Content uploaded by Giovanni Abramo
Author content
All content in this area was uploaded by Giovanni Abramo on Jan 25, 2016
Content may be subject to copyright.
Jointly published by Akadémiai Kiadó, Budapest Scientometrics, Vol. 79, No. 3 (2009) 517–539
and Springer, Dordrecht DOI: 10.1007/s11192-007-2046-8
Received January 30, 2008
Address for correspondence:
GIOVANNI ABRAMO
E-mail: abramo@disp.uniroma2.it
0138–9130/US $ 20.00
Copyright © 2009 Akadémiai Kiadó, Budapest
All rights reserved
Gender differences in research productivity:
A bibliometric analysis of the Italian academic system
GIOVANNI ABRAMO,a,b CIRIACO ANDREA D’ANGELO,a ALESSANDRO CAPRASECCAa
a Laboratory for Studies of Research and Technology Transfer, School of Engineering,
Department of Management, University of Rome “Tor Vergata”, Facoltà di Ingegneria,
Dipartimento di Ingegneria dell'Impresa, Via del Politecnico 1, 00133 Roma, Italia
b Italian Research Council
The literature dedicated to the analysis of the difference in research productivity between the
sexes tends to agree in indicating better performance for men. Through bibliometric examination
of the entire population of research personnel working in the scientific-technological disciplines of
Italian university system, this study confirms the presence of significant differences in productivity
between men and women. The differences are, however, smaller than reported in a large part of the
literature, confirming an ongoing tendency towards decline, and are also seen as more noticeable
for quantitative performance indicators than other indicators. The gap between the sexes shows
significant sectorial differences. In spite of the generally better performance of men, there are
scientific sectors in which the performance of women does not prove to be inferior.
Introduction
The study of differences in productivity between men and women employed in the
scientific world has always attracted interest from a wide range of observers. It feeds a
lively debate that covers at least two themes: psycho-cognitive and sociological. In
particular, during the past two decades, the issues of gender differences in cognitive
abilities has been addressed by numerous meta-analysis studies of verbal [HYDE &
LINN, 1988], spatial [LINNE & PETERSON, 1985; VOYER & AL., 1995] and mathematical
abilities [HYDE & AL., 1990]. Such studies do not indicate a substantial differentiation
in abilities between men and women but they do offer a characterization by typology of
ability and context of application. In addition to this issue, for the sciences in particular,
and especially in the world of research, the feminine presence still seems highly limited
and relegated to marginal roles [UE, 2006]:
x Women represent only one sixth of research workers in the private sector
and one third of the entire community of academic staff, though their
representation has increased over time.
; Published online January 31, 2009
Scientometrics 79 (2009)
518
ABRAMO & AL.: Gender differences in research productivity
x Regarding the composition of academic staff, women tend to be
concentrated in inferior roles. There is only one woman for every 3.5 men
in the top academic ranks.
x In the scientific committees appointed by the European Community the
proportion of women is about 20%, but the leadership of these committees
is entrusted to a woman in only 10% of cases.
Such statistics have stimulated studies of a sociological character to identify and
suggest potential interventions to effectively balance the situation, commencing from
the possible causes: the smaller number of women entering the field, unequal
opportunity and sexual discrimination, or lesser performance with respect to men.
In this last area, one of the most consistent findings in the literature on research
productivity is that women tend to have somewhat lower publication rates than men
[LEE & BOZEMAN, 2005]. The lesser productivity of females has been established in
tens of studies of diverse countries and disciplines, spanning decades and using a wide
variety of measures [COLE & ZUCKERMAN, 1985; FOX, 193; LONG 1987]. Indications
are that the difference in average productivity between the genders is also accompanied
by a diverse distribution of the research product. LE MOINE [1992] shows that the
concentration of women among researchers who publish a single article is greater than
for men, while their representation among “star” scientists is less. COLE & ZUCKERMAN
[1984] neatly label these gender differences as the “productivity puzzle”, although
science sociologists retain that the puzzle exists only for those who refuse to recognize
the impact of sociological determinants.
Zainab’s review of the studies on the subject [ZAINAB, 1999] identifies, among
others, certain classes of personal variables that are significantly correlated to
productivity of scientists. It results that the difference in scientific performance between
men and women is significant, but also emerges that such differences lessen over time
[COLE & ZUCKERMAN, 1984; XIE & SHAUMAN, 1998; LEAHEY, 2006], and can in part
be traced to factors other than gender, such as level of specialization [LEAHEY, 2006] or
academic position. Differences between the sexes in the early stages of career seem to
be more visible [XIE & SHAUMAN, 1998].
Within the vein of these investigations, the present study is intended to provide, for
the first time, the evidence from the Italian academic research system. It proposes to
examine:
x Whether there are differences in research productivity between men and women;
x If such differences can be identified in all the evaluation parameters for
scientific performance
x If such differences are general or present sectorial specificities;
x If such differences remain more or less constant or vary significantly with level
of employment.
Scientometrics 79 (2009) 519
ABRAMO & AL.: Gender differences in research productivity
The current study is not intended to investigate, in this phase, the causes of the
differences encountered, but the authors will indicate further investigations that the
findings could suggest.
The work here is unique with respect to the international state of the art in at least
two aspects. Firstly, for the field of observation studies in the existing literature have
been based only on samples of the population of interest and generally focalize either on
very restricted disciplinary sectors or on single institutions. Instead, the study proposed
here refers to the entire technological-scientific population of Italian universities,
consisting of approximately 33,000 research scientists. Secondly, for the manner of
comparing individual performance each scientist has been classified by role and
scientific field of specialization, with the aim of limiting the inevitable distortions in
productivity due to non-homogeneity of gender distribution among roles and scientific
sectors (see [ABRAMO & D’ANGELO, 2007]). The research products are observed as the
scientific publications in international journals recorded by the Thomson Scientific
Science Citation Index (SCI) during the period 2001 to 2003. The analysis based on the
whole population of academic research staff, avoids problems in robustness and
significance of inferential analyses. It further presents the undeniable advantage of
objectivity and homogeneity of source data, not always found in examinations based on
questionnaires.
Factors affecting scientific production
Research activities resemble a type of input-output process [MORAVCSIK, 1985], in
which the inputs consist of human and financial resources, while outputs have a more
complex character, of both tangible (publications, patents, conference presentations,
etc.) and intangible nature (personal knowledge, consulting activity, etc.). The outputs
most commonly used to evaluate results of research in science and technology are the
scientists’ publications in specialized journals, the par excellence form to communicate
the results of their research to the research community. Through this medium, scholars
obtain the recognition of their peers, a determining factor for further funding researches
and career progression [RAMSDEN, 1994].
