Introduction

Even though the application of bibliometric methods is more popular and adequate in the natural sciences, the possibilities of applying such techniques in the social sciences should be explored (Glänzel and Schoepflin 1999; van Leeuwen 2006). Bibliometric methods for monitoring research performance in the social sciences should explicitly consider the heterogeneity of publication and citation behaviors in these disciplines (Nederhof 2006). In contrast to the natural sciences, social scientists publish in different formats, specifically, they rather produce books and contributions to edited volumes and monographs than journal articles. Naturally, they may focus more on issues that are of national, regional, or even local interest than natural scientists do (Hicks 1999; Nederhof 2006). Therefore, they publish more in the national media and in the local language rather than in sciences lingua franca, English.

Bibliometric studies show that while natural scientists mostly publish journal papers, social scientists publish in a wider range of different types of literature. Bourke and Butler (1996) examined all research output from Australian universities for 1991 and found that natural scientists published about 84 % of works in journal articles or published conference papers, as opposed to the 60 % in the social sciences and humanities. Pestaña et al. (1995) investigated the Annual Reports of the Spanish Scientific Research Council (CSIS) to analyze the research output of the eight divisions of CSIS. 81 % of the output from the seven natural science divisions were in journals while the social science and humanities division published 54 % of its research as journal articles. The results from the Norwegian system providing complete scholarly publications from 2005 to 2008 reported by Sivertsen (2009) show a similar disparity: 90 % of publications in the natural sciences in Norwegian universities are journal articles (articles in series with an ISSN), while 60 % of publications in the social sciences are journal articles.

Not only publication behavior but also citation behavior in the social sciences differs from that of the natural sciences. While the latter prefer to cite journal articles, citation patterns of social scientists are more diverse, with books and monographs being cited the most. Citations to and from books are distributed differently from those to and from journal articles. Additionally, these publications reach their citation peaks much later (Bourke et al. 1996; Clemens et al. 1995; Hicks 1999; Hicks 2004; Line 1979, Nederhof et al. 2010). Samuels (2011, 2013) collected the citations of sets of books and journal articles in political science from both Social Sciences Citation Index (SSCI) articles and books, and found that the average university-press book receives about three times the number of citations received by a SSCI article. Hicks and Potter (1991) collected a bibliography of sociology of scientific knowledge, and found that journal articles were cited 1.2 times while books got 5.7 citations on average (“Books” here includes whole books, edited books, edited journals and monographs). As books play such an important role in publishing academic results in the social sciences, they should be taken into consideration in research assessments. It is important to bear in mind that 40 % of citations that books receive, as reported in Bourke et al. (1996) and Hicks and Potter (1991) by the share of references to monographs, are missed if we use the indicators built from citations by SSCI-indexed journal (Hicks 1999).

Political science was selected as the focus of this study because its average-level bibliometric characteristics, such as the Web of Science (WoS) coverage, the share of book publications or language preference, are typical of many subfields in the social sciences. It is also a relatively empirical field and among the top three fields with the largest increase in citations caused by the inclusion of non-indexed items according to one study (Butler and Visser 2006). Van Leeuwen (2006) used the data of Delft University of Technology over the period 1994–2003 and found that the share of references to Institute for Scientific Information (ISI, nowadays Thomson Scientific) publications in political science and public administration is only about 20 %. It was identified that the ISI internal coverage of political science is 20 %. Political scientists read and cite more articles from local journals or other non-serials sources. In his recent study, van Leeuwen (2013) analyzed all publications in political science and public administration in 2010 and the internal coverage is about 31 %. The ISI internal coverage of German output in political science and public administration in 2011 is about 40 % in the same study.

On the other hand, the ISI external coverage of Australian universities publications from 1999 to 2001 in political science reported by Butler and Visser (2006) is also 20 %. In another study handled by Engels et al. (2012), the publications in the regional bibliographic database of Flanders in the social sciences and humanities (VABB-SHW) show that publications in political science in Flanders are covered around 17 % by WoS during the period 2000–2009. Among these 1,260 publications, 79 % of them are published as journal articles, 16 % are book chapters, 3 % are books as editor, and 2 % are book as author. Political Science is the only discipline of the Social Sciences where book publications represent more than 20 % of the output in this Flemish study. More than half of publications in political science in Flanders are published in English (65 %), and around 35 % are in Dutch. In the Norwegian national publications system (CRIStin), the WoS coverage of all publications in political science from 2005 to 2008 is 27, 64 % of publications are in foreign language, and 45 % of publications in political science are journal articles with an International Standard Serial Number (ISSN), 51 % are articles in book or proceedings with only International Standard Book Number (ISBN), the rest 4 % are books (Sivertsen 2009).

