Acknowledgments Networks in Accounting Scholarship: A Note

Table of Contents

1. Introduction

2. Networks in Accounting Research: Related Literature

3. Sample and Research Design

4. Acknowledgment Networks

5. Conclusions

6. References

1 Introduction

Accounting knowledge is socially constructed. It is constructed in response to the society’s needs to design economic activity so as to foster prosperity and, specifically, in response to necessity for accounting operations in the production process and the urgency of standard-setting regulations for accounting practices. Such needs shape – among other things – the production of accounting knowledge in academia; they affect the design of universities and educational programs, the interaction between accounting academia and the accounting profession, the allocation of resources within universities, the design and the assessment of novel ideas in accounting scholarship (e.g., Almer, Bertolini, & Higgs, 2015; Locke & Lowe, 2008; Mintz, Dang, & Savage, 2013; Rowlinson, Harvey, Kelly, Morris, & Todeva, 2015). Accounting scholars reproduce (or sometimes change) the structural features of the accounting academia and the profession as well. Their responses are related to their personal values, priorities, scientific profile, and professional concerns (Malsch & Tessier,2015). These responses include the accounting researcher’s choice on the topic, the theory, the methodology, the quantity, the diversity, and the publication outlets of his research output. These choices are often articulated in a collaborative manner. Cooperation is needed to produce increasingly numerous and diverse research products in a setting of fierce competition for influential publication space and university employment.

Exploring the cross section of partnership in accounting research, we focus on acknowledgments. Acknowledgments are collaborations which are essential in getting a paper published by helping the authors shape convincing arguments, identify their contribution to the literature, and even correct minor linguistic errors (e.g., Dalton et al., 2016). Our work is motivated by the importance of acknowledgments in the production and dissemination of novel academic output in accounting, findings on the social structure of acknowledgments in other disciplines (e.g., Andrikopoulos & Economou, 2016), as well as existing evidence that coauthorship and citations in accounting exhibit social network properties. In this framework, we seek to answer the following research questions: “What is the social structure of acknowledgment partnerships in accounting research?” and “Who are the most central nodes in the acknowledgment network?”.

Working with a hand collected data set that spans a 20 year period of intellectual partnership evidenced in coauthorship and acknowledgments in six accounting journals, we employ social network analysis in order to map acknowledgment patterns, since intellectual partnership between accounting academics is known to span social networks: such networks can involve coauthorship, citation-based communication, PhD examination committees (Andrikopoulos, Bekiaris, & Kostaris, 2020; Andrikopoulos & Kostaris, 2017; Bonner, Hesford, Van Der Stede, & Young, 2012; Casanueva & Larinaga, 2012; Wakefield, 2008).

Our contribution is threefold: i) we unveil cross sectional and time-series acknowledgment collaborations, ii) we describe the structure of the acknowledgment network in accounting research, and iii) we discover the central nodes in the acknowledgment network. Our findings a) suggest that acknowledgments have been increasing during the last two decades of research in leading accounting journals, b) reveal that the acknowledgment network has small-world properties, being highly clustered and characterized by a giant component which extends across the biggest part of the network, and c) present the most central accounting scholars who – via acknowledgments – help improve and disseminate accounting research.

Section 2 presents prior research in social networks in accounting scholarship. Section 3 describes the methodology and the sample. Section 4 presents and discusses acknowledgment networks in published accounting research. Section 5 concludes the paper.

2 Networks in Accounting Research: Related Literature

Wakefield (2008) produced a ranking of accounting research journals based on citation networks. She identified research journals based on a journal’s stated objectives, the publication of research articles, and the number of authors and editors who were academics. Exploring citations to papers which appeared in a sample of 22 accounting research journals from 2000 to 2006, she assessed a journal’s influence with both the quantity of citations and also the quality of the journals in which these citations appeared. Wakefield also identified groups of journals which were connected through citation networks, documenting, e.g., that the Accounting Review, the Journal of Accounting Research, and the Journal of Accounting and Economics were closely connected through citations.

