Information Technology Reference
In-Depth Information
100%
90%
20,.75,.75
20,.75,.50
80%
70%
20,.50,.75
20,.50,.50
40,.75,.75
40,.75,.50
40,.50,.75
40,.50,.50
60%
50%
40%
30%
20%
10%
0%
0 .1 .20
0.3
0.4
0.5
0.6
0.7
0.8
0.9
true
legend: npub,rttq,rtjq (numbe
manuscript quality, correlation
er of articles published, correlation between talent and
n between true and judged manuscript quality)
Fig. 3.
Percentage of authors p
40 track records
published in the first journal issue who accrued the highest 20
and
4
Discussion
Our simulation examined h
and track record would ch
journal space. The results
competition outcomes whe
authors and the quality of t
of the manuscripts. In pa
proportion of highest-quali
most-talented authors with
to track record also increas
their reputation before ot
published elite who leave n
Although our simulatio
simulation of Squazzoni an
theirs and with the empiri
Bourne & Barbour 2011;
provide additional support
Effect. And they prompt an
The use of track record
correlations between talen
quality, approach r = +1.0
would always produce the
as best, and the past wou
manuscript quality and trac
how changing the relative weights given manuscript qua
ange the outcomes of competitions for a scarce resour
show that changing the weights has a strong effect
en there are imperfect relationships between the talent
their manuscripts, and between the true and judged qua
articular, as more weight is given to track record,
ity manuscripts published declines, and the proportion
the best track records declines. Increasing the weight gi
es the proportion of early-published authors who estab
thers do, eventually creating a “closed shop” of fi
o room for later arrivals.
on models the editorial task differently than does
nd Gandelli (2012), our general findings are consistent w
ical findings of others (e.g., Bornmann & Daniel, 20
Dewett & Denisi 2004; Ferber, 1986). The findi
for Merton's (1968) original observation of the Matth
n important question: Why use track record at all?
for making publication decisions can be justified when
nt and manuscript quality, and between true and jud
0. At the extreme, if r = +1.0, the most talented auth
best manuscripts, the reviewers would always judge th
uld always predict the present. Under these conditio
ck record would be equivalent, but track record would
ality
rce:
t on
t of
ality
the
n of
ven
lish
irst-
the
with
007;
ings
hew
the
dged
hors
hem
ons,
d be