The limited coverage of the WoS databases in the social sciences will certainly lead to errors when bibliometric methods are applied to these subject fields. Accordingly, the bibliometric indicators which are applied in evaluation procedures in the social sciences need to be considered carefully. Thus, this study aims to analyze the publication, citation, and referencing behaviors of the social sciences in a specific empirical field, political science, in order to uncover the characteristics and peculiarities of this field, for more knowledge to apply relative proper evaluation procedures. It should be pointed out right from the start that the analyses in this study may be generalized in assessments of other social sciences, while the outcomes of the citation- or reference-based analyses are more specific to German political science and any extrapolations should be done with caution.

Data and methods

The 5-year publication output (2003–2007) of two top-ranking German political science institutions, Department of Political Science at Mannheim University and Institute of Political Science at University of Muenster (CHE University Ranking 2010/11-Political science; Hix 2004), was chosen as research samples. The 1015 publications of 33 professors in these two institutions were collected from the researchers’ official websites, institutional repositories, and German Social Science Literature Information System (SOLIS). The professor list was updated in November 2010 according to the official websites of these two institutes. Visiting professors, research fellows, and Post-Docs are excluded, since they are not the main contributors to scholarly publications or long-term employees. Fundamentally, this study adopts the classification of document type directly from the professors’ CVs. Each item has only one document type.

After data collection, all publications were sent to the professors for verification. 60 % of the professors confirmed or updated their bibliographic records, especially in the area of the language or document type data. The remaining records from the professors who did not reply are stored as the original data from websites. Citations of these publications and references of WoS articles were obtained from the 2012 version (fixed at the 17th week of 2012) of the WoS in-house database of the Competence Centre for Bibliometrics for the German Science System (Kompetenzzentrum Bibliometrie, KB), including the raw data of Science Citation Index Expanded (SCIE), SSCI, and Arts & Humanities Citation Index (A&HCI) from 1980 to present. The WoS citations mentioned in this study exclude the citations from Book Citation Index (BKCI). Additional citation data were searched in BKCI web version on 6th of November 2013. A sliding citation window of 4 years was applied in this study (e.g., for an item published in the year 2003 citations until 2006 were considered).

Citations to all items were acquired by matching the corresponding search terms to the references in the WoS in-house database, according to a set of rules for different document types. It takes at least two rounds of Structured Query Language (SQL) queries to identify the citing references in the WoS-based database. Take Book Chapter for example, during the first round, the method used in this study searches the first word of the title and the first author name (surname, first initial) in the references of WoS, then filters the results to include only those items listed with publication year within ±1 year of the target year and try to broaden the different abbreviations of journal titles and author names indexed in the database from the primary salvaged results. A specific rule applied to Book Chapter is that the matched references with an accurate first page number may have a ±1 publication year difference from than the original one, but those without accurate first page numbers need to exactly match the publication year. The first page number is allowed to vary ±2 when other conditions are met. The obvious typos, such as p. 267 are indexed as p. 26, are accepted in this study if other metadata are exactly matched. Next, the second round is to repeat the search query by matching the first page and the first author’s last name instead, and filter the results by the rules of year and page again. In the end, duplicates of the combined results from these two rounds are removed. For some source articles showing no exact first page in their references where there are more than two chapters written by the same author in one book, the full-text of the articles was checked manually to make sure which chapter was cited exactly.

The citations discussed in this study include self-citations but self-citation rates were also calculated. Citations are traced diachronically (Aksnes 2003b) and the definition of self-citations proposed by Borgman and Furner (2002) is used, “author self-citation occurs when at least one of the authors of a cited document is the same person as one of the authors of the citing document.”

Publication patterns

Coverage of source items

Among all 1015 publications of these two institutions, 70 of them are indexed in WoS (SCIE, SSCI, and A&HCI). The low external coverage of WoS shows that this citation index covers only a very small part of the German political science publications. Consequently, relying only on this database fails to include the whole scientific communication of the community in this field. The references of the 70 WoS indexed articles were analyzed for estimating the internal coverage. There are 2593 references cited by these articles, including 818 WoS indexed items and 1775 non-WoS indexed items. The internal WoS coverage is therefore 32 %, and the average number of references of these 70 articles is 37.04, which is lower than the aggregated average of references (47.7) in the political science category in the 2013 social sciences edition of Journal Citation Reports (JCR).

