Knowledge linkage structures in communication studies
using citation analysis among communication journals
Scientometrics (forthcoming)
Han Woo PARK[a] and Loet LEYDESDORFF[b]
click here for .pdf>Abstract
This research analyzes a “who cites whom” matrix in terms of aggregated, journal-journal citations to determine the location of communication studies on the academic spectrum. Using the Journal of Communication as the seed journal, the 2006 data in the Journal Citation Reports are used to map communication studies. The results show that social and experimental psychology journals are the most frequently used sources of information in this field. In addition, several journals devoted to the use and effects of media and advertising are weakly integrated into the larger communication research community, whereas communication studies are dominated by American journals.
Key words: Knowledge linkage, communication studies, citation analysis, social network analysis, communication journals
Introduction
Given the notable institutional build-up of media and communication studies since 1990, in not only North America but also other regions including East Asia and Western Europe, the number of manuscripts published in the Institute for Scientific Information (ISI)-rated journals is increasing rapidly (the ISI has recently been renamed Thomson-Reuters but the name ISI continues to be widely used). The conventional perception of communication studies is that it is an incomplete aggregation of atomized research domains.
Scholars have viewed communication studies as a heterogeneous group with different attributes (Delia, 1987; Leydesdorff, 2004), partly due to the duality of communication studies. As Rogers (1994, p. 48) stressed, “Communication is a professional field, as well as a scientific discipline. The mass media industries stand behind the academic field of communication, offering jobs for its graduates, helping fund its research, and providing endowments for professors and schools of communication.” As a result, communication studies have split into many academic entities; for example, speech communication, mass communication, advertising, and journalism. Continual developments in media technology have also forced the founding of new specialties that include telecommunication, new media, computer-mediated communication, informatics, and human-computer interaction.
In 2005, communication studies gained legitimacy as an independent discipline for educating and studying the nature of human communication mediated by various channels (Treadwell, 2006). The National Science Foundation of the USA (NSF) formally accredited communication studies as an academic field with a particular subject in the NSF’s list of disciplines. The representative professional International Communication Association (ICA) was founded more than half a century ago in 1950. Derek de Solla Price (1963) noted that big science refers to a field that has established theories, mature research methods, and scientific collaborations. In this regard, communication studies are ready to become accepted as a big science in the 2000s.
The success in having communication studies recognized as a stand-alone discipline inspired us to write this article. We address the following two research questions: Where on the academic spectrum do communication studies reside? Which subject category is core to communication studies? In order to investigate these topics, we focus on linkage structures in communication studies using citation analysis among communication journals. Citations do not occur in a social vacuum. What communication researchers cite forms how communication studies are constituted. More specifically, we evaluate the development and progress of communication studies by collecting a journal-journal citation matrix from the ISI’s database as an input file to our scholarly investigation.
Related studies
Compared to the speedy diffusion of communication and media studies, scientometric exercises that examine the academic structure of current communication research have been rare. In response to this, the Journal of Communication, a top journal published by the largest association of scholars in communication studies, the ICA, has published two studies that analyze the national origins of communication journals (Lauf, 2005) and professional careers of article authors (Bunz, 2005). Both articles look at the position of communication studies as an independent discipline using bibliometric data obtained from scholarly journals and their authors. In a similar vein, Masip (2005) focused on research articles written by European authors in thirty-five communication journals.
There are more restricted studies that focus on reviewing some specific area in communication studies: Internet research related to advertising (Cho & Khang, 2006), research trends in mass communication (Kamhawi & Weaver, 2003), and media research on the Internet (Kim & Weaver, 2002).
More recently, the Communication Research Centre (CRC, 2007) at the University of Helsinki conducted a comprehensive project about current trends of media and communication research in seven countries (Finland, USA, Germany, France, Japan, Estonia, and Australia). According to the reports about individual countries, communication studies around the world are somewhat differently deployed in some major strands. While in the U.S.-based academic tradition, quantitative investigations of media effects are dominant, communication programs with different approaches (e.g., humanistic and sociological practices) are gradually growing. The CRC’s report on the U.S. also provides the top ten communication and media journals using the ISI’s impact factor from 1998 to 2005. At the top of the top ten list (according to an eight-year mean) is Public Opinion Quarterly, followed by Communication Research, and the Journal of Communication. Media Psychology, Discourse & Society, Human Communication Research, Cyberpsychology & Behavior, Public Culture, Political Communication, and the Journal of Health Communication are also included in the top-ten citation list.
