In an analysis of recent news releases from the top ten US universities, only 59% of them provided links to the peer review paper they were written about. Over 30% did not cite the paper. And less than 10% of the releases used DOI links to the peer reviewed paper. The analysis was based on the most recent ten news releases on peer review papers up to September 12 2014 from the top US universities, as ranked by US News and World Report.
A more complete analysis of 30 of the top US universities is available here.
Frustration
I’ve long been frustrated by news releases from academic institutions that highlight ground breaking and transformative research, and yet fail to link to the original publication, or in some cases even neglect to give enough information to find the paper easily with a citation search. If the release highlights research I’m interested in, I want to read the paper, not the press release. And if I think the research may be of interest to someone else, I need to be able to tell if and when the news release has misrepresented or over-hyped it.
Anecdotally, the practice of promoting research without enabling readers to validate the information for themselves seems commonplace. But, realizing that this may be conformation bias on my part, I recently carried out a quick analysis of fifty recent news releases from top US academic institutions.
Analysis
The methodology I used was pretty straight forward: The institutions I used were the top ten ranked universities in the US, according to US News and World Report. For each of these, I analyzed the most recent ten news releases based on peer review papers, up to and including those released on September 12.
Each news release was assessed to see:
1. Whether it included a full citation to the paper that was being promoted, either as a text citation or a link to the paper;
2. Whether it included an html link to the paper; and
3. Whether it used a DOI link to the paper.
The last point is a little nerdy, but it’s an important one, as journals can and do occasionally change the URLs of papers. The DOI link will follow a paper, regardless of whether the journal URL changes, making it the preferred option for linking to publications online.
Citations to papers included in news releases
67% of the news releases analyzed provided citations to the paper they referred to; over a third of them provided no citation at all.
When the data are broken down by institution, it’s clear that there’s a high degree of variability between institutions. In terms of citations, Yale and Duke come in with the lowest rate at 30% (remembering that with just 10 news releases analyzed, the stats aren’t great). Caltech comes out best, with a 100% record of including citations. (Note: I couldn’t find a central repository of news releases at Harvard, and so used the Harvard School of Public Health as an alternative).
Links to papers included in news releases
59% of the news releases analyzed provided links to the paper they referred to – over 40% of them provided no links at all.
Caltech included links in all the news releases analyzed. Yale and Duke only included a link in one of the ten releases examined.
There was considerable variability between releases within some institutions. Some of this seemed to be associated with news releases being written by different academic units within the institution.
Use of DOI links
Just six news releases used DOI links (as in html links using the format http://dx.doi.org/…), and one of those used a DOI link for only one of two papers it was covering.
Overall, it would seem that the current use of DOI links in university news releases is an anomaly rather than accepted practice.
Why bother linking in the first place?
I suspect that the relatively low incidence of linking to peer review publications is a throwback to when print media was dominant. If a press release was to lead to an informed piece somewhere like the New York Times, it was probably reasonable to assume that the decision to cite the original paper or not resided with the editor, and that journalists could always request a copy of the paper from institution if they were interested in reading it.
In this age of online news and cut and paste reporting though, this makes little sense. Adding a link (including a DOI link) takes marginal additional effort, and provides readers with the ability to cross-check news articles with published findings. This is extremely important where news outlets reproduce press releases word for word and where there is little or no original reporting. It’s also important because institutional press releases don’t always get their representation of the research right.
I’d like to think that the purpose of university news releases is to communicate useful information about new findings to people who will benefit from this – either through education, enrichment, or better-informed decision making. I’m sure that many (hopefully most) news releases aspire to these aims – but the primary driver is usually promotion of the institution and its achievements.
That said, many institutions work hard at balancing promotion and communication as both will typically be part of the institution’s mission. Yet without checks and balances in place, news releases can shift all too easily to promotion, and and even into mis-communication.
One of these checks and balances is providing links to source material.
In fact, without links (or at the very least citations) which allow the reader to do their own fact checking, news releases begin to look suspiciously like promotional material, and little else.
The good news is that from this quick and dirty analysis, a number of institutions are making sure that their news releases are more than just PR. And it isn’t just the top ranked universities. I reached out to long-time colleague Matt Shipman for hist take on linking to original papers – Matt works for the North Carolina State University News Services. This is what he came back with:
I almost always link to a paper (regardless of whether it’s open access). The only times I won’t link are if the paper is not online. E.g., if it’s a conference proceeding and the release is going out in advance of the conference.
Here’s an example of the format I follow for news releases: http://news.ncsu.edu/2014/09/labean-dna-origami/
You’ll note that we provide a hyperlink to the paper, all of the citation information (including DOI), and the paper’s abstract. We want to make it as easy as possible for people to find the paper.
I strongly believe that this is the right way to go – hopefully an increasing number of institutions will continue to follow this lead.
A more detailed analysis of 30 of the top ranked US universities can be found here.
Talking of linking to source materials, if anyone wants to browse through the data for this analysis, it’s downloadable as a Excel spreadsheet here.
Update 09-16-14 12:22 PM EST – Data updated to reflect inclusion of 10 news releases from each institution. Previously, only five releases per institution were used.
Update 09-16-14 1:39 PM EST – updated graphics to remove amateurish use of color!
Excellent piece, Andrew. Ironically, given that I’m quoted in the piece, today is the day I’m issuing a release *without* a link. That’s because PNAS lifts the embargo on all of the papers being released in a given week on Monday…but doesn’t post all of the papers online on Monday. Instead, the papers are published sporadically over the remainder of the week. I *did* still include the DOI, though. Just wanted to call myself out on this one. 🙂
I think it’s also worth reiterating the point about variability _within_ institutions. I just talked it over the DOI thing with the other science PIO who works in Penn’s main communications office (there’s a dedicated comms office for the Medical school, and many, many other disparate communications people all over the university). She uses DOI links and I don’t, and the only reason for that discrepancy is personal habit — which is to say that if you had sampled our news page on a different day and gotten more of her releases than mine (or the med school’s), you’d have a completely different figure for Penn.
Yeah – I need to go back and capture this, but qualitatively there is massive variation where an institution has news release writing groups in different academic units. There are also a number of institutions where there are clearly different styles and templates within the same news office.
I also meant to say that I really appreciate that you’re crunching the numbers on this type of thing, and would love to see more of this kind of analysis in the future.
This is a great piece. Sometimes I feel like authors don’t link, because they don’t want readers to question their analysis or write up about the original source. Maybe they just feel they are so esteemed that we should just take their word for it. As a layperson, I will say that if you want to have a relationship of trust, especially between parents and the science community, providing links, sources, and transparency is the only way to do that.