Where demographics meets scientometrics: towards a dynamic career analysis

Where demographics meets scientometrics:towards a dynamic career analysis

Lin Zhang • Wolfgang Glanzel

Received: 6 December 2011 / Published online: 22 January 2012� Akademiai Kiado, Budapest, Hungary 2012

Abstract In an earlier exercise some demographic methods were reformulated for

application in a scientometric context. Age-pyramids based on annual publication output

and citation impact was supplemented by the change of the mean age of the publications in

the h-core at any time. Although the method was introduced to shed some demographic–

scientometric light on the career of individual researchers, the second component, i.e., the

age dynamics of the h-core can however be applied to higher levels of aggregation as well.

However, the found paradigmatic shapes and patterns do not only characterise individual

careers and positions, but are also typical of life cycles and subject-specific peculiarities. In

the present study, the proposed approach is used to visualise the careers of scientists active

in different fields of the sciences and social sciences and notably the second component,

the h-core dynamics, is extended to the analysis of scientific journals from the same fields.

In addition to the dynamics of productivity and citation impact, the evolution of

co-authorship patterns of the same scientists is studied to capture another facet of indi-

vidual academic careers.

Keywords Demographics � Career analysis � Age-pyramids � H-core dynamics �Co-authorship

Introduction

Since the h-index has been introduced by Hirsch in 2005, the analysis of scientists’

individual careers gained a new impetus. Scientists in different disciplines were very soon

L. Zhang � W. Glanzel (&)Centre for R&D Monitoring (ECOOM) and Department of MSI, K.U. Leuven, Leuven, Belgiume-mail: [email protected]

L. ZhangDepartment of Management and Economics, North China University of Water Conservancy andElectric Power, Zhengzhou, China

W. GlanzelInstitute for Research Policy Studies (IRPS), Hungarian Academy of Sciences, Budapest, Hungary

123

Scientometrics (2012) 91:617–630DOI 10.1007/s11192-011-0590-8

ranked according to their h-index. Although such exercises proved hazardous by many

reasons (cf. Glanzel 2006), the presentation of the h-index of scientists, considered

‘leading’ in their fields, remained quite popular (e.g., Glanzel and Persson 2005; Egghe

2006; Bar-Ilan 2006, 2010; Cronin and Meho 2006, 2007; Levitt and Thelwall 2009) in the

field of information science and scientometrics. However, high h-index values might be

considered kind of confirmation of the supposed prominent position the researcher holds in

the community. Besides its well-known subject-dependence (cf. Batista et al. 2006), the

h-index is sensitive to the scientist’s academic age as well. In order to compensate for this

effect, it was suggested to normalise the individual’s h-index by the respective career

length (e.g., Jensen et al. 2009). On the other hand, such normalisation eliminates

important aspects that are otherwise captured by the h-index. Subject-specific peculiarities

and characteristics of career stages are thus lost if the measure is normalised this way. The

changes in the h-index and the h-core along with the age structure of publications and

citations allow a deeper insight into an individual’s career and might reflect breaks, a

caesura or shift in the scientist’s academic life.

In a previous note (Glanzel and Zhang 2010), we have monitored the changes in the age

structure of the h-core of individual scientists by introducing a new measure. In addition,

we have used an analogon of the well-known age pyramid presentation of the age distri-

bution in a human population, which was re-defined for visualising the change of publi-

cation activity and citation impact of individuals over time. Changes in the number of

different co-authors along the corresponding periods raise another interesting question in

the context of individual career analysis, namely, in how far the achievements are brought

about in changing environment, by mobility and changing collaboration in general.

Methods and results

In some recent studies, the career of scientists in the field of information science has been

analysed in the light of the dynamics of their productivity and citation impact. In particular,

the relationship between creativity and both chronological and professional age in infor-

mation science has been explored by Cronin and Meho (2007). The authors have used

high-impact work and cumulative citation rates to capture the shape of a scientist’s career.

In a more recent study by Levitt and Thelwall (2009), it has been shown that high citation

impact of individual papers in information and library science is not always reflected by a

high h-index of their first authors.

