The unintended impact of impact factors

By -


Dr. Mickey Marks of UPenn stopped by PSPG yesterday to discuss the San Francisco Declaration on Research Assessment (DORA) which calls for new metrics to determine the value of scientific contributions. The system in question is the Thomson Reuters’ Impact Factor (IF) which was developed in the 1970s to help libraries decide which journals to curate. Since then IF has taken on an inflated level of importance that can even influence promotional and hiring decisions.  But can a single number really summarize the value of a scientific publication?
IF is calculated by dividing the average number of citations by the number of citable articles a journal has published over the last two years. One reason Dr. Marks became involved in DORA is because he is co-editor at a journal whose IF had been steadily dropping over the last few years, a trend experienced by numerous other cell biology journals. This led many in the field to question whether IF was really accurate and useful. As you might imagine there are many factors that can skew IF one way or another: for example, in some fields papers are slower to catch on and might not start accumulating citations until well past the two year IF has been calculated. Journal editors can game the system by reducing the number of “citable” articles they publish: citable articles must be a certain length, so if a journal publishes many short articles they can decrease the denominator and inflate their IF. So how reliable is the IF system? Are journals with a high IF really presenting the best science? A few years ago the editors at one journal (Infection and Immunity) set out to address that very question and the answer may (or may not) surprise you. The editors found a strong correlation between IF and retractions (see graph).

Infect. Immun. October 2011 vol. 79 no. 10 3855-3859
          Why are these high impact journals forced to retract at such a higher rate? It might be because these editors are looking for sexy science (because that’s what sells) and might be willing to overlook sloppy research conduct to print an exciting story. Another reason may be that researchers are under extreme pressure to publish in these journals and are willing to omit inconsistent data and allow mistakes and even misconduct slip through to keep the story neat. And this brings us to the real problem presented by IF: in some circles individual researchers’ scientific contributions are being judged almost entirely on what journals they publish in. Scientists learn very early in their careers that if you want a faculty job you need to publish in Science, Nature and Cell. This is because it is faster and easier to draw conclusions based on a journal’s reputation rather than actually read an applicant’s publications.
There are a few alternatives to IF, including Eigenfactor and SCImago which are similar to IF but take into account the overall impact of each journal, and Google’s PageRank which ranks journals based on search engine results. These alternatives generally result in similar rankings to IF, however. The real issue isn’t the rankings themselves but how we as scientists use them. If the system is going to change it will have to start with us.  Scientists must decide together to de-emphasize impact factors and publication rankings when making decisions about promotions, hirings and grants.

Nicole Aiello
PSPG Communications

Comments

Post a Comment

Popular posts from this blog

2015 AAAS Science and Technology Policy Forum Summary

By -
I recently had the opportunity to attend the 2015 AAAS Science and Technology Policy Forum in Washington, D.C. This annual meeting brings together a range of academics and professionals to discuss the broad S&T policy landscape. Below are some of my takeaways from the meeting. I hope to have additional comments from other National Science Policy Group members up soon. By Chris Yarosh The talks and panels at the Forum encompassed a huge range of topics from the federal budget and the appropriations outlook to manufacturing policy and, of course, shrimp treadmills . My opinion of the uniting themes tying this gamut together is just that—my opinion— and should only be taken as such. That being said, the threads I picked on in many of the talks can be summarized by three C’s: cooperation, communication, and citizenship . First up, cooperation. Although sequestration’s most jarring impacts have faded, AAAS’s budget guru Matthew Hourihan warns that fiscal year 2016 could
Read more

AAAS Forum Take #2

By -
Another point of view of the AAAS Forum by Matthew Facciani: I have provided scientific testimony and met with some of my local legislators, but I’ve never had any formal exposure to science policy. I was really excited to hear about the AAAS Science & Technology Policy Forum to learn more about how scientists can impact policy. The information I absorbed at the conference was overwhelming, but incredibly stimulating. Some of the lectures discussed the budget cuts and the depressing barriers for achieving science policy. However, I felt there was definitely an atmosphere of optimism at the conference and it was focused on how we can create positive change. One of my favorite aspects of the conference were the discussions of how to effectively communicate science to non-scientists. Before we can even have discussions of funding, the general public needs to understand how science works and why basic science is so important. For example, science never proves anything with 100%
Read more

Dr. Sarah Cavanaugh discusses biomedical research in her talk, "Homo sapiens: the ideal animal model"

By -
Image
Biology and preclinical medicine rely heavily upon research in animal models such as rodents, dogs, and chimps. But how translatable are the findings from these animal models to humans? And what alternative systems are being developed to provide more applicable results while reducing the number of research animals? Image courtesy of PCRM Last Thursday, PSPG invited Dr. Sarah Cavanaugh from the Physicians Committee for Responsible Medicine to discuss these issues. In her talk entitled, “ Homo sapiens : the ideal animal model,” she emphasized that we are not particularly good at translating results from animal models into human patients. Data from the FDA says that 90% of drugs that perform well in animal studies fail when tested in clinical trials.  It may seem obvious, but it is important to point out that the biology of mice is not identical to human biology. Scientific publications have demonstrated important dissimilarities in regards to the pathology of inflammation , d
Read more