Scientific Studies: Tweets Show No Correlation with Later Citations

By January 22, 2014
publications scientifiques

Although it has been suggested that Twitter has an impact on the life cycle of scientific publications, the number of citations a paper later receives – which is the sacred touchstone of academic success – does not appear to correlate with the number of tweets, as social buzz is generated essentially by and for a non-specialist audience.

Although a recent study purported to show that Twitter can influence the life cycle of scientific publications, a second study by Canadian researchers at the University of Montreal has concluded that there is only a very weak correlation between the indicators from social media, especially Twitter, and traditional academic citations. While the impact differs somewhat depending on the scientific review in which an article is published, the method developed by the research team demonstrates to what extent the number of tweets proves to be a valid measure when assessing the real impact of the academic work. Results show that the scientific quality of a paper had no direct effect on whether or not it was tweeted. “The top-tweeted articles are a mix of curious stories, funny articles or those with medical implications,” pointed out the lead author of the study. So it would appear that while the impact of the tweets on the general public can be assessed, this does not hold for the scientific community.

Scientific community ‘not very connected’

The Canadian researchers first took a set of 1.4 million documents relating to biomedical science covered by the top ‘citation collectors’  PubMed and Web of Science between 2010 and 2012. The first finding that emerged was that very few articles from among these well-cited publications had been brought to the attention of users of the microblogging site. Fewer than 10% of these articles had been tweeted even once. Nevertheless, the use of the network is changing fast. By 2012, 20% of all biomedical articles were being tweeted at least once.  However, as less than 2.5% of the scientific community is active on Twitter, compared with 8.7% of the total North American population, retweets are few and correlation with later citations very weak indeed. Moreover, tweets from accounts which are regularly fed with information do not see comparatively more retweets on the network (0.2 tweets per article) than the tweets sent from sporadically-used accounts (between 0.2 and 0.3 tweets per article). The study therefore concludes that there is no observable correlation between the use of Twitter and formal citations for scientific publications.

Topical or funny stories generate most buzz

While scientists’ low use of Twitter and the network’s low reputation as a tool for scientific communication seem to mean that tweets do not lead to citations further down the line, the Canadian researchers found that most Twitter buzz was generated not so much by the academic quality of the paper as curious and even funny items of information picked up and retweeted by the general public. In other words, the number of tweets circulating does not seem to have anything to do with the intellectual contribution or scientific quality of the paper.  However, apart from humorous side, another factor found to generate tweets was the connection between the scientific paper and current health issues. All in all therefore, it seems that you first need to assess the audience for a particular tweet topic before drawing any conclusions about the likely impact of a piece of scientific research on future work or academic progress.

Legal mentions © L’Atelier BNP Paribas