Tweets don’t bring citations, randomised controlled study finds

It’s generally thought that tweeting about a research paper will get it more citations – one traditional indicator of the scholarly impact of a scientific paper. Now a group of scientists with large followings on X, formerly Twitter, has conducted a three-year study which concludes that, while discussing science on social media has many benefits – and can be ‘a lot of fun’ – increasing a paper’s citations is probably not one of them.

The group of biological and environmental scientists – with followers ranging from 4000 to 48,000 on X – randomly selected five papers from a journal and tweeted about one, while keeping the other four as controls. They repeated this 10 times across 11 journals, recording Altmetric scores, a measure of how much attention a research article receives, the number of tweets, and citation counts before and after tweeting.

Tweeted articles were found to be downloaded 2.6–3.9 times more than controls straight after tweeting. They also gained significantly higher Altmetric scores and number of tweets three years after tweeting. However, while some tweeted papers were cited more than their control papers in the same journal, the team concludes that the overall increase in citation counts after three years was not statistically significant.

‘There have been very few previous controlled experiments looking at this; almost all previous papers on this topic did not conduct controlled experiments,’ says lead author Trevor Branch, a biological scientist at the University of Washington. ‘Ours is the longest running and the one that most closely mimics the actual impact of a scientist tweeting about a paper, instead of a large group of scientists, or a concerted effort across multiple forms of social media.’

Branch now leans towards the conclusion that if a paper is interesting, novel or timely enough, it will be cited anyway, regardless of tweeting about it. ‘Tweeting does not cause more citations. Instead, good papers both get cited more and tweeted about more.’

Nevertheless, Branch still believes social media is important for scientists and society in general. ‘But the platform used may change over time depending on whether it encourages truth and facts or is welcoming to misinformation,’ he adds. ‘I know a lot of scientists that have moved from X to Bluesky [an alternative microblogging social platform] recently.’

The paper concludes with the authors noting how they all have benefited from their foray into social media, for example by building an online community, establishing collaborations and publishing papers that otherwise would not have happened. However, they add that increasing the profile of their scientific papers was not their primary aim.

‘This is a very well-executed study, with a clear focus and clean methodology,’ says Vincent Traag, a specialist in bibliometrics at Leiden University in the Netherlands. ‘The authors found that a single tweet is unlikely to have a large effect on the number of citations. It is still possible that if multiple people start tweeting about a paper, it would increase citations more substantially. Nonetheless, the results of the study do suggest that whatever the size of the effect, it would be relatively modest.’

While there is a relationship between tweeting and citations, it’s most likely confounded by the overall interest in the paper, he continues. ‘That is, papers that are of interest to scientists are more likely to be tweeted about, and also more likely to be cited. So, indeed, even if a paper is not tweeted about at all, it would still be cited because the paper itself is of interest. Nonetheless, there may be a small increase in citations if a paper is tweeted about.’