Citations as media of communication between accounting scholars were also the topic of Bonner et al. (2012). They explored a variety of communication structures based on the theory, the topic, and the methodology of published research. Their sample spanned 25 years (1984–2008) and five journals: The Accounting Review, Journal of Accounting Research, Journal of Accounting and Economics, Accounting Organizations and Society, and Contemporary Accounting Research. Two types of communication structures were observed: normal academic field (multiple disconnected clusters of topical origin) and small-world, with tribalism being present mainly in the early periods of the sample (tribes are shaped on joint influences of topic, theory, and method). They also found the accounting scholars who serve as hubs in the citations network, connecting disparate streams of accounting research.

Casanueva and Larrinaga (2013) extended the discussion on the social structure of accounting academia into gatekeeping procedures such as the examination of PhD theses in accounting. Drawing on sociological arguments about the role of the academic elite in the generation and diffusion of new ideas, they investigated the community of accounting scholars in Spain in order to assess the validity of the “invisible college hypothesis”: “disciplines are characterized by a core and inter-institutional group of highly productive and high-profile scientists that interact, formally and informally, with each other in the field and generate a disproportionate volume of new ideas, including its rules and certain research problems” (Casanueva & Larrinaga, 2013, p. 20). Measuring academic centrality with membership in examination boards of PhD theses, they found that the invisible college hypothesis does not hold in the Spanish community of accounting: there is an oligarchy, but it is not highly productive scientifically.

Andrikopoulos and Kostaris (2017) are probably the closest to our study in terms of research agenda, sample, and methodology. They explored collaboration networks in accounting, focusing on coauthorship partnerships between scholars, affiliations, and countries. Their sample included all papers which were published from 1985 to 2014, in the Accounting Review, the Journal of Accounting Research, Journal of Accounting and Economics, Accounting Organizations and Society. They discovered that coauthorship in accounting is articulated in a small-world of collaborators: this network is highly clustered, is largely covered by a single component and, within the component, there are short distances between accounting researchers and institutions. Extending this study, we investigate the network of intellectual collaboration in accounting research by incorporating collaborators whose contribution was not equal to the author’s, but it was sufficient for them to be included in the acknowledgments; in scientometrics, they are called subauthors.

Overall, prior research has explored social networks in intellectual communication in published account research, mostly in terms of coauthorship and citations. Intellectual partnership, however, is not limited in coauthorship. Scholars that are mentioned in acknowledgments often offer both technical and theoretical aid in the preparation of the published manuscript and, in this regard, their contribution to the production and dissemination of knowledge can be essential (Glänzel & Schubert, 2004; Heffner, 1981). Therefore, the analysis of social networks in published accounting research will help us construct a richer and more refined depiction collaboration patterns in accounting research.

3 Sample and Research Design

Our methodology is social network analysis. A social network is a collection of actors and connections between these actors. A social network can be illustrated with a graph, in which actors are depicted with nodes and connections with edges between the nodes. A graph can be directed or undirected. A coauthorship link has no direction: if A has written with B, then B has also written with A. The subauthorship link, however, is directional: A may acknowledge B’s contribution in her paper, but B may not acknowledge A’s contribution in one of B’s papers; directional connections are depicted with arrows (if A acknowledges B’s help, then an arrow runs from A to B). The number of edges which start or end with a particular node constitute the node’s degree. In a directed subauthorship network the number of edges stemming from a node are the node’s outdegree whereas the number of edges which are directed towards a node are the node’s indegree. Two nodes are considered connected if there is a sequence of nodes that connects them. Such sequence is called a path and the length of the path is the number of the path’s edges; the length of the shortest path between two nodes is called the distance between them. A group of connected nodes form a component within the network. The biggest such group is the network’s giant component.

As an example, Figure 1 presents the network of John Harry Evans III, who is the actor with the highest indegree in the last period of our sample, from 2011 to 2015. This means that he received more acknowledgments than any other accounting scholar during this period.

Figure 1:

The network of John Harry Evans III.