The internal coverage in this study is generally higher than external coverage, showing that German political scientists pay attention to cite source articles while publishing, although they do not publish mainly in WoS journals. According to the data from Delft University of Technology from 1994 to 2003 (van Leeuwen 2006), the share of references to ISI covered publications in political science and public administration is about only 20 %, lower than the 32 % shown in this study. The results shown in another similar study (van Leeuwen 2013) indicate the gradual increase of the share of ISI references in political science and public administration in Germany, from 15 % in 1991 to 40 % in 2011. The share of WoS references of source items in this study increases from 31.6 % in 2003 to 37.5 % in 2007. From these two studies it can be seen that when German political scientists publish WoS articles, they refer more and more to WoS articles. It could be speculated that authors might tend to cite more WoS articles because they are easier to be found. The other possible reason may be that journals often ask authors to cite other articles from the same journal.

Table 1 shows that including BKCI additionally leads to a 20 % increase in citations to the BKCI indexed items, even though there are only 23 items indexed in BKCI (5 edited books and 18 book chapters). The average citation rate of BKCI indexed items (10.9) is higher than WoS indexed items (7.1). Among 23 BKCI indexed items, only 6 of them are provided with reference data in BKCI. It is obvious that BKCI does not yet provide sufficient cited references data currently. The average reference rate of BKCI indexed items with references data (165.2) is also higher than WoS indexed items (37.0). This shows that BKCI edited books and book chapters are cited more in WoS than ISI journal articles are, and also refer to more references than ISI journal articles do due to their longer length. However, the BKCI indexed items refer to much more document types but fewer source references (19 %) than WoS indexed items (32 %).

Table 1 Citation and reference statistics of WoS- and BKCI- indexed items
Full size table

Document type

Table 2 shows that German political scientists publish more books, edited books, and book chapters (51 %) than journal articles (22 %). In comparison, the Norwegian and Flemish data (Sivertsen and Larsen 2012; Engels et al. 2012) suggest that Norwegian political scientists publish these two types nearly commensurately (54 % items with ISBN and 46 % with ISSN) and Flemish political scientists publish much more journal articles (79 %) than books, edited books, and book chapters (20 %). Apparently, there are only two main peer reviewed types of publication (books and journal articles) in their datasets due to the funding allocation purpose. Peer reviewed conference papers are rarely collected in the Flemish bibliographic database (VABB-SHW), and even included in the category of articles in books in the Norwegian system (CRIStin). The larger share of monographs in this study comparing to other studies may be because the non-peer reviewed monographs (especially book chapters) are included in this dataset but excluded in VABB-SHW and CRIStin. It reveals that the evaluation targets in different systems are different as a result of different purposes of systems. In fact, the difference in publication patterns for the same field among different countries should be smaller than the difference across fields. Sivertsen and Larsen (2012, p. 569) state that “publication patterns differ between disciplines but are similar across countries, and that results from studying only one country can be generalized to a certain extent.” The difference among three political science datasets analyzed here shows that there could be a reflection of different system designs instead of the fundamental divergence. On the other hand, it could also be the effect of different country sizes. The bigger local audience would allow a bigger country to keep the traditional publishing culture more than a smaller country.

Table 2 Shares of document types in political science in different countries
Full size table

The publication behavior of German political scientists is changing. The difference in proportions of publication types between the 2003–2004 and 2006–2007 periods is significant at α = 0.05 (using the Chi squared test of difference in proportion). The 2-year publication periods are used to aggregate more analysed samples for each document type category. The relative share of book chapters, edited books and authored books increases during 2003–2007 (from 50.5 % in total to 56.2 %), whereas the share of journal articles (ISI journal articles and non-ISI journal articles) decreases during this time (from 23.6 to 18.1 %). Although the overall share of journal articles decreases, the ratio of ISI journal articles to non-ISI journal articles is increasing every year. The amount of ISI journal articles increases over the 5 years (from 6.3 to 7.9 %) but the amount of non-ISI journal articles decreases (from 17.3 to 10.2 %).