In comparing products of research work between any two individuals and
particularly between the sexes, it is necessary to filter the effects of all factors other than
individual merit that may affect individual performance in a direct or indirect manner.
ZAINAB [1999] groups determinants of scientific performance in two categories:
personal and environmental. The following determinants are noted among the first
category:
x Gender: studies have revealed higher productivity among males, both in
analyzing specific sectors and observing specific research institutions over
Scientometrics 79 (2009)
520
ABRAMO & AL.: Gender differences in research productivity
an extended period of time [FOX, 2005; STACK, 2004; XIE & SHAUMAN,
2004; PRPIC, 2002].
x Age: certain studies seem to show the existence of a peak in productivity in
the years approaching age 40 and the years soon afterwards, followed by a
constant decline with advancing age [FOX, 1983]. Investigation of the
scientists at the National Research Council of Italy seems to confirm these
findings [BONACCORSI & DARAIO, 2003].
x Marriage: Almost all studies agree on the positive effect of marriage on the
scientific fertility of researchers, but certain studies [PRPIC, 2002] show
that men receive the greater share of the benefit due to the presence of a
spouse. FOX [2005] shows that unmarried men are the least productive of
all. Among women, those who are married, and particularly those married
for the second or third time, have a higher level of productivity.
x Children: The results from investigations of the impact of children on
productivity are not always simple to align. According to FOX [2005], the
presence of children, especially of preschool age, increases productivity
among both genders. Evidently children can motivate scientists to work
harder, enabling them to provide a higher standard of living for their
offspring. Women with preschool children are found to be especially
efficient, particularly in their allocations of time. However, in a study of a
much larger sample, STACK [2004] shows that women with preschool aged
children publish less than other women. Obviously, the time, energy, and
money devoted to child-rearing can reduce research productivity. In any
case, men with children continue to be more productive than women with
children [PRPIC, 2002].
x Level of specialization: increases in professional specialization seem to
have a positive of influence on a scientist’s research productivity. Some
studies illustrate that women tend to specialize less than men, which results
to the detriment of their productivity [LEAHEY, 2006].
Certain structural and environmental factors can also be noted:
x Academic rank (role): many studies illustrate a correlation between
academic rank and a scientist’s productivity. In a study sample of
American academics, BLACKBURN & AL. [1978] show that full professors
publish at a higher average rate than associate professors and research staff.
DICKSON [1983] and KYVIK [1990] have illustrated the same effect of
professional role on scientific productivity in their respective studies of
Canadian and Norwegian universities.
x Teaching load: in universities, research and teaching activities accompany
each other. Certain studies of performance evaluation and gender show that
Scientometrics 79 (2009) 521
ABRAMO & AL.: Gender differences in research productivity
men obtain better performance in research while women seem to excel in
(and favour) teaching activities [GANDER, 1999]. In confirmation of this
thesis, XIE & SHAUMAN [2003] reveal that even though the difference in
teaching load between the sexes is on the decline, women continue to
favour teaching activity more than men and thus, on average, devote a
lesser portion of time to research.
x Prestige of the institution or department of affiliation: certain studies
illustrate that productivity may be a function of the researcher’s institute of
affiliation. The presence of “illustrious” colleagues has a positive effect on
the productivity of the other researchers. Further, this effect is seen as more
notable among lower level researchers (post-doctoral student, research
associate, teaching fellow) and continually weaker as careers advance. But
the direction of the cause-effect relationship is unclear, as to whether it is
the better university staff teams that draw the most brilliant minds or vice
versa [FOX, 1983].
In light of the state of the art, for this study the authors chose to take in
consideration all the relevant factors cited above, either directly or indirectly, while also
working with a field of observation much wider and more representative than those of
the preceding studies. The following section of this paper proceeds with a description of
the methodological choices, highlighting their relevance with respect to the present
limits in the state of the art.
Analytical model
For the scope of this study, two variables among those indicated by current literature
were taken into particular consideration: academic role and gender. This choice does
not imply any loss of generality. Given the character of the Italian university system it
can be assumed that all the variables note above are for the most part more than
indirectly linked to the two chosen ones.
With regard to the group of personal variables, age is strongly correlated to
academic role, given the system of career progression in Italian universities. At the
same time, level of specialization of individual professionals is also implicitly taken
into consideration, since the analysis is conducted precisely by scientific-disciplinary
sector. Every academic scientist in Italy is classified in a specific sector of research,
generally very clearly defined in terms of specialization. The Italian academic system is
specifically subdivided into 14 macro disciplinary areas and 370 scientific sectors.
Scientometrics 79 (2009)
522
ABRAMO & AL.: Gender differences in research productivity
The analysis conducted here is concentrated on the 8 areas of a technological-scientific
character,1 which in turn include 183 sectors. The analysis initiates at the level of single
sectors, implying that the individuals under observation are homogenous in their level
of scientific specialization.
The inclusion of the structural-environmental variables listed above was also
considered. An analysis of distribution of personnel among academic roles by
geographic area showed substantial homogeneity, which induces the exclusion of
location-related variables from the investigation. In addition, it seems little relevant to
consider teaching load in the analysis, since the legal framework of the Italian academic
system establishes this load a priori. The variable of relative prestige of department or
institution also appears of little significance, given the publicly regulated nature of the
Italian university system. The characteristics of university degrees are legislated and the
recruitment of teaching personnel is based on rigidly regulated national competitions,
inducing a situation of great homogeneity among universities. Only in recent years have
universities attained a certain financial autonomy, still insufficient to facilitate
differences in reputation.
Data set
The data utilized for the output component of the model were taken directly from
the ORP (Italian Observatory of Public Research), developed in the authors’ home
institute. This observatory extracts data on scientific literature from the SCI and applies
procedures for disambiguation and identification of the exact origin of the
publications.2 It lists all scientific articles authored by Italian university personnel3
holding a position as assistant, associate or full professor during the three years under
consideration (2001 to 2003), in the technological-scientific university disciplinary
areas (UDAs). The research personnel were identified by extraction from a database at
the Ministry of Universities and Research4 and number approximately 33,000 scientists
from the sectors indicated by Table 1. The data show several interesting points. Firstly,
of the total of scientific staff, assistant professors number more than for any other role:
approximately 38%, compared to 33.4% for associate professors and 28.8 for full
1 Mathematics and information sciences; physical sciences; chemical sciences; earth sciences; biological
sciences; medical sciences; agriculture and veterinary sciences; industrial and information engineering. Civil
engineering and architecture, a final technical-scientific area from the national university order, was discarded
from consideration because the SCI would not be sufficiently exhaustive in representing the research output
of this area.