In the present study we will apply the tools developed in the previous study (Glanzel

and Zhang 2010) to two levels of aggregation. The proposed approach is used to visualise

the careers of (1) individual scientists active in different fields of the sciences and social

sciences, and is extended to the analysis of (2) scientific journals from the same fields. We

will analyse ‘career-specific’ patterns of publication activity, citation impact and co-

authorship dynamics along the following research questions.

1. In how far does citation impact mirror the evolution of publication activity?

2. Is there a phase shifting between publication and reception of results?

3. Is there any cumulative effect of impact independently of publication activity?

4. In how far is the influence of ‘hot papers’ or ‘hot topics’ measurable by the age

distribution and the changing age of the h-core?

5. How is the rate of different co-authors over publication records changing along with

the individual career life?

618 L. Zhang, W. Glanzel

123

In order to answer these questions we will apply three approaches. The first approach is

based on a well-known visualisation tool in demographics. It is designed to measure and to

compare publication and citation life-time for a selected individual. Before we apply this

tool, we shortly recall the basic patterns according to typical population pyramids obtained

in demographics. The second tool is based on the calculation of the arithmetic mean age of

publications of the h-core sequence, and the third one is the ratio of the number of differentco-authors over publication output in the corresponding periods. These indicators will be

explained in detail later.

The age pyramids

In demographics, the population pyramid is an elementary tool to reflect the age structure

and the growth characteristics of a given population. In an age pyramid or age structure

diagram, the age distribution in a human population is shown in a double bar diagram,

where the various male age groups are plotted against the corresponding female groups.

Demographers distinguish about 5–7 paradigmatic shapes reflecting different types of

expanding, stationary and contracting population models. From the mathematical view-

point, one can distinguish simple linear, convex and concave shapes as well as more

complex shapes with and without inflection point. Most known shapes in demographic

analysis of human populations are the triangle (reflecting steady growth with high fertility

and high mortality in all age groups), the pagoda shape (with very high fertility and high

infant mortality), the bell shape (typical of the baby boom in the industrial countries after

World War II), beehive shape (reflects a stationary structure, provided infant mortality is

low) and the ‘‘onion’’ shaped (reflecting superannuation of the population).

Here we have to stop for moment since the above characteristics refer to ‘real’

populations. The adoption of demographic model in an informetric context requires some

re-interpretations. While the notion of fertility can still be interpreted as the current pub-

lication activity, mortality does not exist in this context since papers, once published, and

citations, once received, will not disappear from the system any more. And life expectation

can at the best be interpreted in terms of obsolescence as reflected, for instance, by the life-

time distribution of citations (cf. Glanzel and Schoepflin 1994). The concept of mortality

should therefore be renounced when this model is applied to informetrics. This holds,

above all, for the interpretation of the triangle and pagoda shape. Both patterns just express

higher activity in recent years; by contrast, the beehive shape reflects stable publication

activity over time, while the onion or urn shape indicates decreasing activity of the author

in the recent years. The age profile of citation impact is subject to a very special effect.

Papers might still be cited even when a scientist is not active any more, the right-hand side

of the diagram tends to be rather triangle- or pagoda-shaped but beehive or onion shape

might occur as well. Furthermore, citation impact is expected to have a non-degenerate age

distribution when an author has become inactive and publication activity is flatlined.

At this point we have to stress that one has to distinguish between the case of indi-

viduals with their specific life cycles and higher aggregates like journals, topics or insti-

tutes, where individual life cycles do not play an important part and ‘terminators’ (cf. Price

and Gursey 1976) are continuously replaced by new authors and, therefore, both activity

and impact distributions are not mainly determined by the individual’s life cycle.

For the following exercise we use anonymous samples, where we have selected four

individual authors from Thomson Reuters’ Web of Science representing a group of sci-

entists with about 25 or more years of professional experience in four different subject

areas, one each representing the life sciences, natural sciences, mathematics (statistics &

Where demographics meets scientometrics 619

123

probability) and the social sciences. The selected authors are, on one hand, not repre-sentative for their field in the sense that they would represent the standard of their field;

they have a quite long career with large publication output and higher than average citation

impact. On the other hand, they are indeed representative for their field in the sense that

their communication behaviour, i.e., their publication and citation behaviour reflects the

subject-specific peculiarities of their discipline.

