Getting the Impact Factor Genie Back in the Box
On occasion The Official PLOS Blog presents Thought Leadership interviews with scientists leading the way on issues integral to the transformation of science communication and advancement of Open Science. Previous interviewees include Bruce Alberts and Trevor Bedford. Here we present our conversation with Sandra Schmid from University of Texas Southwestern Medical Center.
Over the years, Sandra Schmid has gained a reputation for academic strength and leadership, most recently as Professor and Chairman of the cell biology department at University of Texas Southwestern (UTSW) Medical Center. She’s also gained a reputation for her honesty regarding varied issues, including the position of post-docs, “if it were a job, we’d pay you better and give you retirement benefits,” the training of faculty, “few of us as mentors, as Principal Investigators, were ever taught how to run a lab or how to mentor individuals” and how she participates in open discussion of research before publication “mostly over beers.”
Schmid has been particularly vocal about the misuse of journal impact factors (JIFs) as a way to evaluate researchers and, as she claims, “the unfortunate consequences to the scientific community of their misuse.” At UTSW, Schmid’s home institution, there has been no overt discussion among the leadership regarding JIFs and where faculty should choose to publish. There is no formalized preference for high impact journals. “In fact, we celebrated the founding of eLIFE [a journal which rejects the use of JIFs] and have faculty on the Editorial Board of the journal,” says Schmid. The JIF was “never intended to and indeed does not measure the quality or impact of the individual papers in a journal,” says Schmid. It was originally developed and commercialized by Eugene Garfield to help librarians decide on which journals they should spend their subscription dollars.
“Individuals and institutions are being spuriously judged – by other scientists, funders, governing bodies and administrators – based indirectly on JIF, rather than directly on the quality and impact of their work,” Schmid wrote in “Negative Consequences of the Misuse of Journal Impact Factors for Scientific Assessment” as part of the 8th Forum on the Internationalization of Sciences and Humanities.
Flawed Statistics
The JIF is a statistic calculated based on the average citations of a selection of papers in a given journal. One major problem with the JIF is that citations are highly skewed, with most articles receiving fewer citations. Since citation distributions are skewed, averages are meaningless. “Indeed,” writes Schmid, there are journals that “flaunt their JIF in marketing material to authors that would ironically not accept papers reporting such flawed statistics.” This skewed distribution was clearly demonstrated last year through a collaboration between multiple publishers, including Université de Montréal, Imperial College London, PLOS, eLife, EMBO Journal, The Royal Society, Nature and Science (see Measuring Up: Impact Factors Do Not Reflect Article Citation Rates). The analysis, posted on bioRxiv, showed that citation distributions of journals with clearly distinct impact factors greatly overlap—in other words that all journals publish many papers with similar lower numbers of citations, and few highly cited papers.
A Better Option: Citation Distributions
The authors of the bioRxiv analysis call for publishers to make publicly available the actual citation distributions of their journal’s articles, rather than rely on irrelevant and misleading JIFs. Since journals use many different techniques to artificially increase their impact factor, including publication of review articles (which are often more highly cited than the original research papers they review) and front matter, including commentary and mini-review articles (that generate citations but are not counted as “citable” content) comparison across journals is problematic. It is hoped that public disclosure of article citation distributions will lead to more granular comparisons and better informed decisions by authors on where to submit their work.
Then and Now
From the perspective of a senior investigator with a long-established career and history of publishing quality work at all tiers of influence, what has changed for Schmid when deciding where to publish is that in the past, “journals had different purposes and different scopes” and that was good. Before there was the JIF there was an understanding of what journal went with what type of data. “We sent our best biochemistry to Journal of Biological Chemistry; our best cell biology to Journal of Cell Biology. If we happened upon a new and potentially important discovery, even before we understood mechanism, we’d communicate it rapidly in Science and Nature because they were three figure papers.” Before the advent of supplemental materials, more meaty, in-depth studies were published in non-page limited, subject-specific journals.