2 For an exhaustive description of the development and function of the observatory and the listings of
scientific production by name for Italian university researchers see ABRAMO & AL. [2007].
3 Each indexed article is assigned to all the co-authors, regardless of their position in the listing.
4 http://cercauniversita.cineca.it/php5/docenti/cerca.php. It was impossible to obtain direct information on age,
gender, or marriage of individuals due to privacy regulations. Identification of gender was obtained by
analysis of first names.
Scientometrics 79 (2009) 523
ABRAMO & AL.: Gender differences in research productivity
professors. This division by role is quite different when men and women are observed
separately. The latter, which represent slightly more than one quarter of the population,
are much more concentrated in lesser roles. More than 55% of women fall within the
role of assistant professor, compared to only 32% of men. The inverse situation occurs
for senior roles for each female full professor there are more than 8 males. This
discrepancy, substantially consistent among all disciplinary areas, has historic-social
causes and reflects the state of progress of feminine emancipation in education and
employment.
Table 1. Distribution of Italian university staff by gender, role and disciplinary area:
average of data for years 2001 to 2003
Industrial and
information engineering
Agriculture and
veterinary sciences
Biological sciences
Chemical sciences
Earth sciences
Physics
Mathematics and
information sciences
Medical sciences
Total
M 1,488
(96%)
865
(91%)
1,099
(77%)
832
(88%)
350
(91%)
748
(94%)
793
(85%)
2,260
(92%)
8,435
(89%)
Full
Professors F 60
(4%)
90
(9%)
331
(23%)
109
(12%)
35
(9%)
47
(6%)
144
(15%)
188
(8%)
1,004
(11%)
M 1,299
(90%)
660
(74%)
859
(54%)
823
(70%)
371
(77%)
788
(84%)
678
(62%)
2,684
(81%)
8,162
(75%)
Associate
Professors F 148
(10%)
231
(26%)
729
(46%)
360
(30%)
113
(23%)
146
(16%)
414
(38%)
643
(19%)
2.784
(25%)
M 1,115
(82%)
646
(58%)
747
(40%)
481
(47%)
274
(65%)
572
(73%)
572
(53%)
3,249
(68%)
7,656
(62%)
Assistant
Professors F 241
(18%)
470
(42%)
1,102
(60%)
544
(53%)
148
(35%)
215
(27%)
507
(47%)
1,548
(32%)
4,775
(38%)
M 3,901
(90%)
2,171
(73%)
2,705
(56%)
2,136
(68%)
995
(77%)
2,108
(84%)
2,043
(66%)
8,194
(77%)
24,253
(74%)
Total
F 449
(10%)
791
(27%)
2,162
(44%)
1,013
(32%)
296
(23%)
408
(16%)
1,065
(34%)
2,379
(23%)
8,563
(26%)
Total, both sexes 4,350 2,962 4,867 3,149 1,291 2,516 3,108 10,573 32,816
To permit the analysis of scientific production from single individuals over the
period of observation, scientists who did not hold a staff role throughout the entire
triennium were cut from the initial data set, eliminating all those who were assumed
after December 31, 2000 or exited prior to January 1, 2003. Scientists that changed
scientific disciplinary sector (SDS) for whatever reason were also cut from the data.5
Those who changed their role within an SDS during the triennium due to career
advancement were attributed with their role in the final year of observation (2003).
5 Problems with homonymy in names would have made exact identification of individuals difficult as they
moved from sector to sector, contributing potential errors in attribution, listing and count of publications.
Scientometrics 79 (2009)
524
ABRAMO & AL.: Gender differences in research productivity
The final data set used in the analysis is presented in Table 2. Comparing to Table 1,
one sees a distortion in favour of senior roles (associate professors and especially full
professors), caused by the procedures noted.
Table 2. Distribution of Italian university research staff by gender and role; data set 2001 to 2003
Full Professors Associate Professors Assistant Professors Total
Male 8,686 (88.7%) 7,596 (73.4%) 5,401 (60.7%) 21,683 (74.7%)
Female 1,102 (11.3%) 2,758 (26.6%) 3,493 (39.3%) 7,353 (25.3%)
Total 9,788 (33.7%) 10,354 (35.76%) 8,894 (30.6%) 29,036
Performance indicators
Individual performance was evaluated on the basis of a number of indicators:
x Output (O): the sum of publications realized by the scientist in the triennial
under consideration.
x Fractional Output (FO): the sum of the scientist’s contributions to the
publications realized, the contribution for each publication being
considered as the inverse of the number of co-authors.
x Contribution Intensity (CI): the ratio of FO to output. A value close to 1
indicates that the scientist generally excludes collaboration, publishing
articles alone; the inverse, a value close to 0, indicates that the scientist
tends to publish in co-authorship with many other colleagues.
x Scientific Strength (SS): equals the weighted sum of publications realized
by the scientist. The weight is the normalized impact factor6 of the
publishing journal.
x Fractional Scientific Strength (FSS): analogous to FO but based on the
scientific strength.
x Quality Index (QI): the ratio of scientific strength to output, indicating the
average quality of the publications authored by the scientist.
The authors are aware of limitations arising from some of the methodological
assumptions used. In particular, in this analysis only scientific journal publications are
taken into consideration as research output, which excludes other codified forms of
outputs such as proceedings, monographs, patents or prototypes. However in the
scientific sectors taken into consideration, journal publications are actually highly
representative of real output from research activity. It should be noted that when Italian
6 The distribution of the impact factors of journals is observed to differ substantially from sector to sector. The
normalization of each journal’s impact factor with respect to the sector average permits limiting the
distortions embedded in comparing performances between different sectors.
Scientometrics 79 (2009) 525
ABRAMO & AL.: Gender differences in research productivity
universities submitted their products for consideration in the first national research
evaluation [VTR-CIVR,7 2006], journal articles were a minimum of 85% and a
maximum of 99% of the total products selected by each university. In 7 of the 8
discipline areas under consideration, journal publications exceeded 90% of the total
product submitted.
Overall, the most critical consideration is the correct quantification and
classification of the articles for each university. In this regard, other than errors and
limitations attributable to the data source,8 there may also be those arising from the
identification of scientific production by author and institute name. But, as indicated by
ABRAMO & AL. [2007], the errors in the disambiguation process for author name do not
induce substantial losses in significance for the analysis, due to the limited extent of
such errors (2%) and to their uniform distribution among the data sets of the analytical
model adopted.