The pyramids are presented in Fig. 1. In order to facilitate visualisation and interpre-

tation, we have grouped publication and citation counts by three-year periods. This helps

avoid fluctuations and avoids the occurrence of periods of relative inactivity as well. We

plot the distributions of papers according to their age at the left-hand side of the diagram,

that of citations at the right-hand side. Furthermore, we have rescaled citations by factor

25.

Some typical patterns can immediately be recognised in Fig. 1. The first peculiarity, that

strikes the eye, concerns the asymmetry of publication and citation patterns. While the

publication-age distribution of the first three authors is of beehive/onion type, the fourth

author represents a triangle type. While the second and third author are already active for

35–45 years and their recently decreasing activity seems to be plausible, the onion shape of

the first author is somewhat surprising. The triangle shape of the fourth author is not an

exception to the rule. Besides the individual peculiarities also the influence of the subject

field is visible. The shape of the distribution in the natural sciences and in mathematics is

more stretched and the absolute frequencies are lower than in the life sciences and the

social sciences. This effect is even more obvious if the age of citations is considered. The

gap between the citation impact in the natural sciences, notably in mathematics, on one

hand, and the life sciences, on the other hand, is large. The triangle and pagoda type can be

found in the first and the fourth diagram, respectively. While the beehive in the second case

Fig. 1 Scientometrics age pyramid for four scientists (top left life sciences, top right natural sciences,bottom left mathematics, bottom right social sciences) [Source Thomson Reuters Web of Knowledge(formerly referred to as ISI Web of Science)]


123

and the onion in the third one by and large mirror the corresponding shapes of publication

age, the onion shape of publication age is contrasted by citation triangle in the first case.

This might have two reasons, particularly a phase shift between publication and reception

of results, or some recent hot topics in the work of the author in question. The latter case

can be observed for the fourth author in the social sciences. Also the author in the natural

sciences is characterised by contrary trends of publication and citation age in the most

recent periods. The trend might point to a certain citation delay in this case.

Further examples of triangle/pagoda or beehive/onion shapes of the scientometric age

pyramid for individual authors in the field of information science can be found in the piece

by Glanzel and Zhang (2010).

The age pyramid of scientific journals will structurally differ from the previous one as

has pointed to above. The distributions of publication and citation age will not reflect

peculiarities of individual life cycles. However, the same basic patterns can be observed for

journals too. In order to illustrate this we have chosen four journals from the same

domains, namely Journal of Infection for the life sciences, Superconductor Science &Technology for the natural sciences, Probability Theory and Related Fields for mathe-

matics (statistics & probability) and Scientometrics for the social sciences. The results are

presented in Fig. 2. Journal of Infection is a dynamically growing journal with linear

shapes of the age distribution of both publications and citations. The evolution of

Scientometrics is even more dynamic; both distributions have pagoda shape. This is cer-

tainly a result of the sharp rise of quantitative science and technology studies having taken

during the last three decades. The interpretation of the pyramid of the journal Supercon-

ductor Science & Technology is not so easy. The frequency of younger citations increases

linearly but the publication age has rather a beehive, almost an onion shape. The reasons

for that are not clear. The publication age distribution of Probability Theory and Related

Fig. 2 Scientometrics age pyramid for four journals (top left Journal of Infection, top right SuperconductorScience & Technology, bottom left Probability Theory and Related Fields, bottom right Scientometrics)[Source Thomson Reuters Web of Knowledge (formerly referred to as ISI Web of Science)]


123

Fields reflects superannuation; however the shape of the citation age is linear. This sub-

stantiates that increasingly recent citations are predominant. There is indeed cumulative

effect of impact independently of publication activity, and citation impact may not mirror

the evolution of publication activity in the same period of career. On the other hand, the

linearly increasing impact also shows the strong position the journal holds among the

theoretical journals in the field of statistics and probability.

The h-core dynamics

The Hirsch core, or h-core, is formed by those papers that have received at least h citations,

where h denote the actual value of an h-index (Jin et al. 2007). Liang (2006) proposed the

h-index sequence in order to measure the dynamics of the h-index in a scientists career.