The citation flow of ISI journal articles shows that German political scientists publish mostly in European journals and get their citations from them (Journal of European Public Policy, Politische Vierteljahresschrift, West European Politics, and European Union Politics), indicating that there is a continental community in the field (for the two departments studied). However, they cite American journals, American Political Science Review and American Journal of Political Science, the most (23 % in total). To illustrate the background of this phenomenon, one of the professors from the samples was consulted. According to this personal communication, American political science journals, especially American Political Science Review and American Journal of Political Science, are regarded as the best journals in this field with very high quality criteria and very high rejection rates.Footnote 1

The influence of the American community on German political scientists can also be shown in the location of the conferences they attended from their conference papers. About 26 % of the conferences where these political scientists published their conference papers with location information are in USA, while 27 % of the conferences are in Germany. Apart from the German conferences, German political scientists frequently attend conferences in the United States, especially the Annual Meeting of the American Political Science Association. The above two findings in terms of the referencing and conference attendance show that even though the local German communication network is important, the American academic community still has a large impact on German political scientists.

Language

Among these 1015 published items, around 57 % of the 1,015 published items were in German and 40 % were in English. About 70 % of the published book chapters were in German. Figure 1 shows that the dominant position of German is also prevalent in other publication types, such as edited books, books, and non-peer reviewed journal articles. Compared to WoS papers which are published mostly in English, the other 91 non-WoS peer reviewed papers are more often published in German (60 %) than in English (37 %). German political scientists use German to publish in books and regionally-oriented journals. By contrast, English is used more often than German in peer-reviewed journal articles and conference papers, which serve more international communication purposes.

Fig. 1
figure 1

Shares of items in different languages by document types. The chart is ordered by the share of publications in German

Full size image

Another study (van Leeuwen 2013) shows a trend in political science and public administration: the number of German authors’ ISI-indexed articles in English from 1981 to 2010 increased, but those in German remained stable. In the present study, German political scientists tend to publish non-ISI journal articles more in English, while no clear trend of language change is apparent in ISI journal articles. In fact, the percentage of ISI journal articles in German to all languages even increases slightly from 2003 to 2007 (from 15.4 to 22.2 %), but it could not be regarded as a trend, due to the actual numbers of ISI journal articles in German being very small (2 in 2003 and 4 in 2007).

Items in German (no matter which document type) are observed to be cited by more WoS citations in German than items in English (Fig. 2). The books in German have 62 % of WoS citations in German, while books in English have only 8 % of WoS citations in German. On the other hand, BKCI does not have a sufficient coverage of books in German in political science, reflecting a very poor percentage of BKCI citations in German to all items. This implies that BKCI is not ready to be one of the citation sources for citation analysis in political science in Germany, even though it could be a useful supplement to collect citations from books to non-indexed items in the future. Furthermore, the percentage of citations from books to books is expected to be much higher (and could possibly reach the 40 % observed by Hicks 1999) if the coverage of BKCI is improved.

Fig. 2
figure 2

Shares of citations in German from WoS and BKCI

Full size image

Individual publishing behavior

To investigate the different publishing preferences among different generations, all professors were classified by age (excluding two professors whose ages are unknown), in order to find differences in publication patterns by age. Professors who are older than 65 years old (12 persons) publish around 26 publications on average during the reporting period and obtained 0.36 citations on average, see Table 3. Instead of publishing journal articles (6 ISI journal  articles and 46 non-ISI journal  articles, in total 16 % of all), this group of professors publishes more in books, resulting 71 % of their publications are book chapters, books, and edited books. Furthermore, the ratio of ISI journal articles to non-ISI journal articles is 0.14 (6:46).

Table 3 Publication and citation statistics of three age groups
Full size table

Researchers aged between 50 and 65 (11 persons) in this study publish around 40 publications and obtained 0.84 citations from all items on average. They publish more journal articles (24 %) but fewer books (54 %) than the older group. Their ratio of ISI journal articles to non-ISI journal articles is 0.37 (28:75). Those who are younger than 50 (8 persons) publish around 37 publications and have the highest average citation rate, 1.13. 24 % of their publications are journal articles and 28 % are books, showing that they publish more diversely than other two groups. The ratio of ISI journal articles to non-ISI journal articles in this group is 1.03 (36:35).

The analyses above show that the publication behaviors are changing with generations, which has an influence on the citation impact of their works. The older researchers publish fewer publications during the period 2003–2007 due to the inactivity in retirement. However, it is obvious that they have a higher preference for publishing books than journal articles, and they published in non-ISI journal articles more than ISI journal articles while publishing journal articles. By contrast, the younger researchers publish more ISI journal articles and fewer books, thus achieving a higher citation impact than older researchers observable in WoS.