A further critique could concern the use of the impact factor for the journal as a
proxy for the publication quality, and therefore for the scientist’s production. This
assumption imposes a bias, but the bias diminishes at the moment that citations of
single articles are considered (as amply described and analyzed in publications such as
[WEINGART, 2004; MOED, 2002]), and in the judgment of the authors the assumption
does not significantly alter the study or the conclusions to which it gives rise.
With regards to assumptions concerning input, the major limitation lies in the
impossibility to quantify the time dedicated to research activity by university
professionals over the period under consideration. Further, there is no information on
the frequency or duration of maternity time for women9 or of sick leave in general.
Although there is no reason to expect any gender diversity in distribution of illness, the
negative impact of maternity on productivity by women could be notable, especially for
assistant professors, where the average age is 43.
Results
In the triennial period under observation, over 61.5% of Italian academic research
personnel participated in at least one scientific publication listed in the SCI™ (Table 3).
7 Triennial evaluation (2001-2003) of research activity in universities and major public research institutions;
for details see http://vtr2006.cineca.it/
8 The SCI™ lists approximately 4,800 international journals, which represent only a sample of the global
scientific press. It also lacks uniform representation from the disciplines, for example being greatly weighted
towards the life sciences.
9 Under current law the normal duration is five months, but longer leaves are frequent and are permitted by
the provisions. Leave is also permitted for men but they rarely take advantage of such provisions.
Scientometrics 79 (2009)
526
ABRAMO & AL.: Gender differences in research productivity
There is no significant difference in the data respecting men and women; 38.6% of the
latter result as inactive, versus 38.5% among men. However, when the data are
disaggregated by role and re-considered, this slight difference is overturned due to the
fact that women are primarily concentrated in “lesser” or “less active” roles, as
previously illustrated in Table 2. Among full professors, woman show 1.1% more
activity than men. Among associate professors the difference in favour of women rises
to 1.5% and among assistant professors as high as 5.2%.
When sorted by disciplinary areas the data do not show any strong lack of
homogeneity between the sexes (Table 4). The maximum and minimum values refer
respectively to male full professors in chemical sciences (where only 8.2% fail to
produce any publications during the triennium) and to female associate professors in
mathematics and information sciences (slightly less than two thirds fail to realize any
publications in the triennial under observation).
Table 3. Distribution of scientists who publish, by gender and role
Full Professors Associate Professors Assistant Professors Total
Male 5,895 (67.9%) 4,526 (59.6%) 2,921 (54.1%) 13,342 (61.5%)
Female 760 (69.0%) 1,684 (61.1%) 2,071 (59.3%) 4,515 (61.4%)
Total 6.655 (68.0%) 6,210 (60.0%) 4,992 (56.1%) 17,857 (61.5%)
Table 4. Percentage of scientists who publish, by gender, role and discipline area
I
ndustrial and
i
nformation
e
ngineering
Agriculture and
veterinary sciences
Biological sciences
Chemical sciences
Earth sciences
Physical sciences
Mathematics and
information sciences
Medical sciences
Total
M 52.7% 50.6% 80.9% 91.8% 56.0% 72.9% 56.5% 73.8% 67.9%
Full
Professors F 50.7% 61.0% 72.3% 87.6% 62.9% 66.0% 52.5% 76.6% 69.0%
M 52.3% 45.9% 69.4% 83.9% 48.8% 62.3% 44.4% 60.7% 59.6%
Associate
Professors F 56.0% 50.0% 71.2% 81.3% 52.1% 56.9% 37.1% 61.3% 61.1%
M 51.4% 43.8% 69.5% 86.4% 42.6% 72.9% 46.5% 47.8% 54.1%
Assistant
Professors F 48.7% 52.0% 68.4% 81.6% 50.0% 66.1% 40.7% 55.3% 59.3%
M 52.3% 47.5% 74.7% 87.8% 50.3% 68.9% 50.2% 60.6% 61.5%
F 52.0% 52.7% 70.2% 82.3% 52.7% 62.5% 41.1% 59.6% 61.4%
Total
Total 52.2% 48.9% 72.8% 86.1% 50.8% 67.9% 47.1% 60.4% 61.5%
Independently of their role, women consistently result as more active than male
colleagues in the areas of medical sciences, agriculture and veterinary sciences, and
earth sciences. The opposite is true for the areas of industrial and information
Scientometrics 79 (2009) 527
ABRAMO & AL.: Gender differences in research productivity
engineering, chemical sciences, physical sciences, and mathematics and information
sciences. In addition, it is evident that the percentage of scientists that result as active in
a specific area is correlated to the average intensity of publication in the area itself
(Figure 1).10
Regarding only those scientists who publish, analyses at the aggregate level shows a
noticeably skewed distribution in frequency of output (Figure 2). A 38% share of the
active scientists produces less than one article per year over the arc of the triennium.
The more productive scientists, however, contribute a notable portion of the scientific
production: 20% of scientists realize over 53% of the scientific production in the whole
of the areas under consideration for the national academic system. Significant and
notable data also emerge concerning the difference between the sexes – the curve for
distribution of publication frequency by women is more tapered than that for men, as
can be seen from the inversion of the bars in the graph shown in Figure 2 (the skewness
is 2.25 for women and 3.46 for men). The same occurs at single professional role level
(Figure 3–5). Although with few differentiations, findings show that in general men are
more concentrated than woman in the top-productivity ranks in each role.
Figure 1. Percentage of scientists who publish, by gender and discipline area.
The intensity of publication is indicated in parentheses (average annual publications per active professional)
10 Certain areas result as more productive than others either for internal reasons (time to complete projects and
develop results is substantially less) or external reasons (the number of journals listed by the SCI™ in the area
is larger than in others).
Scientometrics 79 (2009)
528
ABRAMO & AL.: Gender differences in research productivity
Figure 2. Average frequency of scientific production by Italian academic research personnel;
data from 2001 to 2003 for scientists who publish
Figure 3. Average frequency of scientific production by Italian academic research personnel;
data from 2001 to 2003 for full professors who publish
Scientometrics 79 (2009) 529
ABRAMO & AL.: Gender differences in research productivity
Figure 4. Average frequency of scientific production by Italian academic research personnel;
data from 2001 to 2003 for associate professors who publish
Figure 5. Average frequency of scientific production by Italian research academic personnel;
data from 2001 to 2003 for assistant professors who publish
Scientometrics 79 (2009)
530
ABRAMO & AL.: Gender differences in research productivity
The analysis of the homogeneity of scientific production in different professional
roles is illustrated in Table 5, which presents indexes of concentration for both genders
in all three roles. The indexes used are the cumulative production of the first two deciles
and the Gini coefficient. Considering each role, both the Gini coefficient and the
cumulative production of the first and second deciles are consistently higher for men
than for women. This phenomenon could be due, at least in part, to the presence of a
relatively high number of “star” scientists among the male population.