She defined the h-index sequence hk as the h-index of the papers published by the author in

question in the time interval [n - k ? 1, n], where n is the most recent year. Citations are

counted for the same period. However, unlike Liang’s approach aiming at constructing the

h-sequences in a retrospective fashion, we plotted the evolution of an author’s or a jour-

nal’s h-index through his/her or its career by calculating the h-index from the beginning

year of their career, that is, in a prospective manner. In other words we first calculated the

h-index for papers published in the first year of their career, then the first two years, the first

three years, and so on until the most recent year is reached. This can be done even if the

author’s productive career has come to an end before. This method of calculation is also in

line with Burrell’s study (Burrell 2007) of time-dependence of the h-index using stochastic

models.

Then a h-core sequence is defined analogously to the h-core for the h-index sequence.

Here we include the most recent ones in the h-core if there are several publications with

exactly h citations. The calculation of the age of publications in each h-core is also based

on the three-year sub-periods. The age of papers in each h-core is equivalent to the

difference between the ‘‘current unit’’ (i.e., the period for the h-core in question) and the

time unit of publication. For instance, a publication in unit 1 has an age of 2 if it appears in

the h-core of unit 3. Here we define the ‘‘time zero’’ for the author in question as the time

unit when the author’s first publication appeared in the WoS.

The arithmetic mean age of publications of this h-core sequence is calculated, which

expresses whether the more recent or the older publications are predominant in the

respective h-core. The obtained patterns reflect different aspects of changing impact from

the pyramid approach. Also for this indicator, we can find four paradigmatic patterns.

1. A linear shape of the mean age of the h-cores plotted against time reflects steady

growth of the age of most cited publications.

2. A convex shape reflects accelerated growing age of the most cited papers. This means

that the ‘‘top’’ papers were rather published in earlier stages of the scientist’s career.

3. A concave shape reflects decreasing age of highly cited papers, that is, recent papers

by the author are the more cited ones.

4. ‘‘Indefinite’’ shape. This covers all cases not listed above.

Case 1 can be considered a standard situation. It can be expected that a paper remains in

the h-core once it has already received a sufficient number of citations. If an author

becomes less active or inactive, the age of the h-core will disproportionally increase after a

while. In extreme cases this might result in a convex shape. If, however, an increasing

number of recent papers enter the h-core, the age curve turns concave. For instance, a new


123

emerging topic or a ‘hot paper’ and its follow-ups might cause this phenomenon (cf.

Glanzel and Zhang 2010).

The mean age sequences of the h-core for the four selected authors are presented in

Fig. 3. The most striking common feature is probably the subject dependence of the mean

age. The mean age of the h-core for the author in the life sciences does even not exceed

4 years in the most recent year. By contrast, the mean age in the natural sciences and

mathematics with 7–8 years in 2010 is pronouncedly higher. However, this might also

partially be caused by the different career lengths of these scientists, where the scientist in

life science has a relatively shorter career compared to the scientists in the natural sciences

and mathematics. The fourth case is most interesting because of the shape. While the

shapes in the natural sciences and mathematics are nearly linear, this one turns from a

linear to a concave graph. The reason is a ‘hot topic’ on which the author published in the

second half in the last decade. The first case is also interesting as it basically experienced a

‘‘concave shape’’ around the middle phase of the career and then turned to a linear graph.

When taking a deeper look into the ‘‘concave’’ phase of this author, we found relatively

many highly cited publications appearing during the period 1996–2001. For instance, 5

new papers entering the h-core in 1998 (with 13 publications in total) were published in

the same period 1996–1998 and it was striking that in 2001, among the 21 publications in

the h-core, 15 papers have been published in 1996–2001. This gives an explanation of the

temporary concave graph of the h-core age sequence. However, there have been much less

‘‘recent’’ new-comers into the h-cores afterwards, thus a linear graph is observed since

2001. It is also interesting that most of the highly-cited publications appearing in the

‘‘concave’’ phase (1996–2001) have always remained in the subsequent h-cores, and

several of them are among the most cited publications of the author. On the other hand,

there are still new members from the ‘‘concave’’ period entering the most recent h-cores.