Our sample spans the period from 1996 to 2015 and includes all papers which were published in the Accounting Review (TAR), the Journal of Accounting Research (JAR), the Journal of Accounting and Economics (JAE), Accounting Organizations and Society (AOS), Contemporary Accounting Research (CAR), and the Review of Accounting Studies (RAS). We chose these journals because they all received the highest rating (4 or 4*) in the Academic Journal Guide of the Chartered Association of Business Schools in 2018, thereby giving us the opportunity to study intellectual partnerships in published research that may be perceived as “leading”. While we acknowledge the validity of this journal quality ranking has limitations in the case of accounting scholarship (Rowlinson et al., 2013), the choice of these journals facilitates comparisons with prior scientometric research in accounting; these journals, except for RAS, were explored in previous studies of social networks in accounting research (Andrikopoulos & Kostaris, 2017; Bonner et al., 2012). Our sample period starts in 1996 since this was the year when the youngest journal in our sample was founded (RAS). We included original research articles, excluding items like book reviews and editorials. We ended up with 3950 papers and 9922 authors and subauthors. The data set was hand collected and, in this sense, it is unique. The network’s nodes sample contains both authors and subauthors (acknowledged individuals) since our network is a directed one (acknowledgments are directed from authors to subathors). E.g., in a paper with three authors and six subauthors, there are 18 directed connections in the network, six directed edges from each of the authors to each of the subauthors. In our sample, the roles of author and subauthor are interchangeable, since most authors are subauthors as well: 96.68% of authors are also subauthors (and 21.66% of subauthors are authors as well). Tables 1a and 1b presents an overview of our sample which is split in four five year periods. The number of published articles has increased by 76% across the five year periods of our sample, largely due to increases in published output in TAR, CAR, and RAS. Published accounting research is becoming increasingly collaborative, with single-author papers falling from 31.08% of the sample in 1996–2000 to 16.57% in 2011–2015 (AOS is the journal which publishes the least amount of collaborative research: 35.68% of published research in AOS has been written by a single-author). The average paper was authored by 1.984 writers in 1996–2000 and 2493 writers in 2011–2015. Intellectual partnership has been increasing in terms of subauthorship as well; the average number of subauthors has increased from 8.22 in 1996–2000 to 11.43 in 2011–2015, with JAE exhibiting the largest number of subauthors and AOS the lowest. The increase in collaborative research can be related to the length of published research; research endeavors of greater length may be of greater scope thus necessitating a larger number of collaborators (e.g., Laband & Tollison, 2000). Indeed, the length of the average paper has increased from 23.47 to 28.52 pages (the shortest papers were published in AOS in all sub-periods of our sample period). Finally, our results corroborate previous findings in the literature about the dominating presence of USA-based affiliations in published research (e.g., Raffournier & Schatt, 2010): the percentage of papers which were authored or coauthored by USA-based scholars were 69.9% in 1996–2000 and 58.1% in 2011–2015, with JAE being the journal which published the largest number of papers which were written or cowritten outside the USA.

4 Acknowledgment Networks

Figures 2 and 3 show the distribution of degree centrality in the acknowledgment network. Out-degrees have similar distribution as in-degrees. The majority of authors (8402) have not thanked more than 10 people in their papers whereas the majority of subauthors (8798) have not been thanked by more than 10 people. Table 2 demonstrates average degrees for all journals and periods. Authors and subauthors in JAE and TAR are more connected than the ones in other journals, whereas AOS has the least connected nodes in its network. These qualitative characteristics of connectedness are also evident in the number of cliques in these networks. Cliques are sets of interconnected nodes. Since any pair of authors or subauthors is connected, we defined cliques as three nodes who are interconnected through links of subauthorship at least twice; these are the two-level cliques. As implied by degree distribution, Table 2 presents evidence that the number of two-level cliques is the highest in JAE and TAR and it is the lowest in AOS. Table 3 presents the 20 most central nodes, according to four measures of centrality. These four top-20 lists are populated by 45 individuals which means that there is substantial overlap across the centrality measures; 44 out of 45 of them have served in the editorial board of at least one of the journals of the dataset. The 20 most prolific authors in the dataset have also served in the editorial boards of sample’s journals (Table A1 in the Appendix). We observe, therefore, an overlap between accounting scholars who run leading journals, publish their work in the same journals and get thanked for assisting the journals’ authors.

Figure 2:

Indegree distribution.

Figure 3:

Outdegree distribution.