Table 5. Indexes of concentration of scientific production by gender and role;
data from 2001 to 2003 for scientists who publish
Full Professors Associate Professors Assistant Professors
M F M F M F
Gini coefficient 0.509 0.448 0.488 0.449 0.472 0.437
Cumulative production 1st decile 47.5% 34.0% 30.4% 19.9% 22.4% 13.1%
Cumulative production 2nd decile 65.7% 56.9% 48.4% 40.1% 39.8% 29.0%
Differences in performance
The data presented in Table 6 now permit us to address the research questions posed
at the origin of this study concerning the potential presence of significant differences in
performance between men and women. For each indicator considered, the table reports
the average general performance (
g
k
P),11 by gender and role. Table 6 also shows the
average percentile rank (Rank %) for all the sectors included in the field of observation.
These rank totals derive from the simple aggregation of the rank data for males and
females in each sector. Basing analyses on the linearization of the data along an
invariant scale from 0 to 100, the table permits evaluation of performance by
individuals, independent of the sector in which they operate (i.e. independent of the
number of scientists falling in the sector and of the sector’s fertility in publications) In
particular, Table 6 reports average data for men and women according to their role and
the performance indicators considered. Higher performance for men can be observed
along all dimensions of the evaluation. Notably, the overall average output per male
scientist is 16.8% superior to that of female scientists. But the average quality of
11 Average general performance ( gk
P
) of scientists of gender “g” and role “k” is calculated as:
1
1
average performance of scientists of gender "g" and role "k", in sector "j"
average performance of scientists of role "k", in sector "j"
number of s
SDS
ngjk
gk gjk
jk
j
gk
gjk
jk
gjk
Y
P Staff
Staff Y
Y
Y
Staff
¦
cientists of gender "g" and role "k", in sector "j"
total of scientists of gender "g" and role "k"
total of scientific sectors
gk
SDS
Staff
n
Scientometrics 79 (2009) 531
ABRAMO & AL.: Gender differences in research productivity
production is only 4.5% superior (Quality Index, total in far right column of Table 6).
Women also tend to collaborate more, seeing as their average contribution intensity (CI)
is 6% inferior to that of men. The spread between the qualitative productivity indicators
(SS and FSS) is obviously greater due to the combined effect of the differences found
above.
Table 6. Average general performance (Pgk) and average percentile rank (Rank %) of men and women
by role (percentage differences indicated in brackets)
Full Professors Assistant Professors Assistant Professors Total
Index Gender gk
P
Rank % gk
P
Rank % gk
P
Rank % gk
P
M 1.252 (+13.3%) 63.0 0.94 (+12.3%) 55.4 0.848 (+17.5%) 53.4 1.032 (+16.8%)
O F 1.105 61.5 0.837 52.4 0.722 48.6 0.884
M 1.286 (+19.7%) 56.8 0.939 (+15.9%) 48.2 0.828 (+20.2%) 46.6 1.038 (+21.8%)
SS F 1.074 55.3 0.81 45.9 0.689 42.4 0.853
M 1.238 (+15.7%) 56.9 0.965 (+16.5%) 49.9 0.871 (+25.0%) 47.3 1.042 (+21.3%)
FO F 1.07 55.0 0.828 46.3 0.697 41.9 0.860
M 1.287 (+22.6%) 56.4 0.956 (+23.0%) 48.9 0.846 (+27.6%) 46.8 1.049 (+27.5%)
FSS F 1.05 54.4 0.777 45.5 0.663 42.0 0.823
M 1.03 (+2.7%) 53.0 0.986 (+1.4%) 48.6 0.994 (+1.7%) 49.5 1.005 (+4.5%)
QI F 1.003 51.2 0.972 48.5 0.977 48.3 0.961
M 0.983 (+2.9%) 50.6 1.032 (+5.0%) 52.4 1.021 (+4.4%) 51.5 1.010 (+6.0%)
CI F 0.955 49.0 0.983 50.5 0.978 48.8 0.953
Analysis by role shows that, in terms of output, the average general production of
men is greater than that of women in all of the three roles considered: +13.3% for full
professors, +12.3% for associate professors and +17.5% for assistant professors. When
considering the qualitative dimension of scientific production, the performance
difference seems to increase the general average scientific strength (SS) of men is
19.7% greater than that of women among full professors, 15.9% greater among
associate professors and 20.2% greater among assistant professors.
Finally, the data also indicate a certain difference between the sexes relative to
contribution, with a spread of once again in favour of men. In terms of contribution
intensity (CI), the men’s figures are greater than that of women by 2.9% among full
professors, 5% among associate professors and 4.4% among assistant professors. The
data for average percentile rank of the two sexes can be superimposed on the
consideration of average performance. In particular, the percentile rank compresses the
difference, reducing the weight of particularly anomalous data such as those that may be
attributed to star scientists, but the overall higher performance of males still rests
unchanged for all the roles and indicators considered.
It can also be noted that the difference in average percentile rank between men and
women generally tends to decrease with increased stature of professional role. Looking
at output (O), the difference in percentile rank between genders is 4.8% among assistant
professors, 3% among associate professors, and only 1.5 per cent among full professors.
Scientometrics 79 (2009)
532
ABRAMO & AL.: Gender differences in research productivity
The same trend can be seen with other indexes of observation. However, for the role of
assistant professor, the difference in performance may have been amplified by not
having taken into account the probable occurrence of maternity leaves.
As an alternative means of examination, the difference in performance between men
and women was also calculated by applying the casual variables sequence criterion to
the entire active population. Beginning with the performance ranking of each male
scientist in his discipline sector, the distance between the ideal and effective cases was
measured:
eff
jM
jM
diff
jM RRR
max (1)
where:
maxjM
R = sum of the ranks of males in sector j under the hypothesis of maximum
differentiation*
eff
jM
R= sum of the ranks of males in sector j
* “maximum differentiation” is understood as the situation in which the highest
performing woman is still ranked below the lowest performing male
The value diff
jM
R therefore represents the “distance” for the ideal situation of
maximum performance difference between genders in favour of males. The same
calculation is completed for women, and through comparison between diff
jM
R and
diff
jF
R, it can be determined which of the two populations, male or female, obtains a
higher overall ranking. The simple sum of the data by sector provides the overall view
at the level of discipline area.