Citation patterns

Average citation rate

In general, books (including edited books) have higher impact than other non-source items in the social sciences as shown in Table 4, especially in the field of law in Australia. In Amez’ study (2013), the analysis of the 610 publications in VABB-SHW by Vrije Universiteit Brussel (VUB) authors shows that in the social sciences and humanities books as editor have the highest average citation rate (3.54). The high impact of edited books among non-indexed items is shown in both Amez’ and the present study (deviations from other document types in each study are both positive). However, the authored books in German political science have a higher citation impact than books as author in the Belgian SSH (deviation −0.16: −0.60), compared to the difference of average citation rates of book chapters between two studies (deviation −1.12: −0.70).

Table 4 Average citation rates in German, Belgian, and Australian studies
Full size table

In order to assess the citation impact of the German political scientists in this study at a worldwide level, the average citation rate of all the articles (document type is limited to “Article”) published from 2003 to 2007 in the category “Political Science” of WoS are calculated in the in-house database of KB with a sliding 4-year citation window. Figure 3 shows that the average citation rate of ISI journal articles in this study is much higher than the average of WoS articles in the whole world and the average for all of Germany in political science. Although we should take into account that this study collects only publications published by professors, which might be more qualified than publications from other researchers or students, Fig. 3 still reveals the high impact of the ISI journal articles of these two German political science institutions.

Fig. 3
figure 3

Average citation rates of political science papers in WoS

Full size image

Uncited rate

In this study, the overall uncited rate of all publications of two German political science departments is about 80 %. The document types having the highest citations per item also have the lowest uncited rates, revealing that some document types have higher influence than some other types. Only 19 % of the ISI journal articles are never cited 4 years after they were published, and 61 % of the edited book and 64 % of the books are uncited. On the other hand, non-ISI journal articles and conference papers, reports, magazine/news articles, and others have uncited rates even higher than 80 %.

Two Australian universities (Butler and Visser 2006) have publications with a lower share of uncited items than the German study in political science (Table 5). In particular, their research monographs are cited more often than the German ones. Disciplinary difference and the language influence could be possible reasons affecting the uncited rates. The share of uncited German political science source items (19 %) which are mainly in English (56 out of 70) is not very different from the Australian source items’ (14 %). Conference papers are rarely cited in both German and Australian studies and have relatively low impact as proved in German, Belgian, and Australian studies (see Table 4). These robust convergent results strongly suggest that conference papers which contain preliminary research results are not the main reference for citation.

Table 5 Uncited rates in German and Australian studies
Full size table

In comparison to the Belgian study (Amez 2013) which reports on non-ISI publications in social sciences and humanities from 2002 to 2008, Table 6 shows that the share of German political science books (64 %) and edited books (61 %) which are never cited by WoS-indexed items is lower than Flemish ones (both 78 %). These two document types also have relatively higher impact than other document types in this study. By contrast, the non-ISI journal articles and book chapters which have relatively low impact are cited less often than the SSH publications of the VUB. Amez points out that contrary to the Australian case (Butler and Visser 2006), the VABB-SHW contains only items from the social sciences and humanities, showing a more prominent role played by book contributions. It is even stronger in the present study.

Table 6 Uncited rates in German and Belgian studies in the social sciences
Full size table

Self-citation rate

Table 7 shows that the average self-citation rate of German political science publications is about 21 % (the average self-citation rate without a 4-year citation window is approximately 16 %). The overall self-citation rate for the Norwegian national ISI articles from 1981 to 1996 considering citations from year of publication until 2000 reported by Aksnes (2003b) is 21 %. According to Aksnes, the self-citation rate of physics in the Netherlands (1985–1994) is 29 % (van Leeuwen et al. 1996), and is the same as reported in a study on Dutch chemistry (1980–1991) (Moed and van der Velde 1993). The observed rate in this study (21 %) is lower but not far from other fields.

Table 7 Self-citation statistics within different citation periods
Full size table

Another trend shown in these studies is that with longer citation life the self-citation rate declines. In Aksnes’ (2003b) study, the self-citation rate is 36 % within a 3-year citation window, 29 % within a 5-year window, and 21 % without any citation window. This is similar to the results in this study, which shows a self-citation rate of 21 % with a 4-year citation window and 16 % when there is no citation window applied. It may reveal that authors start to cite more recent articles when citing themselves.