When asked in the post-print era, how do researchers decide where to publish, Schmid replies, “That is the unfortunate part.” A lot of the decisions are being made by postdocs telling her about impact factors, although she cautions that “publishing in high impact factor journals doesn’t mean it’s high quality work.” Early career researchers are looking at numbers as a distinguisher between journals, says Schmid, so her efforts are focused on getting these scientists to think more broadly. Her response and recommendation? First and foremost is to choose the journal where the work will get in front of the audience that matters the most. Schmid is crystal clear when outlining her main considerations for deciding where to publish her work and the work from her lab:
- Are the people who handle my paper able to identify qualified referees?
- Are the editors going to understand the discussion and criticisms and be helpful in handling my papers; do they understand my field?
- Do my peers read and respect the content in this journal?
Unintended Consequences
The real question for Schmid is how to get the “impact factor genie” as she calls it, “back in the box.” Why is this so important? Scientists and publishers often focus on the limitations of JIFs and the benefits of evaluating work at the article rather than journal level. However, there are more than just limitations to the JIF. According to Schmid there are very “specific and unintended consequences of the abuse of JIF as a tool for individual and institutional assessment.” Many of these, she notes, are direct; others are subtle, downstream ramifications:
- Deferred communication of discoveries that might launch new fields as reviewers and editors demand more information per paper
- Discouraged follow-up or augmentative studies to verify results due to over-interpretation of findings for the purpose of artificially inflating a work’s value
- Misguided evaluation of graduate students, postdoctoral fellows and junior faculty by their individual papers rather than the combinatorial impact of their work in context
- Wasted time and resources spent satisfying unnecessary demands of reviewers and editors in high-impact journals
- Demoralized early career researchers forced to package an entire thesis or postdoctoral project into one comprehensive paper
A Better Option: Article-Level Metrics
Perhaps wanting to get that impact factor genie back in the box was more than a mistaken mixing of two idioms. The difficulty of reverting to a situation that formerly existed (putting the genie back in the bottle) combined with the repercussions of doing something that causes unexpected and unintended negative consequences (opening a Pandora’s box) does describe the situation the scientific community has with JIFs. Fortunately, this is not an impossible situation to remedy. Article-Level Metrics were developed by PLOS as a better means to assess research value in an electronically networked world. They are gaining acceptance across a broad swath of the scientific community, from scientists to funders and more, since they provide granularity, breadth and proximity (PLOS ALMs are updated daily to monthly, depending on source and age of the individual article). ALMs also allow different scholarly research outputs to be tracked, such as policy impact, datasets, software and code. Schmid also recommends simply using PubMed as a portal for assessing the influence of an article, stating, “from title to abstract to download is a good metric,” although not as complete as a suite of ALMs.
Leadership in Practice
In 2013 as Schmid took up the position of Chair of the Department of Cell Biology at UTSW, she offered an employer’s manifesto (published as a Science Careers column) on the approach her department would take in hiring new assistant professors. This manifesto promised “a better job of screening applicants—and to avoid inappropriate criteria such as journal impact factors.” The idea was to encourage applications from qualified candidates who “might feel sidelined because their paper has yet to be, or perhaps won’t be, published in a high-impact journal.” Schmid closed her column with an enthusiastic “Let’s run this experiment!” Four years later, she shared some of the results with PLOS. Using their Academic Jobs portal the entire faculty is engaged in viewing applicants and every candidate that has piqued the interest of even one faculty member is interviewed via Skype, removing the need for reaching a ‘consensus’ that might rely more on JIFs. Those few candidates whose programs are most likely to thrive in the department’s specific environment are invited to campus to visit. Since taking this approach “our new faculty are indeed thriving,” says Schmid.
This approach suggests that a reduction on emphasis of JIFs in favor of more constructive and meaningful measures of evaluation, both quantitative and qualitative, fosters an assessment program that is both fair and thoughtful. This is how science should be; if it works for people it can work for research outputs as well.
*************
Sandra Schmid is Cecil H. Green Distinguished Chair in Cellular and Molecular Biology, Professor and Chairman, Department of Cell Biology at the University of Texas Southwestern Medical Center. She was co-founding editor of Traffic, Editor-in-Chief of Molecular Biology of the Cell and president of the American Society for Cell Biology. Schmid was elected fellow of the American Association for the Advancement of Science and Vice-Chair of the European Molecular Biology Laboratory Scientific Advisory Committee.