Reposted from Science Magazine, 23 September 2005
CHICAGO, ILLINOIS--It's the closest most scientists will come to picking their own jurors. Amid all the checklists, bibliographic information, and file-attachment instructions, the manuscript submission forms of many journals ask authors a simple question: Are there any individuals you would like to suggest or exclude as potential reviewers?
Having a say over who will review one's work should be a good thing. Authors may be better placed than editors to know who is best qualified to evaluate their findings, and they may have valid reasons for keeping sensitive results out of the hands of a close competitor. Yet many decline to suggest reviewers, and only a small percentage opt to exclude them.
That may change, thanks to the results of three studies presented here last week at the Fifth International Congress on Peer Review and Biomedical Publication, organized by the Journal of the American Medical Association and the British Medical Journal (BMJ) Publishing Group. Either suggesting or excluding reviewers, the studies show, can significantly increase a manuscript's chances of being accepted.
"The studies point out a potential for bias in the peer-review system," says R. Brian Haynes, a clinical epidemiologist at McMaster University in Ontario, Canada, and the editor of two clinical journals. "If that's the case, this is something we should be taking a closer look at."
Journal editors who use author-suggested reviewers tend to disagree about their value, says Sara Schroter, a senior researcher at the BMJ Publishing Group. So she and colleagues compared author-suggested reviews to those solicited by editors at 10 journals owned by the company, including Heart, Tobacco Control, and BMJ itself. In a 9-month survey of 788 reviews for 329 manuscripts, the team found no significant difference in the quality (as measured by widely agreed upon criteria judged to be essential for a good review) or timeliness of reviews between the two groups. However, they did find that, compared to editor-suggested reviewers, author-suggested reviewers were more likely to recommend manuscript publication (55.7% versus 49.5%) and less likely to recommend rejection (14.4% versus 24.1%).
"Editors and authors can be confident that either group will do an adequate job at reviewing the manuscript," says Schroter. "But editors should be a bit more cautious about relying on the recommendations of author-suggested reviewers."
Choose wisely. Author-suggested reviewers are more likely to recommend manuscript acceptance and less likely to advocate rejection than editor-suggested reviewers, according to studies led by Sara Schroter (above) and Elizabeth Wager (below).
Schroter's findings are reinforced by a study conducted by journal consultant Elizabeth Wager and colleagues at BioMed Central, an open-access publisher of online journals. Wager's team compared editor-chosen and author-suggested reviews submitted to 40 of BioMed Central's journals. Using criteria similar to Schroter's, the researchers found little difference in quality between the two groups of reviews. And, like Schroter, they found that author-suggested reviewers were more likely to advocate manuscript acceptance (47% versus 35%) and less likely to recommend rejection (10% versus 23%).
Opting to exclude reviewers may have an even more dramatic effect on a manuscript's success. Lowell Goldsmith, a dermatological geneticist at the University of North Carolina, Chapel Hill, and the editor of the Journal of Investigative Dermatology, and colleagues looked at 228 consecutive manuscript submissions to the journal in 2003. The team found that the odds of acceptance were twice as high for manuscripts for which authors had excluded reviewers compared to those whose authors had not done so. "Excluding reviewers ends up being very, very important," says Goldsmith. "People know their assassins."
What's driving these numbers is not clear. If authors tend to suggest sympathetic reviewers and exclude nitpicky ones, for example, the findings could spotlight biases in the peer-review process. Similarly, bias may be introduced by reviewers in journals at which reviews are not anonymous. Says Wager: "Author-suggested reviewers don't want to be the person that killed their recommender's last study."
But David Nordstrom, an epidemiologist at the University of Minnesota, Twin Cities, and an adviser on grant applications and peer review, isn't as cynical. "I take a fairly benign view," he says: Author-suggested reviewers tend to be familiar with the author's field and may be in a better position to recognize the potential impact of a paper. And Haynes says that more-established researchers, who may have the hubris to exclude reviewers, may also have a better chance of getting manuscripts accepted.
Are such author-tailored reviews likely to increase? Matthias Egger, an epidemiologist at the University of Bern in Switzerland and an associate editor of the International Journal of Epidemiology, says it's hard to predict. Many authors are loath to exclude reviewers because it goes against their ideal vision of what science should be about, he says: "Scientists like to believe that personal factors shouldn't play a role in science."
At the same time, he says, there are valid reasons to single out reviewers. Some scientists hold grudges, Egger says. Others may have conflicts of interest or are just not qualified to evaluate certain topics. So suggesting or excluding reviewers may help limit bias rather than introduce it. "I've never excluded a reviewer," he says, "but perhaps it isn't such a bad thing to do."