Page 59 - Becoming Metric Wise

49
Publishing in Scientific Journals

3.1.7 Rejection Rate
The RR of a journal (over a certain period) is the ratio of the number of
rejected manuscripts and the number of submitted ones (during that
period). Similarly, the acceptance rate (AR) is the ratio of the number of
accepted manuscripts and the number of submitted ones. Clearly,
AR 1 RR 5 1 for a given submission period. Most highly visible journals
have high RRs of more than 90% (McCook, 2006). Actual RRs are
field-dependent. Starting journals usually invite well-known colleagues to
submit articles for their first issues, and hence have initial low RRs.
Sugimoto et al. (2013) note that, simple as these notions may seem,
they are in reality not straightforward. Problematic issues are resubmis-
sions counted as new submissions (and hence the original submission is
counted as a rejection), the inclusion/exclusion of invited papers or spe-
cial issues, an unclear timeframe used, and the inclusion/exclusion of
book reviews. Moreover, if data are obtained by surveying editors they
may try to bias the outcome in one or the other way.
Already in the seventies, Zuckerman and Merton (1971) examined
disciplinary variation in RRs. They found substantial variation with RRs
of 20 40% in the physical sciences, and 70 90% in the social sciences
and humanities. They suggested two sources for the wide variation in
RRs: differences in consensus between fields of scholarship, and differ-
ences in space shortages. Note though that this is a very old article and
circumstances have changed quite considerably. Probably RRs in journals
covered by the Web of Science (WoS) have increased considerably since
this investigation.
Using Cabell’s Directories of Publishing Opportunities, for which data
are provided by editors, Sugimoto et al. (2013) found that statistically sig-
nificant differences in ARs were found by discipline, country affiliation of
the editor, and number of reviewers. In their study, including social health
science, business, computer science, education and psychology, the
health journals had the highest ARs (a median value of about 50%),
while business journals had the lowest (a median of about 25%).
Generally, journals with very high impact factors have low ARs, which
may be explained by a reinforcement cycle consisting of authors trying
to publish their work in journals with a high impact factor (even if not
really appropriate) thereby reducing the AR. Such authors’ unrealistic
expectations lead to an inefficient and clogged publication system.
Sugimoto et al. (2013) also found that OA journals had significantly
higher ARs than nonopen access ones.