There are many proposed benefits associated with the use of social media by scientists. There have been a couple of excellent pieces recently published that provide an overview of social media and some of the potential benefits of its use. Last month I wrote about tracking the exposure and reach of my tweets to measure the potential impact of public health awareness activities. Twitter seems to work well in providing exposure for public health messages, could it be used to increase exposure of new publications?
The role of social media in the promotion of research and publications has already received some attention. A study published in the Journal of Medical Internet Research (JMIR) measured the quantity of tweets linking to publications in JMIR. The authors found that 4208 tweets cited 286 distinct JMIR articles and concluded that “highly tweeted articles were 11 times more likely to be highly cited than less-tweeted articles”. Similarly, a study analysing access to and citation of pre-print publications posted to the arXiv database (http://arxiv.org) found that “the volume of Twitter mentions is statistically correlated with arXiv downloads and early citations just months after the publication of a preprint”.
There have also some interesting observations by Melissa Terras on her blog about the use of social media to increase exposure of publications. Melissa found that publications she blogged or tweeted about had at more than 10 times the number of downloads than her other publications. In particularly, Melissa posted a nice piece on increased access to one of her recently published open access papers after she had tweeted about it.
I’ve been planning to do something similar with my publications but just hadn’t had an opportunity to do it. One of the other issues is that I generally don’t publish in open access journals. I’ve been guilty of simply submitting articles to journals that I had previously published in or that were considered the key journals of mosquito research or were of regional importance (e.g. Australian Journal of Entomology, Journal of the American Mosquito Control Association, Journal of Vector Ecology).
On 7 May 2013, a publication that I was co-author on was published online in the open access journal PLoS ONE (Gonsalves L, Law B, Webb C, Monamy V (2013) Foraging Ranges of Insectivorous Bats Shift Relative to Changes in Mosquito Abundance. PLoS ONE 8(5): e64081. doi:10.1371/journal.pone.0064081). I was looking forward to the publication of this paper. It presented some of the research from a very exciting project investigating the ecological role of mosquitoes in coastal environments. I was also excited about publishing in PLoS ONE and having some research more widely (and freely) available.
Since PLoS ONE provide metrics on each of their publications, I thought I’d take the opportunity to track some of the basic metrics to see if activity on social media may influence exposure of the publication. Each day, for almost four weeks, I made a record of the page views, downloads and “social shares” (Facebook and Twitter mentions). I made a conscience effort to split my “self-promotion” tweeting into three distinct periods, the first few days after publication, a week or so later and then an additional week later.
Rather than just track our paper, I thought I’d also track some other papers published on the same day. I choose two “mosquito-related” papers, one “general health” related paper and two “ecology” papers. In selecting these papers, I simply browsed the list of publications to see what else had been published that day, I didn’t give any consideration to what impact or “newsworthiness” these papers may inherently have.
The five additional papers selected were:
Kim J-Y, Ji S-Y, Goo Y-K, Na B-K, Pyo H-J, et al. (2013) Comparison of Rapid Diagnostic Tests for the Detection of Plasmodium vivax Malaria in South Korea. PLoS ONE 8(5): e64353. doi:10.1371/journal.pone.0064353 [the “malaria” paper]
Villabona-Arenas CJ, Mondini A, Bosch I, Schimitt D, Calzavara-Silva CE, et al. (2013) Dengue Virus Type 3 Adaptive Changes during Epidemics in São Jose de Rio Preto, Brazil, 2006–2007. PLoS ONE 8(5): e63496. doi:10.1371/journal.pone.0063496 [the “dengue” paper]
Winkvist A, Bertz F, Ellegård L, Bosaeus I, Brekke HK (2013) Metabolic Risk Profile among Overweight and Obese Lactating Women in Sweden. PLoS ONE 8(5): e63629. doi:10.1371/journal.pone.0063629 [the “lactation” paper]
Jessopp M, Cronin M, Hart T (2013) Habitat-Mediated Dive Behavior in Free-Ranging Grey Seals. PLoS ONE 8(5): e63720. doi:10.1371/journal.pone.0063720 [the “seal” paper]
Moody EK, Sabo JL (2013) Crayfish Impact Desert River Ecosystem Function and Litter-Dwelling Invertebrate Communities through Association with Novel Detrital Resources. PLoS ONE 8(5): e63274. doi:10.1371/journal.pone.0063274 [the “crayfish” paper]
On the day of publication, as is usually the case with new publications, I tweeted about the paper and provided a link to PLoS ONE. I also sent an email around to my “mosquito research” colleagues. In addition, I wrote a blog post, the buzz of bat conservation, that put the paper into context with the broader research project. I tweeted about that too, in fact I probably tweeted links to the blogpost more than the paper directly during those first few days. The paper’s lead author, Leroy Gonsalves, tweeted about the paper and blog post too.
I think it is important to note that, much to my disappointment and notwithstanding media releases from both the Australian Catholic University and University of Sydney media offices, I am not aware of any substantial coverage of our paper in the online or traditional media. All the promotion for this paper seemed to come from social media.
So, what happened?