Fig. 3 Mean age sequence of the h-core for four scientists (top left life sciences, top right natural sciences,bottom left mathematics, bottom right social sciences) [Source Thomson Reuters Web of Knowledge(formerly referred to as ISI Web of Science)]


123

By 2010, 26 papers from 1996–2001 exist among the 40 publications in h-core, which

means roughly two-thirds of the most important publications of the author were from the

‘‘concave’’ period. It could be concluded that the author had published in ‘‘hot topics’’

during the corresponding period and apparently these topics remained ‘‘hot’’ or important

even after a decade. Combined with the age pyramid analysis, the most productive phase of

the author was found in the period 1999–2001. Though the author is still active in his field,

we may conclude that the period of ‘‘concavity’’ might be the one with highest impact in

his publication career. To confirm this conjecture, we further have a look at the h-index

sequence of the author (Fig. 4), not surprisingly, there was indeed an accelerated

increasing trend of h-index in the period of 1996–2001. The concave turn of the mean age

graph of the forth author is even more striking.

In order to compare the career evolution of these two authors, the h-index sequence of

the latter author is also presented in Fig. 4. The h-index is accelerated increasing in the

most recent periods, corresponding to the ‘‘concave’’ phase of his h-core age, and on the

other hand, is also in line with his most productive and most cited period. Back to our

research question, the influence of ‘‘hot papers’’ or ‘‘hot topics’’ are indeed measurable by

the age distribution and the change of the age of the h-core in time and the ‘‘hot papers’’

also mark some important points in the author’s career.

The idea of using h-indexes for scientific journals was introduced as early as in 2005 (cf.

Braun et al. 2005, 2006). Although it was shown that there is a strong correlation between

the h-index and hybrid measures calculated on the basis of the number of publications in a

journal and their citation impact (Schubert and Glanzel 2007), through its ‘‘size depen-

dence’’ the journal h-index reflects somewhat different aspects than other journal impact

measures, notably the ISI Impact Factor. And these aspects are indeed closer to what we

can consider as or characterise by the expressions ‘‘career’’, ‘‘demography’’ or ‘‘life time’’.

The mean age sequence of the h-core for the four journals can be found in Fig. 5. There

are, of course, similarities between the charts for the individuals and the journals; above

all, the ‘‘high age’’ in mathematics. At a first sight, the low mean age of the h-core for

the physics journal is somewhat striking. However, in the light of the scope of the journal

the reasons for the almost concave curve with low gradient become somewhat clearer. The

journal is (1) focussed on a hot subject, (2) primarily publishes an experimental papers and

accepts theoretical articles only if those are clearly linked to experiments, and (3) publishes

rapid communications along with regular research articles. The fast ageing is also reflected

by the low Cited Half-life reported for this journal in the annual Journal Citation Reportsof Thomson Reuters. In the ISI Subject Category physics, condensed matter this journal,

Fig. 4 H-index sequence for scientists in life sciences (left) and in social sciences (right) [Source ThomsonReuters Web of Knowledge (formerly referred to as ISI Web of Science)]


123

together with journals on nanoscience and -technology, has one of the lowest Cited

Half-life.

The shape of the age diagram of the h-core sequence of the fourth journal, Sciento-metrics, resembles to that of the fourth author in Fig. 3. For this journal we have found

several hot topics that might be responsible for the dramatic drop around 2004. Above all,

the h-index and its derivatives has speeded up scholarly communication and provoked

extended discussions with numerous citations. In addition, highly topical issues like sci-ence and technology in emerging economies, author self-citations and mapping andvisualisation of networks have also contributed to this effect. Actually, if we calculated the

average age of new papers entering in each h-core, the mean age of these ‘‘new-comers’’ in

h-core of 2007 was only 2.85, and that same value in 2010 was even lower, particularly,

2.25, in contrast to the much higher mean age (4.4) of the new members of the h-core in

2004. Due to a certain citation delay, some ‘‘hot papers’’ appearing around 2004 have

experienced a citation boom in the subsequent periods. In particular, one-third of the

highest-cited publications in 2010 were from 2002–2006, where the two top cited papers

were published in 2004 (Ho, Citation review of Lagergren kinetic rate equation on

adsorption reactions) and in 2006 (Egghe, Theory and practise of the g-index). In this

context we would like to mention that among the 55 publications in the h-core of 2010, the

year 2006 was, with 6 highly cited contributions, the most important contributor. Another

interesting phenomenon is that though the last decade was a prosperous period for

Scientometrics, there is no paper published in the year of 2000 appearing in h-core in any

period under study. The graph of h-index sequence of Scientometrics is presented in Fig. 6,

where we can observe a clear turning point of increasing h-index since 2004. For both the

scientist and the journal in social science, the ‘‘concave’’ period of their h-core age

sequence is exactly corresponding to their phase with the highest citation impact.

