Re: Peer-Review Costs, Rejection Rates and Journal Quality

From: Stevan Harnad <>
Date: Sat, 15 Mar 2003 18:03:47 +0000

On Sat, 15 Mar 2003, Jim Till wrote:

> <>
> See: <>
> [Pieter Bolman]: "They [PLoS] charge $1,500 an article [in a page fee
> charged to the author], and I think that's $1,500 per accepted article. So
> say it costs the same as BioMed Central thinks it does. BioMed Central
> charges $500, and they base that on a rejection rate of 50 percent, which
> means every article costs $250 to put through the refereeing process and
> produce. If you apply that same number to the PLoS, then their rejection
> rate is going to be 83 percent. That strikes me as, first of all, pretty
> elitist and sort of a drop on a hot plate. That will never become a large
> journal."

(1) An 83% rejection rate is not "elitist" (and there are journals with
even higher rejection rates!). It is simply an indicator of the high
quality-level for which PLoS is aiming (to compete with Science, Natures,
Cell, NEJM). A journal-name is a quality-control tag, indicating the level
in the quality hierarchy (usually a bell curve) at which the journal's
contents tent to be situated. The reader can then decide for himself at
what level he wishes to read and rely on the literature.

(2) Peer-review is not the passive acceptance of the top 17% of what
is submitted; it is a dynamic interaction between referees and author,
mediated by the editor, and often invoving one or more rounds of revision
and re-refereeing. It is less a measure of the quality of raw submissions
than a measure of what percentage of them were successfully revisable
to meet the journal's quality standards.

(3) The figures will eventually become known, but I do not believe for
one minute that the true cost of peer review alone (none of the other
journal functions wrapped into it!) averages more than $500 per
*accepted* paper with the costs of the rejected papers already factored
into that figure (like a "shoplifting" surcharge). So this inflation of
the peer-review cost by the rejection rate is almost certainly
spurious. (Many of the rejected papers are rejected summarily by the
editor, without undergoing full peer review; other rejections involve
only one round of refereeing and no revision. But a lower "submission
fee", creditable toward the peer-review fee if the paper is accepted,
might help discourage nuisance-submissions, wasting the refereeing time
of the entire hierarchy of journals, often with no effort at revision,
until these papers find their way to their proper level in the quality

> [Jim Till]: If the APF (article processing fee) for any particular
> open-access journal will, in the longer term, be determined mainly by that
> journal's rejection rate, then might the APF come to be regarded as a
> proxy for the quality of the journal? If this does happen, the rejection
> rate might be a rather poor proxy for quality, because of evidence that
> the rejection rate varies across journals in different research fields.

(1) Distinguish APF -- which may have many other things such as
copy-editing, tagging, markup, composition, and even printing and
distribution costs factored into it -- from the costs of implementing
peer review alone. (The peers review for free.) Any journal that does
more than peer review (especially one that produces a paper edition too)
will have difficulty estimating what its costs for peer review alone are
or would be. But $500 (per accepted paper, with rejections factored in)
is almost certainly the ceiling.

(2) There are many reasons why the peer-review cost is unlikely
to co-vary with or be predictive of journal quality. Among them are
factors such as submission rate, economies of scale (publishing multiple
journals vs. single journals), and research field differences (as Jim
notes). Old and new measures of journal impact are likely to be much
better predictors, though they too have to be weighted differently for
different fields, depending on size and on rate of research progress.

Stevan Harnad
Received on Sat Mar 15 2003 - 18:03:47 GMT

This archive was generated by hypermail 2.3.0 : Fri Dec 10 2010 - 19:46:55 GMT