Firstly, how much exposure did our publication get via social media? The chart below shows the total daily blog post views and “social shares” of the publication itself. You can see the results of my tweeting about the blog post early on. I guess it is hard to say how many people who read the blog post then went on to view or download the paper. There were, however, also number of tweets linking to the paper directly over the first few days. The additional periods of active “tweet promotion” were between 15-17 May and 24-25 May and you can see the resulting increase in “social shares” during those periods. It is also interesting to note that “social shares” on the 19 May too that, from what I understand, was not prompted by any active tweeting on my behalf.
How did this change the exposure of the publication?
Looking at the chart of cumulative page views, you can see that, as expected, all the papers had a quick jump in the first couple of days following publication and then the number of page views remained the same for the rest of the week. A couple of interesting observations from that first week. Our paper, along with the “malaria” paper, had the most page views at around 400. While we had put our efforts into social media (with plenty of tweets and a total of 14 “social shares”), there were no “social shares” of the “malaria” paper. The paper with the most “social shares” was the “seal” paper with 28 but less than 200 page views had occurred in that first week.
After a week or so, I put some effort into tweeting about the paper, this time linking directly to the publication. I also tried tweeting at times when people in the US and UK may be more likely to be online. Over this time, there was approximately another 20 “social shares” of our paper. The chart shows the resulting boost in page views over that next week or so. While all the other papers maintained a relatively consistent number of page views, ours jumped substantially so that by the end of the second week we’d had almost twice as many page views. There was no substantial boost in social shares of the other five papers.
A week or so later (now about two weeks after publication), I repeated the same amount of tweets with links to our publication. This third burst of tweets failed to repeat the noteworthy increase of earlier efforts. Why? Perhaps by the time I got around to my third burst of tweeting, any of my followers who were interested in this work had already checked out the paper or had already retweeted my messages on earlier occasions.
There is also the possibility that the spike in page views of our article may not have been the result of that second batch of tweeting. Perhaps there was some kind of delay between people seeing the links and accessing the paper? Could you trace that spike in interest back to the initial “social media push”?
Clicking a link is one thing but was the paper downloaded?
It is interesting to compare the cumulative rates of page views to the cumulative rate of downloads. In the chart of cumulative daily downloads of our publication below, you can see that a very similar trend is followed. After an initial rise and plateau, there is a secondary jump in downloads. There is a similar increase in the number of downloads of all the papers but it is quite dramatic in ours. However, after two weeks or so, and despite additional tweets with links to the paper, downloads grow at a very slow rate. This trend is also shown in the download data of the other publications.
Putting aside the debate around the timing of tweets their resulting influence on metrics, at the end of the three week period, our paper had received almost twice as many “social shares” as any of the other papers, and subsequently, substantially more page views and downloads. Surely the social media effort assisted in this result? I don’t want to draw too much from this relatively simple analysis but I think the resulting increase in exposure of the publication has been worth the relatively small amount of time invested in spreading the word via Twitter.
Lastly, I think it is important to make a note about the importance of the “traditional” media. As I mentioned earlier, I was both surprised and disappointed at the lack of coverage the publication received. I thought a new study that contributes some answers to one of the most commonly asked questions I get, “are mosquitoes good for anything?”, would have generated more interested. I guess all researchers think their research will attract wider interest!
So what happens when a paper is picked up and widely publicised? It is interesting to look at another recently published paper in PLoS ONE. Smallegange RC, van Gemert G-J, van de Vegte-Bolmer M, Gezan S, Takken W, et al. (2013) Malaria Infected Mosquitoes Express Enhanced Attraction to Human Odor. PLoS ONE 8(5): e63602. doi:10.1371/journal.pone.0063602 was published a week or so after ours on the 15 May. It is a great study with an interesting story. The researchers found that mosquitoes infested with malaria parasites are more attracted to humans than non-infected mosquitoes. It was really no surprised that it has been picked up by media outlets across the globe. There have been around 60 news items online that reference the paper, 325 “social shares” and the paper has been downloaded over 700 times. Could these numbers have been generated by social media alone? I have a sneaking suspicion that traditional media played a significant role in the promotion of this publication but social media, and the spread of links to both news coverage and the paper directly, must have played a role as well.
UPDATE [6 March 2014]
It has now been around 10 months since our paper was published so I thought I’d revisit the metrics for all these papers to see if the trend we observed over the first few weeks continued.
It is interesting to see that our “Bats” paper received the most page views (2,186) over that time while the others ranged from 715 through to 1312. However, there wasn’t such a dramatic difference in the number of downloads with our paper downloaded 342 times compared to the 298 of the “Malaria” and 252 of the “Dengue” papers.
There is no doubt that our paper received the most “social shares” but it is also worth noting that there were plenty of Facebook shares of the “Seal” paper but that didn’t result in a boost to either page views or downloads compared to the other papers. In fact, there wasn’t much difference compared to papers with minimal or no “social shares”.
What does this mean almost a year on from publication? With regard to our paper, “social shares”, particularly Twitter, seemed to boost the number of page views we received. Given the result with the “Seals” paper, it is tempted to suggest that Twitter shares are more important than Facebook shares but I suspect I may be drawing a long bow on that one.
I don’t see anything in this analysis that suggests it isn’t worth putting in a bit of effort to promote new publications via social media. It would have been interesting to see what these metrics were like had one of these six papers tracked had been picked up by traditional media outlets. I suspect that working closely with your institution’s media office will be just as important, probably more so, than just relying on send out a few tweets.