ART: Phelan & Cole, International Differences in Scientific Productivity

Gretchen Whitney gwhitney at UTKUX.UTCC.UTK.EDU
Tue Jan 11 18:21:59 EST 2000

Explaining International Differences in Scientific Productivity

Tom Phelan and Stephen Cole discuss their research on national variations
in scientific productivity.

        By any reasonable measure, it is clear that a relatively small
group of wealthy nations produce an extraordinarily large proportion of
world scientific research.  Most experts agree that national wealth and
the size of the scientific labour force are two primary influences on
national levels of scientific production. These variables, however, do not
explain why rates of scientific production vary greatly among nations with
similar levels of wealth or why individuals are more prone to choosing
scientific careers in some nations than in others.

Theories of Scientific Productivity

        Several theories have been proposed that are relevant to
understanding national scientific differences.  One influential theory
suggests that some cultures place a higher value on scientific activity
than others.  More specifically, this theory suggests that in Protestant
dominated societies, individuals are more likely to choose scientific
careers than in Catholic ones.
        A second theory is that the structural organisation of a national
university system will influence the amount of science produced.  This
theory suggests that a very competitive and decentralised university
system will be more productive than a highly centralised one because it
will produce more job opportunities in science.
        Both the cultural theory and the structural theory take for
granted that there is a proportional relationship between scientific
productivity and the number of scientists a nation employs.  This is to
suggest that each additional scientist employed will, on average, increase
scientific knowledge by the same amount.  In contrast to this view, a
third theory suggests that the pool of talented individuals will gradually
deplete, and an increasingly larger number of scientists must be employed
to produce equal increments of scientific advance.  This issue is
important as it bears on the amount of return nations are likely to gain
from increasing their investments in science.

Limits to Scientific Talent

        Earlier work published in Science by Stephen and Jonathan Cole
supported the hypothesis that only a relatively small number of scientists
produced the vast majority of published research and these individuals
tended to cite each other. This being the case, then why bother educating
large numbers of scientists at great expense when most of the worldÕs
science is produced only by a relatively small number of individuals who
draw on each others work?
        In later research, however, Stephen Cole and Gary Meyer examined
cohorts of physicists hired in the United States. The number of physicists
hired varies dramatically over time due changes in funding levels.  If
productivity diminishes as numbers of scientists increase because of the
depletion of the pool of talented individuals, then during time periods
with fewer jobs, the average quality of the cohort ought to be greater
than during years when it was comparatively easy to be hired as a
physicist.  In fact, however, in terms of numbers of citations received,
regardless of how many physicists were employed, the average quality of
the work of members of each cohort was about the same.  Cole and Meyer
were able to show that as the number of positions in physics was reduced,
the attractiveness of the job declined and thus candidates with high
ability as well as candidates with lesser ability were discouraged from
entering the field. This suggests that expanding the size of the
scientific community remains an efficient strategy for a nation to pursue
to increase scientific output.

National Contributions to Science

        Clearly, there was a need to further explore these issues, but
first it was necessary to determine a reasonable method of measuring
national contributions to science.   Simply counting publications in
scholarly journals and attributing these to specific nations would not be
adequate.  Many scientists would argue that much published science is of
little importance.

        As an alternative, numbers of highly cited articles were counted
and credited to the nations in which their authors were affiliated.   Most
scientific articles receive at most one or two citations and many
scientific publications are never cited at all.  We used a cutoff of 40
citations to identify an important article.  An article receiving this
many citations could reasonably be said to have attracted considerable
attention within the scientific community.  The United States, the United
Kingdom, Germany, France and Japan were found to produce the largest total
number of highly cited articles.  On a per capita basis Switzerland,
Sweden and Israel were also very big producers of ÔqualityÕ science. These
data leave little doubt that modern science is dominated by a set of
wealthy countries.

The impact of national wealth

        It is evident from the data presented in figure 1 that the amount
of high quality science a nation produces is strongly related to the
number of research scientists in that nation.  Figure 2 shows that the
wealth of a nation strongly predicts the number of research scientists it
employs.  There were a number of exceptions.  Countries with large numbers
of research scientists such as India, China, Egypt, Romania and Nigeria
nevertheless produced only a small amount of highly cited research.
Comparatively low levels of national wealth relative to population could
well explain the difficulty some countries have in producing science that
attracts international attention, although a focus on narrower regional
concerns or a bias by editors or citing scientists in favour of ÔWesternÕ
science could also play a role in determining these results.