International impact

Figure 4 shows that more than one-third of citations of all items are from articles with at least one address in Germany. Only working papers and books are highly cited by articles written by authors only from Germany. The 46,849 Norwegian scientific articles from 1981 to 1996 (Aksnes 2003a) show that only 19 % of the citations come from Norwegian scientists due to the size of the nation. The author expects that the share of domestic citations would be much higher for a larger scientific nation like Germany. The political science articles in this study support this prediction. German articles garner a higher rate of domestic citations than Norwegian ones.

Fig. 4
figure 4

Shares of papers from different countries citing different document types

Full size image

Political science articles here have a large share (60–80 %) of the citations from articles with at least one address from non-German countries. They are not predominantly cited by authors from Germany. This result might challenge the notion of political science as a locally-oriented field in the social sciences, since these German political science articles attract most citations from countries other than Germany.

However, publications in German have a more locally-oriented impact than those in English. In Fig. 5, the difference between the two groups is significant (Chi squared test, p < 0.0001). About 70 % citations of items in German are from articles written by authors from Germany exclusively, while articles in English have a wider international influence, with 78 % citations from other countries except for Germany.

Fig. 5
figure 5

Shares of papers from different countries citing items in different languages

Full size image

Conclusions

This study examines the publication patterns of German political scientists, and provides an insight into their citation and reference characteristics. The results may serve as background information for peer review committees conducting research evaluation in the social sciences, especially in political science. German political scientists in this study mainly present their academic results in three channels, monographs, journal articles, and conference papers, to disseminate their ideas to their peers. About half of the publications are published in monographs, and around one-fifth of them are in journals. Their publication behavior is changing: the relative share of monographs increases during the 5 years, whereas the share of journal articles decreases during this time. However, they publish more ISI journal articles rather than non-ISI journal articles when they are publishing in journals. The impact of the two top departments' ISI journal articles is much higher than the average impact in political science compared to the baselines of Germany and the whole world.

The usage of German in publishing research output is decreasing, while the publications in English and usage of the more international English-language channels, such as ISI journal articles and conference papers, are increasing year by year. However, English is not dominating in all publications of the two German political science institutions after all. We could argue that the local language is an important issue which needs to be addressed concerning the completeness of coverage of political scientists’ publications. Without including publications in German, for both monographs and journal articles, the dataset is far from complete.

The percentage of overall WoS coverage is about 7 %, while the BKCI coverage is about 2 %. The internal BKCI coverage of BKCI-indexed items is 19 %, while the internal WoS coverage of WoS-indexed items in this study is 32 %. The lower external coverage of WoS/BKCI and higher internal coverage of WoS/BKCI in this study as compared to other studies show that ISI journal articles reside within a relatively small and closed publishing channel in political science in Germany compared to other countries. Furthermore, the lower share of journal articles and the even lower share of ISI journal articles imply that the current coverage of citation databases may be far from representing the outputs of researchers in the social sciences, and at least books and book chapters should be included in quantitative and qualitative evaluations of political scientists.

The uncited rate may be influenced by the language or the accessibility, resulting in different numbers for different document types, fields, or countries. Basically, the document types with higher citation impact have a lower uncited rate, though working papers and discussion papers have low uncited rates which may be influenced by their high self-citation rates. However, the observed self-citation rates do not vary across fields as does the uncited rate. The trend found in different studies is that the longer time for citing, the lower the self-citation rates.

German political science publications do not have a strongly locally-oriented impact, except for the publications in German. Although about 40 % of citations of all items, a higher rate of domestic citations than Norwegian ones (19 %, see Aksnes 2003a), likely because of the size of the nations, are from articles with at least one address in Germany, the publications have a larger share (68 %) of citations from articles with at least one address from non-German countries. However, we should keep in mind that it reflects only the biased characteristics of citations from WoS and therefore the level of local orientation in this field may be underestimated.

All in all, there are two main networks of academic communication in the publication pattern of German political scientists. The significant local communication network covers monographs and regionally-oriented journals that are mainly written in German. Its importance has slightly decreased over time. By contrast, the relatively smaller international one, which covers international publications in English, enlarged its coverage slightly. In the international communication network, the influence of the American community is large, likely owing to the American studies’ quality, impact and numbers. The different publishing behaviors between generations also support the increasing trend in international orientation. An initial increase of the international orientation can be observed. Even though it is not significant yet, its emergence demonstrates an agreement with the results of other studies in the social sciences.