Fig. 5 Mean age sequence of the h-core for four journals (top left Journal of Infection, top rightSuperconductor Science & Technology, bottom left Probability Theory and Related Fields, bottom rightScientometrics) [Source Thomson Reuters, Web of Knowledge]


123

The co-authorship dynamics

Numerous studies have investigated the phenomenon of scientific collaboration (e.g.,

Beaver and Rosen 1978, 1979; Glanzel 2002; Kretschmer 1994, 2002). The increasing

overall rates of co-authorship over the years, as well as the total number of authors

involved in productions, has already been well examined in a number of fields (e.g.,

Endersby 1996; Kliegl and Bates 2011). The interaction between co-authorship and pro-

ductivity as well as citation impact have been variously studied (e.g., Braun et al. 2001;

Glanzel 2002; Glanzel and Thijs 2004; Gazni and Didegah 2011). A positive correlation

between the number of co-authors and the citation frequency has been reported in several

studies (Baldi 1998; Beaver 2004). As the case in most of the other bibliometric measures,

patterns of co-authorship often differ between specialties within various scientific fields.

Theoretical specialties tend to have lower co-authorship rates than scientists in the applied

science field (Glanzel 2002; Moody 2004).

The relation between collaboration and productivity was first reported by Beaver and

Rosen (1979). The authors concluded that collaboration is associated with higher pro-

ductivity. Braun et al. (2001) and Glanzel (2002) found that while co-publication activity

had grown considerably, the extent of co-authorship and its relation with productivity and

citation impact largely varies among fields.

The question arises of how the size of co-authorship is changing along with the indi-

vidual career life, and whether the changing size of cooperation has any influence upon the

authors’ productivity. ‘‘The extent of co-authorship’’ here denotes the number of different

co-authors of the scientist under study.

Figure 7 presents the cumulative number of different co-authors and publications for the

four scientists in the career periods under study. The subject specific characteristics are

rather obvious. The two scientists in life science and natural science had, in general, more

co-authors than publications, and the cases of mathematics and social sciences show the

opposite picture. On the other hand, the number of different co-authors of the scientist in

the life sciences has been dramatically increasing since 1995 (over 900 recently), while the

co-authorship of the mathematician is somewhat limited within a smaller community, and

has been quite stable during the recent years. The most interesting case is the scientist in

natural sciences: the extent of his co-authorship mirrors the evolution of his publication

activity in an quite consistent manner.

Fig. 6 H-index sequence for Scientometrics [Source Thomson Reuters, Web of Knowledge]


123

The ratio of different co-authors over publication output in the corresponding cumu-

lative periods reveals further details of the academic career of each individual scientist.

The mean number of different co-authors sequence for the four scientists is shown in

Fig. 8.

If an author is part of a stable team and as such co-author of its publications, normally

he/she shares most of the papers with the other members of the team. In this situation, the

rate of different co-authors over publications should normally be large in the beginning of

his/her career (C1) since most authors start their careers as co-author, then it should

quickly decrease to smaller values. However, real life is different, so that we have found

cases that considerably deviate from the above-mentioned ideal case. Scientists #4, who is

active in the social sciences, first follows the stable-team model till the ratio slightly

increases after 1992 and around 2002. Since the author is relatively productive, the changes

in the ratio are rather small albeit measurable. This is in line with this author’s mobility.

New affiliations, a new environment should normally increase the number of the co-author

in question. This explains the increase of the ratio in the recent periods.

The extreme 1995 peak of the ratio of the scientist representing the life sciences is

caused by one single ‘‘hyper-authorship’’ with 66 co-authors. After this publication the

author’s co-authorship ratio follows again the ‘‘regular’’ trend.