)( max
1
eff
jM
jM
n
j
diff
AM RRR
A
¦ (2)
where:
diff
AM
R = distanse from the situation of maximum differentiation for area A
nA = number of sectors included in area A
This analysis once again gives a comparison between diff
AM
R and diff
AF
R, indicating
which of the two populations, male or female, obtains a higher average overall ranking
at the level of discipline area (Tables 7, 8, 9). Although men do have an overall higher
performance than that of women (see the last row of each table), the contrary occurs in
some specific areas: for full and associate professors in the industrial and information
engineering area and for full professors and assistant professors in agriculture and
Scientometrics 79 (2009) 533
ABRAMO & AL.: Gender differences in research productivity
veterinary sciences. Female full professors result as having higher output than men in
the physical and chemical sciences areas, but for scientific strength in the latter area it is
males that obtain a higher performance. For the physical sciences, a similar inversion in
favour of males occurs when examining the fractional dimension of performance.
The case of agriculture and veterinary sciences is singular – women shine in the
roles of full professor and assistant professor, but they are unseated by men in the
intermediate role of associate professor. In earth sciences, women assistant professors
seem stronger than their male colleagues. With the advancement of professional role the
situation changes – among associate professors men achieve the higher performance
in 5 indicators out of 6; among full professors men score better in all 6 indexes.
Table 7. Analysis of the average position of male and female full professors
using the casual variables sequence criterion
Area O SS FO FSS QI CI
Industrial and information engineering F F F F F F
Agriculture and veterinary sciences F F F F M F
Biological sciences M M M M M M
Chemical sciences F M F M M M
Earth sciences M M M M M M
Physical sciences F F M M M M
Mathematics and information science M M M M M F
Medical sciences M M M M M M
Total M M M M M M
Table 8. Analysis of the average position of male and female associate professors
using the casual variables sequence criterion
Area O SS FO FSS QI CI
Industrial and information engineering M F F F F F
Agriculture and veterinary sciences M M M M M F
Biological sciences M M M M F M
Chemical sciences F F M F F M
Earth sciences M M M M F M
Physical sciences F F F F M F
Mathematics and information sciences M M M M M M
Medical sciences M M M M F M
Total M M M M F M
Table 9. Analysis of the average position of male and female assistant professors
using the casual variables sequence criterion
Area O SS FO FSS QI CI
Industrial and information engineering M M M M M M
Agriculture and veterinary sciences M F F F F F
Biological sciences M M M M M M
Chemical sciences M M M M M M
Earth sciences F F F F F M
Physical sciences M M M M F F
Mathematics and information science M M M M M M
Medical sciences M M M M F M
Total M M M M M M
Scientometrics 79 (2009)
534
ABRAMO & AL.: Gender differences in research productivity
Analysis at the level of single sectors
Moving down to the level of sectors, there are further interesting points for
consideration. Tables 10, 11 and 12 indicate, for each area, the number of sectors in
which the average percentile rank of women is not inferior to that of men. The tables
also indicate, in parentheses, the weight of each sector in terms of the total number of
professionals in the area. It should be noted that the comparison is only possible in the
sectors where both males and females hold professional roles. Since the representation
of women varies among professional roles there is also variation in the total number of
sectors per role in which a comparison between is possible: 110 sectors for full
professors, 146 for associate professors and 147 for assistant professors.
Considering the population of full professors first (Table 10), the average percentile
rank for output by women is not less than that of men in 43 sectors out of 110 (39.1%)
and 28.6% of the professors are employed in those sectors. Very similar indications are
obtained for other performance indexes, while important differentiations emerge at the
level of disciplinary area. For agriculture and veterinary sciences, still referring to the
role of full professor, women demonstrate performance not less than men in 14 sectors
out of 20 in terms of output, 11 out of 20 if considering scientific strength (SS) and 13
out of 20 considering fractional output (FO). The case of physical sciences also presents
an interesting situation. Women register productivity data that are not inferior to those
of men in four sectors out of five, in terms of output and scientific strength. However,
when the fractional dimension is considered, the difference between men and women is
inverted in the sector PHY/01 (experimental physics), which is the most populous
sector (38% of all scientists in the area fall within this sector). Still with regards to full
professors, it can be noted that industrial and information engineering, an area where
the representation of women is truly marginal (see Table 1: one female professor for
every 25 males), women register a performance which is not inferior to that of men in
six sectors out of 14 for output, and five out of 14 for scientific strength. Finally, in
medical sciences (an area which represents 31% of the entire field of investigation, in
terms of number of professionals), comparison indicates a performance by women
which is not inferior to that of men in 10 sectors of 28 for scientific strength and 11 of
28 for output. In the mathematical sciences area, the average percentile rank for males is
not less than that of women in 6 out 7 cases, for output and scientific impact. However,
in terms of contribution intensity, it can be seen that performance by women is not
lesser in 4 sectors out of 7.
Finally, an examination conducted on the data within the sectors reveals the absence
of correlation between the numbers of women and their general ranking with respect to
male colleagues in each sector.