These previous studies are limited to individual characteristics of communication studies. The past literature is unable to discover the networked traits within communication studies by ignoring structural properties among communication journals and their neighboring journals in other fields. As alluded earlier, communication studies as a disciplinary field are made of several scientific domains. Perhaps, this field can be captured better through network-oriented indicators and visualizations than by indicators which normalize over the journals subsumed under the ISI-category of “communication.” The connectivity patterns represented in the citation behavior of authors in communication studies journals may reveal how this fragmented community perceives and reconstructs its relevant scholarly environments.
Methods: Social network metrics and data collection
Indicators and visualizations based on social network analysis are used in this research (for a detailed explanation about social network metrics and technical procedures, see Hanneman & Riddle, 2005). Previous studies (Kim et al., 2006; Leydesdorff, 2005; Park et al., , 2005; Park & Thelwall, 2006; Park & Leydesdorff, 2008) have shown that the structural pattern of citations among a set of authors, articles, or journals can be better examined using a network perspective. For instance, Freeman (2004) suggested that a network approach utilizing citation data may be a more robust means of studying the inside world of an invisible research community. Using social network techniques, Freeman (2004, pp. 165-166) showed that there were surprisingly few bridging linkages between physicists and the traditional social network analysts who cited Milgram (1967)’s famous article entitled “The Small World Problem”. The network perspective, developed within the social and behavior sciences, is currently identified as an influential method in library and information science, and scientometrics (Otte & Rousseau, 2002; Thelwall, 2004, pp. 213-217).
Among the many techniques for gathering network data, snowballing is frequently used in the case where the entire number of nodes in a given social system and whether they entertain relations are a priori unclear (Garton et al., 1997). For example, researchers often get a particular person to list their acquaintances and report their interactions with all other network members in terms of face-to-face meeting frequency per week, the type of communication media used, and the closeness of relationship. Leydesdorff (2007) took advantage of this network sampling technique to visually map the specific journal’s relevant citation environment and succeeded in detecting a collection of journals that make a network having a particular journal as an entrance point. More recently, Park and Leydesdorff (2008) applied this sampling technique to a scientometrics exercise by focusing on the cited dimension of a Korean ISI-listed journal.
The data for this research were harvested from the CD-Rom versions of the Journal Citation Reports (JCR) 2006. More specifically, the 107 journals in the 2006 ISI database that cited the Journal of Communication were taken and an asymmetrical (valued) matrix—that is, a matrix in which the frequencies of citations between journals are provided—was constructed for social network analysis. Next, ISI’s subject categories and publication places of individual journals listed in the sample were collected. Additionally, the journals cited in the Journal of Communication were gathered and a citation matrix among these (154) journals was made. However, our analysis focuses on the domain of 107 journals in the citation impact environment. Aggregated citation data and journal names for these journals are provided in Appendix I. All values in the JCR-data refer to unique citation relations at the article level.
The Journal of Communication was selected as the focal journal for the mapping of communication studies because, as signaled in its title, this Journal is a flagship publication in the communications research community. The ICA, the world’s largest scholarly body in terms of the participatory countries, has an institutional responsibility for the Journal. Compared to the ICA’s other titles (e.g., Communication Theory, Human Communication Research, and Journal of Computer-Mediated Communication), the Journal is relatively open to many studies with varied theoretical and methodological backgrounds. Most articles published in Human Communication Research, for example, seem to contain a purely quantitative analysis. Therefore, studying the journal structure of communication studies as a discipline with the Journal of Communication as an exemplar is probably the best approach, although this is indeed a specialist universe of publications.
UciNet for Windows (Borgatti et al., 2002), a commonly used software program for network analysis, was used to calculate the various metrics and NetDraw packaged with UciNet was used for the graphic illustrations. In social network research, visualization is often an informative tool to reveal the relationship between different network nodes, in this case, journals. NetMiner software (Cyram, 2003) was also used for the visualizations.
Results
Citation structure among 107 journals citing the Journal of Communication
The Journal of Communication was cited by 107 journals in 2006. The network analysis of the 107 * 107 citation matrix reveals that the clustering coefficient of this network is 0.59. The clustering coefficient—defined as the proportion of links between the vertices within the neighbourhood of a vertex divided by the number of links that could possibly exist between all these vertices—indicates the degree to which friends of a person know each other (for a mathematical definition, see Watts, 1999). In this research, the clustering coefficient measures the extent to which a journal’s neighboring journals are connected to one another. Averaging these proportions over all journals in the network shows how closely interconnected neighbors in the network are. Interestingly, the coefficient value (0.59) found in the study is very close to the result (0.56) that network physicists have identified in social networks of scientific collaboration (Barabási, 2002).