The mathematician started his career as an author with a single-authored paper—as

usual at that time. The huge increase in the second period and the stagnation during the first

years of his career are due to the relatively small publication numbers, which are typical of

this field. The explanation of the peak in the fourth period and the slight increase in the

most recent periods of the career of scientist #2 is less obvious. Nevertheless, authors #1

and #3 follow the assumed trend in later phases of their career, while authors #2 and #4

Fig. 7 Co-author and publication sequence for four scientists (top left life sciences, top right naturalsciences, bottom left mathematics, bottom right social sciences) [Source Thomson Reuters, Web ofKnowledge]


123

constantly increased collaboration including different co-authors in later periods without

contributing to ‘‘hyper-authored’’ publications. The shape of the co-authorship ratios of

author #4 in the first phase of his career is completely in line with the ideal case since he

published for about 10 years as a member of a small stable team.

Conclusions

Both the scientometric age pyramid and the age curve of the h-core proved interesting tools

for the dynamical career analysis of individuals and journals. Another possible extension of

this type of demographic–informetric analysis might form the application of ‘real’ author

populations such as research groups, departments or even institutes. The h-index and its

derivatives have already been introduced in institutional evaluation (e.g., Molinari and

Molinari 2008). Thus the applicability of h-sequences and the mean age of their h-cores to

institutions will work as well as in the case of journals. Perhaps more pronounced shapes

and trends can be expected. The changing constitutions of teams with stable cores (con-

tinuants) but with varying extend of transients, newcomers and terminators (cf. Price and

Gursey 1976) might have strong influence on the age structure of the team’s publications

and citations received by those. It might then be interesting to identify young, dynamic

teams focussing on hot research and to monitor the ups and downs in activity and impact of

groups and departments with long tradition and experience in their research areas.

Fig. 8 Evolution of number of different co-authors over papers for four scientists (top left life sciences, topright natural sciences, bottom left mathematics, bottom right social sciences) [Source Thomson Reuters,Web of Knowledge]


123

The authorship analysis reveals further aspects of integration in co-authorship teams.

According to the basic assumption stable team work should—at least at later phases of the

career—result in decreasing ratios of co-authors over publications with concave patterns.

Two of the four selected authors showed contradicting patterns, one of which was partially

conditioned by mobility and multiple affiliations. The reasons for a steady increase of the

number of different co-authors are, however, not unique, rather complex and not only

caused by mobility or hyper-authored papers alone. Enlarging partnership networks and

multilateral projects might also contribute to this effect.

The bundle of time series proposed in this study allows insight not only in basic aspects

of individual scientists’ publication activity and citation impact, but also into particular

phases in their career and in changes of their communication patterns in the course of their

academic lifetime.

Acknowledgments The present study is an extended version of a paper presented at the 13th InternationalConference on Scientometrics and Informetrics, Durban (South Africa), 4–7 July 2011 (Zhang and Glanzel2011). The first author would like to acknowledge the support from the National Natural Science Foundationof China under Grant 71103064.

References

Baldi, S. (1998). Normative versus social constructivist processes in the allocation of citations: A network-analytic model. American Sociological Review, 63, 829–846.

Bar-Ilan, J. (2006). H-index for Price medalists revisited. ISSI Newsletter, 2(1), 3–5.Bar-Ilan, J. (2010). A follow-up on the h-index of Price medalists. ISSI Newsletter, 22(2), 39–43.Batista, P. D., Campiteli, M. G., Kinouchi, O., et al. (2006). Is it possible to compare researchers with

different scientific interests? Scientometrics, 68(1), 179–189.Beaver, D. D. (2004). Does collaborative research have greater epistemic authority? Scientometrics, 60(3),

399–408.Beaver, D. D., & Rosen, R. (1978). Studies in scientific collaboration. Part I. The professional origins of

scientific co-authorship. Scientometrics, 1, 65–84.Beaver, D. D., & Rosen, R. (1979). Studies in scientific collaboration, part II. Scientific co-authorship,

research productivity and visibility in the French elite. Scientometrics, 1, 133–149.Braun, T., Glanzel, W., & Schubert, A. (2001). Publication and co-operation patterns of the authors of

neuroscience journals. Scientometrics, 51(3), 499–510.Braun, T., Glanzel, W., & Schubert, A. (2005). A Hirsch-type index for journals. The Scientist, 19(22), 8.Braun, T., Glanzel, W., & Schubert, A. (2006). A Hirsch-type index for journals. Scientometrics, 69(1),

169–173.Burrell, Q. L. (2007). Hirsch index or Hirsch rate? Some thoughts arising from Liang’s data. Scientometrics,