Increasing numbers of Scientists

        To address the issue of whether there is a decline in marginal
productivity as greater numbers of scientists are employed within a
nation, the correlation between the number of research scientists per
capita and the number of high quality papers per scientist was calculated.
If hiring more scientists led to a decrease in the average quality of
scientists, this correlation should have been negative, but in fact it was
moderately positive.  This suggests that nations tend to have a adequate
reserve of potential scientific talent among their citizens and they need
not worry that they may reach the Ôbottom of the barrelÕ and lower the
quality of science if they expand programs to train more scientists.

Factors affecting production

While the size of a nationÕs economy is strongly related to the amount of
highly cited science produced and to the number of scientists a nation
employs, this correlation is less than perfect.  Israel, for example, was
found to produce more than five times the amount of high quality research
expected considering the size of its economy.
On the other hand, Japan and Italy were found to produce about 40% less
science than is typical for countries of their economic size.  This means
that factors other than national wealth such as social and cultural
differences may partly explain differential rates of scientific production
among nations.

Science Funding

        There is a strong correlation between the amount of funding per
capita that a nation spends on basic scientific research and the number of
highly-cited papers produced.  There are two different mechanisms through
which this money can be converted into science, both of which could be
operating.  First, spending money on research could serve to make science
a more attractive occupation Ð both by increasing the number of positions
available for scientists and by paying them well.
Second, money spent on science can also be devoted to increasing the
productivity of scientists.
        The amount of non-salary funding per scientist, however, was found
to be unrelated to the average number of papers produced per scientist.
Nor was it significantly related to the production of highly-cited
articles.  But how could this be true Ð could non-salary funding levels in
science really be irrelevant?  Upon closer examination, what appears to
occur is that nations with low levels of non-salary funding tend to focus
on the production of low-cost science such as mathematics and theoretical
physics while nations funding science at a higher level, produce more
research in expensive areas such as experimental physics.  The amount of
non-salary funding appears to influence the foci of attention of
researchers more than it alters the amount of new knowledge produced.

Organisation of Science

        A famous historical study by Joseph Ben-David attributed the rise
of German science to the structure of the German university system.  In
the mid nineteenth century, Germany had a relatively large number of
universities that competed for scientists and attracted many individuals
to scientific careers.  Might a strongly competitive university structure
still be an important influence on scientific productivity today?

        It was found that a nation with many scientific centres relative
to the number of science graduates in the country tends to have a higher
proportion of its population in scientific careers than do nations with a
lower density of tertiary institutions.  This finding suggests that
competitive national scientific systems offer more opportunities for
scientific employment, and thus attract more people to scientific careers.

Cultural Influences

        In his famous book, The Protestant Ethic and the Spirit of
Capitalism, Max Weber suggested that Protestant students were more likely
than Catholics to study science.  The well known theorist Robert Merton in
a study of science in 17th century England extended this idea by
suggesting that Protestant theology promotes values that encourage
scientific inquiry.
Examining wealthy industrial countries it is possible to confirm that the
higher the percentage of Catholics in the population, the lower tended to
be the production of science even while controlling for national wealth.
The positive effect of a nation having a highly competitive university
structure, however, was stronger than the effect of religious composition
on levels of scientific output.


        It can be concluded that at the end of the twentieth century, the
total amount of research that a nation produces is strongly influenced by
its wealth.  Wealth works by providing opportunities for talented
individuals to enter scientific careers.  The total number of research
scientists almost completely explains a countryÕs scientific output.  It
is reasonable to suggest that national wealth is a prerequisite for making
substantial contributions to the growth of scientific knowledge. One might
also argue that the relationship is likely reciprocal, and that a strong
effort in science might itself contribute to greater national wealth.
Still, it is clear that other factors are at work as well. University
funding levels appear to influence both the numbers of people choosing
scientific careers as well as the intellectual focus of those individuals
who become scientists.  Finally, it is worth noting that other cultural
and social factors such as dominant national religion and a more
competitive university system also appear to influence national levels of
scientific productivity.


Tom Phelan is a research fellow at the Institute of Advanced Studies at
the Australian National University.  Stephen Cole is a Professor of
Sociology at the State University of New York at Stony Brook.  The
information provided here is based on a article by Cole and Phelan, "The
Scientific Productivity of Nations" published in the Spring 1999 edition
of the journal Minerva.

Reprinted with permission.

Gretchen Whitney, PhD                                     tel 423.974.7919
School of Information Sciences                            fax 423.974.4967
University of Tennessee, Knoxville TN 37996 USA           gwhitney at

More information about the SIGMETRICS mailing list