original thought"
Researchers analysed 8000 tweets referencing articles published in dental journals. The tweets were posted by 2000 accounts and referenced 4000 papers.
When examining the tweets that referenced the ten most tweeted papers they discovered that:
- Counts would reduce by 56% if accounts tweeting rather than actual tweets were counted
- If unique, non-duplicate tweets were counted, the number would reduce by another 50%, reducing counts to 22% of their original number
- Examining tweeting accounts established that 77% of tweets were mechanical in nature
- 23% of tweets showed a more human, creative character though most such accounts tweeted about dental papers only a few times.
- 6% of accounts were human in character and sustained their tweeting about dental papers.
Analysts should note that tweet counts alone cannot measure engagement with academic literature. At a minimum, analysts should identify and block bots and fake followers. They would also do well to count truly unique text variants in referencing tweets. They would then do well to identify the tweets and accounts that are truly informative.
Librarians use such metrics to identify recommended readings. Almetric.com’s famous doughnut displaying the altmetric score of papers, appearing on ever more journal websites, is largely driven by Twitter counts.
"Simplistic and naïve use of social media data risks damaging the scientific enterprise, misleading both authors and consumers of scientific literature".
The article The unbearable emptiness of tweeting—About journal articles is published on PLOS One.
Original source: Inside Higher Education;