73(1), 19–28.Cronin, B., & Meho, L. (2006). Using the h-index to rank influential information scientists. Journal of the

American Society for Information Science and Technology, 57(9), 1275–1278.Cronin, B., & Meho, L. (2007). Timelines of creativity: A study of intellectual innovators in information

science. Journal of the American Society for Information Science and Technology, 58(13), 1948–1959.Egghe, L. (2006). Theory and practise of the g-index. Scientometrics, 69(1), 131–152.Endersby, J. W. (1996). Collaborative research in the social sciences: Multiple authorship and publication

credit. Social Science Quarterly, 77, 375–392.Gazni, A., & Didegah, F. (2011). Investigating different types of research collaboration and citation impact:

a case study of Harvard University’s publications. Scientometrics, 87(2), 251–265.Glanzel, W. (2002). Coauthorship patterns and trends in the sciences (1980–1998): A bibliometric study

with implications for database indexing and search strategies. Library Trends, 50(3), 461–473.Glanzel, W. (2006). On the opportunities and limitations of the h-index. Science Focus, 1(1), 10–11. (in

Chinese).Glanzel, W., & Persson, O. (2005). H-index for Price medalists. ISSI Newsletter, 1(4), 15–18.Glanzel, W., & Schoepflin, U. (1994). A stochastic model for the ageing analyses of scientific literature.

Scientometrics, 30(1), 49–64.


123

Glanzel, W., & Thijs, B. (2004). Does co-authorship inflate the share of self-citations? Scientometrics, 61(3),395–404.

Glanzel, W., & Zhang, L. (2010). A demographic look at scientometric characteristics of a scientist’s carrier.ISSI Newsletter, 6(3), 66–84.

Ho, Y. S. (2004). Citation review of Lagergren kinetic rate equation on adsorption reactions. Scientometrics,59(1), 171–177.

Jensen, P., Rouquier, J. B., & Croissant, Y. (2009). Testing bibliometric indicators by their prediction ofscientists promotions. Scientometrics, 78(3), 467–479.

Jin, B. H., Liang, L. M., Rousseau, R., et al. (2007). The R- and AR-indices: Complementing the h-index.Chinese Science Bulletin, 52(6), 855–863.

Kliegl, R., & Bates, D. (2011). International collaboration in psychology is on the rise. Scientometrics,87(1), 149–158.

Kretschmer, H. (1994). Co-authorship networks of invisible colleges and institutional communities.Scientometrics, 30(1), 363–369.

Kretschmer, H. (2002). Similarities and dissimilarities in coauthorship networks: Gestalt theory as expla-nation for well-ordered collaboration structures and production of scientific literature. Library Trends,50(3), 474–497.

Levitt, J. M., & Thelwall, M. (2009). The most highly cited Library and Information Science articles:Interdisciplinarity, first authors and citation patterns. Scientometrics, 78(1), 45–67.

Liang, L. (2006). H-index sequence and h-index matrix: Constructions and applications. Scientometrics,69(1), 153–159.

Molinari, A., & Molinari, J. F. (2008). Mathematical aspects of a new criterion for ranking scientificinstitutions based on the h-index. Scientometrics, 75(2), 339–356.

Moody, J. (2004). The structure of a social science collaboration network: Disciplinary cohesion from 1963to 1999. American Sociological Review, 69(2), 213–238.

Price, D. D., & Gursey, S. (1976). Studies in scientometrics. Part 1. Transience and continuance in scientificauthorship. International Forum on Information and Documentation, 1, 17–24.

Schubert, A., & Glanzel, W. (2007). A systematic analysis of Hirsch-type indices for journals. Journal ofInformetrics, 1(3), 179–184.

Zhang, L., & Glanzel, W. (2011). Where demographics meets scientometrics: Towards a dynamic careeranalysis. In: E. Noyons, P. Ngulube, & J. Leta (Eds.), Proceedings of ISSI 2011—The 13th Interna-tional Conference on Scientometrics and Informetrics, Durban, 4–7 July 2011, pp. 914–923.


123

Date post:	19-Aug-2016
Category:	Documents
Upload:	lin-zhang
View:	216 times
Download:	1 times

Where demographics meets scientometrics: towards a dynamic career analysis

Documents