Scientometrics 79 (2009) 535
ABRAMO & AL.: Gender differences in research productivity
Table 10. Number of technical-scientific sectors in which the average percentile rank
of female full professors is not inferior to that of males
(figures in brackets indicate weight of sectors in terms of percentage of total professionals in the area)
Area # SDS O SS FO FSS QI CI
Industrial and information
engineering 14 6
(33.8%)
5
(31.7%)
7
(36.9%)
5
(24.8%)
6
(24.7%)
7
(41.2%)
Agriculture and veterinary
sciences 20 14
(59.5%)
11
(48.6%)
13
(56.6%)
11
(50.8%)
8
(32.2%)
10
(43.4%)
Biological sciences 18 4
(10.4%)
6
(26.2%)
4
(12.5%)
5
(26.0%)
9
(57.0%)
4
(27.1%)
Chemical sciences 10 2
(23.1%)
3
(39.5%)
3
(45.0%)
2
(23.1%)
4
(42.3%)
5
(54.7%)
Earth sciences 8 1
(14.0%)
2
(22.3%)
1
(14.0%)
1
(14.0%)
3
(23.9%)
2
(25.0%)
Physical sciences 5 4
(70.9%)
4
(70.9%)
3
(32.9%)
3
(32.9%)
2
(22.0%)
1
(10.9%)
Mathematics and
information sciences 7 1
(4.0%)
1
(4.0%)
1
(4.0%)
1
(4.0%)
2
(20.9%)
4
(62.8%)
Medical sciences 28 11
(27.8%)
10
(22.2%)
10
(24.6%)
10
(22.2%)
10
(27.6%)
12
(24.0%)
Total 110
43
(28.6%)
42
(31.1%)
42
(27.9%)
38
(24.7%)
44
(33.7%)
45
(34.0%)
Table 11. Number of technical-scientific sectors in which the average percentile rank
of female associate professors is not inferior to that of males;
(figures in brackets indicate weight of sectors in terms of percentage of total professionals in the area)
Area # SDS O SS FO FSS QI CI
Industrial and information
engineering 24 13
(41.6%)
14
(44.6%)
14
(56.2%)
14
(54.1%)
14
(48.9%)
12
(52.7%)
Agriculture and veterinary
sciences 26 7
(16.6%)
7
(17.6%)
10
(27.0%)
8
(25.0%)
12
(39.6%)
15
(59.7%)
Biological sciences 19 5
(28.2%)
6
(30.4%)
4
(27.0%)
6
(30.4%)
9
(43.8%)
7
(49.4%)
Chemical sciences 11 4
(31.6%)
4
(31.6%)
4
(31.6%)
4
(31.6%)
4
(50.7%)
6
(39.9%)
Earth sciences 12 6
(41.0%)
6
(52.4%)
7
(46.6%)
5
(50.1%)
5
(45.9%)
6
(47.3%)
Physical sciences 8 4
(63.7%)
3
(52.8%)
4
(63.7%)
3
(52.8%)
2
(45.0%)
6
(74.9%)
Mathematics and
information sciences 9 2
(17.8%)
1
(13.8%)
1
(4.0%)
0
(–)
4
(36.3%)
2
(8.2%)
Medical sciences 37 14
(32.7%)
18
(46.9%)
10
(20.2%)
16
(43.9%)
24
(68.4%)
17
(42.9%)
Total 146
55
(33.4%)
59
(37.7%)
54
(30.9%)
56
(37.3%)
74
(52.2%)
71
(46.3%)
Comparative analysis for the data in Tables 10, 11 and 12 offers interesting
highlights concerning variability of performance differentials with respect to
professional role. The realities that emerge are not completely consistent from one
indicator to the next. For example, a review of output would seem to suggest that the
gap between men and women tends to lessen with increased professional status. For
assistant professors, the performance of women is not inferior to that of men in 49
sectors out of 147 (33.7% of cases), compared to 55 out of 146 (37.7%) for associate
professors, and as previously indicated, 43 out of 110 (39.1%) for full professors.
Scientometrics 79 (2009)
536
ABRAMO & AL.: Gender differences in research productivity
Table 12. Number of technical-scientific sectors in which the average percentile rank
of female assistant professors is not inferior to that of males;
(figures in brackets indicate weight of sectors in terms of percentage of total professionals in the area)
Area # SDS O SS FO FSS QI CI
Industrial and information
engineering 23 9
(22.4%)
10
(30.3%)
12
(29.0%)
11
(26.4%)
12
(35.5%)
12
(43.0%)
Agriculture and veterinary
sciences 27 12
(44.5%)
13
(49.5%)
15
(55.4%)
13
(44.7%)
16
(59.2%)
16
(52.9%)
Biological sciences 19 6
(14.8%)
7
(19.6%)
6
(28.2%)
5
(25.7%)
5
(15.5%)
6
(37.8%)
Chemical sciences 11 3
(15.3%)
4
(16.1%)
4
(20.4%)
4
(20.4%)
7
(42.1%)
3
(26.4%)
Earth sciences 11 4
(38.6%)
5
(50.4%)
4
(35.1%)
4
(41.4%)
6
(64.2%)
4
(35.1%)
Physical sciences 7 3
(19.9%)
3
(19.9%)
3
(19.9%)
3
(19.9%)
2
(45.0%)
4
(69.3%)
Mathematics and
information sciences 8 1
(16.9%)
2
(30.7%)
1
(16.9%)
1
(16.9%)
2
(30.7%)
3
(53.0%)
Medical sciences 41 11
(16.2%)
14
(35.9%)
9
(14.1%)
11
(17.1%)
23
(55.6%)
14
(30.9%)
Total 147
49
(19.7%)
58
(29.4%)
54
(23.5%)
52
(23.1%)
73
(42.2%)
62
(39.6%)
However, when the qualitative dimension is considered, there is no evidence that the
gap varies with role. For example, in scientific strength, the performance of women
remains quite uniform it is not inferior to that of men in 58 sectors out of 147 for
assistant professors (39.5% of cases), in 55 sectors out of 146 for associate professors
(37.7%) and in 42 out of 100 sectors for full professors (38.2%). Contribution intensity
also seems to flatten the variability by role in performance gap between men and
women. In general, it is in the population of associate professors that the higher
performance of males is challenged in the greatest number of sectors. With career
progress, the sectorial gaps between the sexes vary greatly from discipline to discipline.
For example, in earth sciences, the number of sectors in which performance by males
registers not less than that of women actually seems to increase with career progression.
This occurs in almost every sector for full professors (seven out of eight for output, and
six out of eight for scientific strength), but only in 6 out of 12 sectors for associate
professors and 6 to 7 out of 11 for assistant professors.
Final considerations
The analysis of productivity differences between men and women employed in
research has always attracted interest among science sociologists, whose studies agree
in acknowledging a higher performance among men than women. The study reported
here, analyzing the technological-scientific disciplines of the entire Italian academic
system, confirms the existing literature but also brings to light significant differences in
the distribution of performance between the sexes. Males do demonstrate a higher
average productivity with respect to that of females for all the performance indicators
Scientometrics 79 (2009) 537
ABRAMO & AL.: Gender differences in research productivity
considered. However one of the new and interesting facts is that it is above all in the
quantitative dimension of output where the major gap is recorded. In terms of quality
index and contribution intensity, the gap between the sexes, though still present, seems
less pronounced. The performance gap also seems to reduce with career advancement.
This could in part be due to the effect of maternity, it being reasonable that the
experience of motherhood would be more frequent, for age reasons, among the lesser
university career roles. This result seems to coincide with the conclusions by STACK
[2004], whose work suggests that women with preschool aged children publish less than
others. In effect, the average age of female research professionals in the Italian
academic system for the period under observation is 43 years, falling within the final
family life phase for the presence of very young children.
Although this study does not permit proceeding to inter-temporal comparisons, the
average gap revealed, while still significant, is notably reduced compared to results
reported by other authors. This lends value to the increasingly common thesis of a
progressive reduction over time for the performance gap between the sexes, as proposed
by COLE & ZUCKERMAN [1984], XIE & SHAUMAN [1998] and LEAHEY [2006].
