Research disciplines that take off on Twitter are also well-funded, with a couple of interesting exceptions.
Over two million research documents have been shared on social or mainstream media. Twitter accounts for roughly 20 times the number of mainstream media mentions; a tweet is less effort than a newspaper article.
Over half of these documents are about medical research, referring to around 800,000 original academic papers (at July 2014). New research from Tamar Loach and Jonathan Adams at Digital Science in London examines these papers. Many of these are only shared once – perhaps by the author or their university. But there is a significant group that are shared by thousands of people. These super-papers can be everything from novelty articles about how dogs look like their owners to major breakthroughs in cancer treatment.
Tamar and Jonathan took the top 1% of papers by media mentions, and divided them by disease. They then compared the number of super-papers with UK charity research funding for that disease. They produced this graph by adding up the number of media mentions of super-papers in each disease category:
Most articles that pique the interest of Twitter are in areas with lots of research funding like cancer and neurological disorders. There is a general trend, but the correlation between mentions and spend is not quite statistically significant (0.06>P>0.05).
Digital Science suggests a couple of reasons for this trend.
It might be the case that charity research funding is a proxy for public, and therefore social media, interest in a disease. This makes sense for cancer, where media attention and funding are both high because many people have suffered or know someone who has suffered from cancer.
Some biomedical researchers and clinical and health professionals may use media mentions to draw attention to valuable and innovative outcomes, techniques and recommendations. The relative frequency of mentions for papers that review the state of the art in a field does suggest that, but the data cannot be deconstructed sufficiently to confirm it.
For immune system diseases, few papers get the public traction despite large amounts of research in those areas. Infectious diseases - including virus outbreaks and first aid - are subject to social media discussion disproportionately often relative to their funding levels. Cardiovascular disease also receives much more attention in this analysis than would be expected given the apparently lower level of charitable research spend.
These differences add more colour to the Digital Science team’s suggested reasons for the relationship between social media attention and funding.
It is easy to see how the public imagination is captured by papers that discuss the spread of infectious disease. Social media responds directly to events in the world. It is plausible that some papers are shared because of fears of real world epidemics. Heart disease also figures high amongst public fears, and it makes sense that papers covering cardiovascular disorders receive almost as much attention as cancer papers. What is surprising is that these receive such a small proportion of public funding. It would be an oversimplification to turn this into an argument for heart and infectious disease research – Twitter attention shouldn’t be directing research funding. But it does raise a question about why these levels are so low given public interest in these areas.
Immune system diseases include asthma and type 1 diabetes. These persistent diseases have large patient communities, but do not relate to the same kind of public fear as epidemics or heart attacks. This might account for why papers in this area lead to fewer social media spikes.
Looking to healthcare professions, it’s easy to see why they would be sharing advice on first aid and the latest infection risks. However, the low level of interest in common conditions like asthma points away from this professional education use of Twitter.
A new way to measure the wider impact of research?
Government assessment of research is sensitive to different cultures; surgeons publish more often than health economists. In the same way, any future use of social media to understand research impact will need to be sensitive to the biases of a medium like Twitter – its time sensitivity, the need for a subject to capture someone’s imagination in 140 characters.
This analysis of what does and doesn’t spark a social media conversation says something about the way that wider digital communities form around technology or scientific research. It is quite different to vast web scraping behind Google Trends, where contextual information about why a discussion takes off is even harder to pin down.
Methodologies for analysing mention data need to progress. For example, it would be valuable to develop archetypes of different kinds of users. Who are the facilitators or brokers who communicate research results between academic and non-professional networks? Who are these nodes? Are they Kardashian-like scientists, whose media profile is bigger than their research reputation? Or are they super-patients, who spend their spare time researching their own disease? Understanding the digital social dynamics at play here – understanding where the real effort and impact is happening – starts with counting mentions. But this needs to be complemented by other kinds of analysis too.