Theoretical studies of the historical development of the accounting discipline: A review and evidence

Abstract

Many existing studies of the development of accounting thought have either been atheoretical or have adopted Kuhn's model of scientific growth. The limitations of this 35-year-old model are discussed. Four different general neo-Kuhnian models of scholarly knowledge development are reviewed and compared with reference to an analytical matrix. The models are found to be mutually consistent, with each focusing on a different aspect of development. A composite model is proposed. Based on a hand-crafted database, author co-citation analysis is used to map empirically the entire literature structure of the accounting discipline during two consecutive time periods, 1972–81 and 1982–90. The changing structure of the accounting literature is interpreted using the proposed composite model of scholarly knowledge development.

Keywords:

• accounting theory
• author co-citation analysis
• history of accounting thought
• scholarly knowledge development
• theory closure
• theory groups

Acknowledgments

This paper has benefitted greatly from the helpful comments of Professor Bill McInnes, the late Professor Segun Wallace and two anonymous reviewers. Thanks to Gillian MacIver and Jacky Pierpoint for assistance with the empirical mappings.

Notes

1. An understanding of the development of scholarly accounting knowledge can also benefit non-accounting scholars. Accounting represents an important case study for those working in the field of scholarly knowledge development generally, since it possesses certain distinguishing characteristics whose impact has not been considered widely. In particular, accounting is an applied social science whose theory is embedded in other disciplines, such as economics, psychology, and sociology. In addition, the practice of accounting has professional status, and since the profession is largely self-regulated, research can fulfil a strategic role by representing an externally validated form of knowledge which may be used in the regulatory process. The accounting discipline therefore offers the opportunity to explore its relationships with related disciplines and its associated profession.

2. The three earlier approaches were the traditional philosophy of science (which emphasised internal cognitive factors); the sociology of knowledge (which emphasised external factors); and the sociology of science (which emphasised internal social factors).

3. Fogarty (2004: p. 37) observes that this domination is also reflected in the organisation of the American Accounting Association.

4. Other writers (e.g. Merino, 1993) have focused upon the development of practical, rather than academic, accounting knowledge.

5. Donaldson also argues for distinct stages in the ‘life’ of an idea. He likens major innovations to ‘intellectual waves’ and likens most academics to ‘surfers’ who ‘mount and ride the crest until the wave loses its energy and breaks as a gentle ripple on the shore of established thought’ (1977: p. 9). However, he notes that we are lucky if each decade has one or two ‘waves’. An interesting empirical analysis would be to track the progress of specific waves. This would be complementary to the present paper's analysis, which reviews the entire ‘sea’.

6. Kuhn's view was that those who shared a disciplinary matrix shared the same ‘way of seeing’.

7. A better description of this model might be the ‘environmental determinism model’, however the general usage term is employed here.

8. Murphy (1988) does acknowledge that Bourdieu's (1981) earlier analysis of scientific communities is wholly consistent with closure theory.

9. Citation analysis can be used for purposes other than the study of literature structure and research networks. For example, Brown (1996) identifies influential individuals and articles using citations from only five, predominantly North American, journals. However, Shockley and Stratton (1995) are critical of using citations to evaluate individual faculty members for two main reasons. First, citations do not necessarily reflect ‘value or quality’, due to negative citations, self-citations and ‘halo’ effects. Second, the usefulness of a citation index based on a specific source journal set depends on its inclusiveness and representativeness.

10. In accounting, document co-citation analysis has been used to reveal the structure of individual research areas (Gamble & O'Doherty (1985) examine income smoothing and Gamble et al. (1987) examine agency theory).

11. The journal selections were influenced by comments from two anonymous reviewers and the editor.

12. The joint editors of the journal (T. Tinker and D. Cooper) are based in the US and Canada, respectively. D. Cooper originates from the UK.

13. The four journals in continued existence throughout the period represented all three key countries: Accounting Review, Journal of Accounting Research, Accounting and Business Research and Abacus.

14. For example, ‘Beaver W’ and ‘Beaver WH’ initially showed up as two distinct authors and had to be combined in the database.

15. Authors were excluded in period 1972–81 if either the number of zero co-citation counts exceeded 53 or the mean co-citation count was less than 6; authors were excluded in period 1982–90 if either the number of zero co-citation counts exceeded 15 or the mean co-citation count was less than 10. This difference in cut-off levels reflects the generally larger size of the database in the second period.

16. The ‘elbow test’ is a common heuristic used to determine the dimensionality of the data. Stress values are plotted against number of dimensions and joined up. If a noticeable ‘elbow’ occurs, this indicates that fit is not significantly improved by the addition of a further dimension. It is also recommended that stress values of 0.05 or less indicate a good fit (Kruskal & Wish, 1978: p. 56). In both time periods, the elbow occurred at two dimensions and the stress value was less than 0.05 (0.037 and 0.045 respectively).

17. In agglomerative hierarchical clustering, each author starts as a cluster of one, and at each stage in the agglomeration process the closest two clusters join until all authors are in a single cluster. This process can be visually represented on a dendrogram, with large distances between stages indicating a high level of dissimilarity between the cluster pairs (Aldenderfer & Blashfield, 1984: pp. 53–57).

18. It is interesting to speculate whether a map covering the previous ten years would locate these normative accounting theorists in a central position.

19. Green, who appears at the bottom of the generalist cluster (2b), was the founder editor of the Journal of Accounting Research, and authored several general papers on empirical research.

20. Although these three authors do form a cluster, it is so loosely connected and unfocussed that the boundary has not been shown.