Adding to current literature, the study reported also highlights that there are
important sectorial specificities in the differences between the sexes, but these generally
do not raise any challenge to the higher performance of men in all dimensions of the
evaluation of scientific performance. But if it is true that the average performance of
men is higher than that of women, this is not the case in all sectors of research
professionals. In terms of output by full professors, for 43 sectors out of 100 women do
not demonstrate any lesser performance than that of men. For associate professors this
occurs in 55 sectors out of 146 and for assistant professors in 49 out of 147. Further,
certain areas result as particularly interesting for interpretation of the gender gap. In
industrial and information engineering for example, the feminine presence is truly
marginal, representing 10% of scientists, and among full professors only 4% of the
total. Yet women in this area demonstrate performance not less than their male
colleagues in just less than half of the sectors. This could suggest further questions
revolving around the hypothesis of discrimination between sexes in this area in
particular and in the entire academic system in general. In partial contrast to tendencies
in the literature, certain results emerge concerning non-productive professionals. For
each role, the percentage of non-productive males is higher than that for females, while
the reverse is true in the overall population, though with a minimal difference. However
women show a higher concentration than men in the lowest levels of productivity. The
contrary situation registers for the highest levels of performance. It is therefore possible
that males are characterized by a higher concentration of “star scientists”, and this, with
all probability, would play a significant role in the generally higher performance of men
than women. The authors will attempt to respond to this suggestion in future work.
Scientometrics 79 (2009)
538
ABRAMO & AL.: Gender differences in research productivity
*
Authors are deeply indebted to Giorgia Barbetta for her invaluable support in data processing.
References
ABRAMO, G., D’ANGELO, C. A. (2007), Measuring science: Irresistible temptations, easy shortcuts and
dangerous consequences. Current Science, 6 : 762–766.
ABRAMO, G., D’ANGELO, C. A., PUGINI, F. (2008), The measurement of Italian universities’ research
productivity by a non parametric-bibliometric methodology. Scientometrics, 76 : 2.
BLACKBURN, R. T., BEHYMER, C. E., HALL, D. E. (1978), Research notes: Correlates of faculty publication.
Sociology of Education, 51 : 132–141.
BONACCORSI, A., DARAIO, C. (2003), Age effects in scientific productivity. The case of the Italian National
Research Council (CNR), Scientometrics, 58 (1) : 49–90.
CIVR (2006), Comitato di indirizzo per la Valutazione della Ricerca, Valutazione Triennale della Ricerca
Italiana: VTR 2001-2003, http://vtr2006.cineca.it/.
COLE, J. R., ZUCKERMAN, H. (1984), The productivity puzzle: persistence and change in patterns in
publication of men and women scientists. Advances in Motivation and Achievement, 2 : 217–258.
DICKSON, V. A. (1983), The determinants of publication rates of faculty members at a canadian university.
Canadian Journal of Higher Education, 13 (2) : 41–49.
FOX, M. F. (1983), Pubblication productivity among scientists: a critical review. Social Studies of Science, 13
(2) : 285–305.
FOX, M. F. (2005), Gender, family characteristics, and publication productivity among scientists. Social
Studies of Science, 35 (1) : 131–150.
GANDER, J. P. (1999), Faculty gender effects on academic research and teaching. Research in Higher
Education, 40 (2) : 171–184.
HYDE, J. S., LINN, M. C. (1988), Gender differences in verbal ability: A meta-analysis, Psychological
Bulletin, 104 : 53–69.
HYDE, J. S., FENNEMA, E., LAMON, S. J. (1990), Gender differences in mathematics performance: A meta-
analysis, Psychological Bulletin, 107 : 139–155.
KYVIK, S. (1990), Motherhood and scientific productivity. Social Studies of Science, 20 : 149–160.
LEAHEY, E. (2006), Gender differences in productivity: research specialization as a missing link. Gender and
Society, 20 (6) : 754–780.
LEE, S., BOZEMAN, B. (2005), The Impact of Research Collaboration on Scientific Productivity, Social
Studies of Science, 35 (5) : 673–702.
LEMOINE, W. (1992), Productivity patterns of men and women scientists in Venezuela, Scientometrics,
24 (2) : 281–295.
LINN, M. C., PETERSON, A. C. (1985), Emergence and characterisation of sex differences in spatial ability:
A meta-analysis, Child Development, 56 : 1479–1498.
LONG, J. S. (1987), Problems and Prospects for Research on Sex Differences in the Scientific Career.
In: L. S. DIX (Ed.), Women: Their Underrepresentation and Career Differentials in Science and
Engineering, National Academy Press. 157–169.
MOED, H. F. (2002), The impact factors debate: the ISI’s uses and limits. Nature, 415, 731-732.
MORAVCSIK, M. J. (1985), Applied scientometrics: an assessment methodology for developing countries.
Scientometrics, 7 (3–6) : 165–176.
PRIPIû, K. (2002),Gender and productivity differentials in science. Scientometrics, 55 (1) : 27–58.
RAMSDEN, P. (1994), Describing and explaining research productivity. Higher Education, 28 : 207–226.
STACK, S. (2004), Gender, Children and research Productivity. Research in Higher Education, 45 (8)
891–920.
UE, (2006), Women and science-statistic indicators, SHE FIGURES 2006
Scientometrics 79 (2009) 539
ABRAMO & AL.: Gender differences in research productivity
VOYER, D., VOYER, S., BRYDEN, M. P. (1995), Magnitude of sex differences in spatial ability: A meta-
analysis and consideration of critical variables, Psychological Bulletin, 117 : 250–270.
WEINGART, P. (2004), Impact of bibliometrics upon the science system: inadvertent consequences? In:
H. F. MOED, W. GLÄNZEL, U. SCHMOCH (Eds), Handbook on Quantitative Science and Technology
Research. Dordrecht (The Netherlands), Kluwer Academic Publishers.
XIE, Y., SHAUMAN, K. A. (1998), Sex differences in research productivity: new evidence about an old puzzle.
American Sociological Review, 63 : 847–870.
XIE, Y., SHAUMAN, K. A. (2003), Women in Science: Career Processes and Outcomes, Harvard University
Press, Cambridge, Massachusetts, and London, England.
XIE, Y., SHAUMAN, K. A. (2004), Women in science: career processes and outcomes (review), Social Forces,
82 (4) : 1669–1671.
ZAINAB, A. N. (1999), Personal, academic and departmental correlates of research productivity: a review of
literature. Malaysian Journal of Library & Information Science, 4 (2) : 73–110.