The subauthorship network exhibits small-world characteristics (small-world evidence has also been found in coauthorship networks between accounting scholars [Andrikopoulos & Kostaris, 2017]). A typical property of small-world networks is that the giant component covers a large part of the network; in our case it is 95.96%, while it is steadily above 90% in all sub-periods of our sample. Another major characteristic of small-world networks is the small average degree, compared to the size of the network (while a network may be large, the average node is directly connected to only a small number of other nodes). Average degree is only 6.704 out of 9.921 available connections for every node in (stand-alone journal networks, TAR exhibits the highest average degree, 4.919). Moreover, there are small average distances between the nodes of the giant component in a small-world network; such distances are considered are considered small when the ratio of the distance over the number of nodes in the giant component approaches one as the number of nodes approach infinity. In the subauthorship network, the average distance is steadily smaller than one as the number of nodes increases, so we can take it to be sufficiently small (the small-world findings are similar if we take each journal’s network separately (Table A1 in the Appendix)).

Finally, small-world networks exhibit high clustering coefficients. The clustering coefficient of a node k is the ratio of the number of connections (edges) among k’s connected nodes over the maximum number of connections (edges) among them (this can be, e.g., the fraction of a researcher’s subauthors, who are themselves connected with subauthorship collaborations). The average clustering coefficient is considered to be high if it is a lot greater than the probability of tie formation in the network (which is the ratio of the average degree over the number of nodes in the network). In our case the average clustering coefficient is almost 400 times greater than the probability of tie formation (Table 4). The nodes in small-world networks communicate fast which, in our case, means that novel ideas in accounting scholarship flow efficiently across accounting scholars (after all, subauthorship is about just that: the circulation of novel works of science). On the other hand, a tightly connected network of scholars may be so solid that it may not accommodate the influx of radical innovation, spawned by heterodox, disconnected nodes whose disconnectedness might be a symptom of intellectual freshness (such evidence could help explain findings on the normatively isomorphic status of accounting research [Tuttle & Dillard, 2007]).

Since most of the most central nodes in our sample have served as editorial board members, our results may, in part, reflect standard practice to thank the editor for their help in the acknowledgments. To explore the effect of editorial board membership in our network, we conducted an experiment with TAR, which is the journal with the highest number of two-level cliques in our sample. We removed all editorial members from the network, both as authors and as subauthors. Editorial board membership was assessed on yearly basis. Therefore, if someone were a member of the board in 2000 but not in 1999, she was removed from authorship and subauthorship connections for 2000, but not for 1999. About 3.85% of published output in TAR has been written solely by editorial board members, whereas 29.72% of published output has been produced by authorship teams that include at least one board member. We ran two social network analyses for TAR on with and one without board members (Table 5 presents some of our findings). The effect of editorial board membership on network structure is evident in our results. While the size of the giant component as percentage of network size exhibits only a small change, the removal of board members substantially reduces average degree centrality, network density and the average clustering coefficient, while it lengthens the average distance between nodes in the giant component.

5 Conclusions

Exploring the relational structure of intellectual partnership in accounting research, we studied the social network of acknowledgments in accounting scholarly journals. Our sample included 20 years of publications in six accounting journals. Our findings suggest that most authors also act as subauthors, the average paper has probably been cowritten in the USA and it has been increasing in length, in the number of authors and subauthors. We found that the acknowledgments form a network of intellectual partnership which has small-world properties, being highly clustered and covered – at its biggest part – by a giant component of connected accounting scholars. A densely connected network in which almost all nodes are directly or indirectly connected can be a good thing for the fast flow of ideas across accounting academics. However, a densely connected network could be evidence of intellectual homogeneity across network members, and, in this regard, it may be associated with limited openness to heterodox scholars and radical innovations in accounting thought. Openness to a broader range of approaches to accounting theorizing and empirical research is important for improving the quality of accounting research (Zeff, 2019).

Future research in this area could explore the identity of the network’s nodes to account for the structure of acknowledgment patterns. Academic affiliation (current and previous), involvement in the accounting profession, editorial board membership, PhD affiliation, gender, and age can help highlight the path that connects authors and their colleagues in acknowledgments. Moreover, one can extend acknowledgment networks to include institutional acknowledgments, apart from individual ones: grants, scholarships, conferences, and professional associations can also help explore the flow of ideas, texts, and resources among accounting scholars.