The diameter value of 5.00 implies that all journals are connected to the extent of five hops when the direction of the citations is not considered. The mean distance is 2.08. That is, if two journals in the network are selected they are on average about two steps away from each other. Reciprocity is 0.55: more than half of all the journals exchange their citations. For example, the authors of journal A chose the articles published in journal B as their references and so did the journal B’s authors with reference to journal A. All of these network metrics (of course, with the exception of reciprocity) were done after binarization and symmetrization.
More specifically, the Psychological Bulletin is the journal most cited by others (1,524 times) in this network. Conversely, the Journal of Communication was cited 781 times. In network analysis, “indegree centrality” refers to the total number of citations a journal receives from the other journals at the article level and “outdegree centrality” measures how many times a journal cites the articles published in other journals in the same network (Freeman, 1979). For the calculations of both these centrality measures, within-journal self citations were excluded.
Degree centrality measures are primary indicators for the selected node’s network activity. The dominance of the Psychological Bulletin in terms of the being-cited dimension, that is, indegree centrality, means that research articles in communication science journals connected to the Journal of Communication provide references mainly to psychology and its related concepts. As Delia (1987, p. 23) stated, the most influential sources for the construction of communication research come from theoretical concepts and methodological behaviorism in psychology, and the quantitative research orientation in sociology and political science, that is, those specialties among the social sciences in which issues of measurement and operational procedures are of central concern.
In the citing dimension, Psychological Bulletin had a relatively small number of citations which make reference to 267 articles published in 23 other journals in the network. The most central journal in the citing dimension is Sex Roles with 813 citations to 46 other journals, followed by Social Science & Medicine, the Journal of Youth & Adolescence, and the Journal of Communication with 635, 542, and 512 citations, respectively. The difference between these highest numbers in the cited and citing dimensions are reflected in the network centralization metrics. The outdegree network centralization (citing dimension) is 32.73 percent while the indegree centralization (cited dimension) is 85.11 percent. The higher this percentage the more centralized a network is. Therefore, these 107 journals are connected to each other in the cited network more than in the citing network. In the dominant dimension (cited), the Psychological Bulletin is the most central source of citations.
Table 1 lists the most highly and the most rarely cited ten journals with their respective citation counts. Figure 1 illustrates the 107 journals according to their indegrees, that is, times cited in the network. The most highly cited Psychological Bulletin occupies the central position of the target board and the next most frequently cited journals are assorted up and down. The remaining journals are scattered around the target.
Table 1. The most highly and most rarely cited journals in a network using the Journal of Communication as the seed journal
|
Highly cited ten journals |
Indegree (times cited) |
Least cited ten journals |
Indegree (times cited) |
|
PsycholBull |
1524 |
NewZealJPsychol |
9 |
|
DevPsychol |
848 |
BritJEducTechnol |
7 |
|
JCommun |
781 |
InteractLearnEnvir |
5 |
|
CommunRes |
674 |
WorldEcon |
4 |
|
SocSciMed |
623 |
Policing |
3 |
|
SexRoles |
552 |
ZKlPsychPsychoth |
2 |
|
PsycholRep |
388 |
FoodDrugLawJ |
0 |
|
HumCommunRes |
386 |
LangLearnTechnol |
0 |
|
PsycholWomenQuart |
362 |
PolitSci |
0 |
|
JBroadcastElectron |
351 |
TextTalk |
0 |
· The abbreviations used for the journal titles are available at
http://images.isiknowledge.com/help/WOS/A_abrvjt.html. (See also Appendix I.)
Figure 1. Indegree distributions of the 107 journals under study.
l The data used for this Figure were logarithmically scaled.
Additionally, the inter-citation matrix among 107 journals was aggregated according to the disciplinary categories listed in the ISI’s database. Despite the debate about the ISI classification system (Boyack et al., 2005; Leydesdorff, 2006), this perspective can be complimentary to the above findings. The 107 journals belonged to a total of 21 subject categories. The majority of journals is classified as “communication” (28 journals) or “psychology” (24 journals), followed by “Criminology & Penology” (7 journals) and “Business” (7 journals). “Information Science & Library Science”, “Political Science”, and “Sociology” are included with six journals. Nine categories contain only a single journal each.
According to the network analysis results (see Table 2), 24 journals belonging to “psychology” as a block were cited 2,260 times outside the psychology domain and 28 “communication” journals were cited 1,418 times by journals of other groups. The next most cited categories are “public, environmental & occupational health” (3 journals, 755 being cited) and “sociology” (6 journals, 555 being cited). Furthermore, the most preferred category by “communication” journals was “psychology” journals (565 citations). Conversely, “communication” journals were cited 487 times by “psychology” journals. This tells us that the relational strength between the two fields—“communication” and “psychology”—is the highest among all other subject pairs.