Authorship productivity in the knowledge management literature

Authors


Abstract

A Lotka analysis was conducted of 14,149 references to knowledge management publications retrieved from a search of the ABI/Inform database. Results indicate that the sample of knowledge management publications conforms to the negative power distribution typical of Lotka analyses. Calculated Lotka constants were within expected parameters. A Kolmogorov-Smirnov test revealed no significant differences between observed and expected distributions. It is concluded that the knowledge management literature conforms to the principles of Lotka's Law. Analysis of author affiliations revealed that academics were prevalent among the most productive authors but that the top five authors in terms of productivity were practitioners.

BACKGROUND AND LITERATURE REVIEW

Author Productivity

In 1926, Alfred J. Lotka published his pioneering article “The Frequency Distribution of Scientific Productivity,” in which he described a predictable pattern for the relative contributions of a body of authors to a body of literature. Lotka's basic proposition was that about 60 percent of authors who contribute to a given body of literature make only one contribution to that body of literature, and the pattern of contributions of more productive authors can be described by the equation

equation image

where yx is the number of authors making x contributions and n and c are two constants characteristic of the specific body of literature. Lotka examined selected data drawn from the decennial index to Chemical Abstracts for the years 1907 to 1916 and for Auerbach's Geschichtstafeln der Physik, which was intended as a bibliography of the best works in physics up to 1900. Lotka found that, although the distributions were slightly different for the two sources, the general pattern held.

Lotka's Law has subsequently been studied, updated, and modified by Naranan (1971), Schorr (1974), Vlachy (1976), Potter (1981), Pao (1985), Fang (1995), Huber (2002), Burrell (2008), Egghe (2010), and others. Studies of the applicability of Lotka's Law have been conducted in wide variety of subject areas, including history of technology (Murphy 1973), chemistry, physics, and mathematics (Allison and Stewart 1974), computer science (Radhakrishnan and Kernizan 1979), finance (Chung and Cox 1990), management information systems (Nathan Jackson 1991), accounting (Chung, Pak, and Cox 1992), dental science (Kawamura et al 1999), sport psychology (Baker, Robertson-Wilson, and Sedgwick 2003), financial risk (Chun-Hao and Jian-Min 2011), and sports economics (Santos and García 2011). Rowlands (2004) extended the application of Lotka's Law to the productivity of a multidisciplinary journal publisher.

Bibliometric Studies of the Knowledge Management Literature

There have been very few bibliometric or related studies of the knowledge management literature. Scandura and Williams (2000), Schultze and Leidner (2002), Schultze and Stabell (2004), Guo and Sheffield, and Wallace, Van Fleet, and Downs (2010) used content analysis to examine the core characteristics of the research literature of knowledge management. Ponzi (2003) explored the evolution and nature of knowledge management through citation and cocitation analysis. Wallace, Van Fleet, and Downs (2011) conducted a bibliometric analysis of the research literature of knowledge management and concluded that “the scholarly literature of knowledge management appears to be consistent with that of other fields that have been studied using Bradford analysis” (p. 19).

Serenko and Bontis (2004) conducted a “meta-review” of the knowledge management and intellectual capital literature in which they explored both research productivity and research impact of the authors of scholarly articles appearing in three knowledge management journals. Serenko and Bontis did not frame their work in terms of Lotka's Law, nor did they conduct a specific Lotka analysis. In a subsequent study, Serenko and Bontis (2009) developed a model for ranking knowledge management and intellectual capital academic journals. Identifying only scholarly journals was an explicit goal of the study, which focused on twenty journals. In an adjunct study, Bontis and Serenko (2009) explored alternative approaches to ranking academic journals.

Serenko et al (2010) conducted a bibliometric analysis of 2,175 knowledge management and intellectual capital articles published in eleven peer-reviewed journals. The variables explored in the study included national productivity, institutional productivity, author productivity, and varieties of research methods employed in the knowledge management and intellectual capital literature. One of the study's research questions was “Does the frequency of publication by authors in the KM/IC field follow Lotka's law?” (8). Serenko et al found that the distribution of articles among authors substantially conformed to Lotka's Law, with an n constant of 2.82.

RESEARCH QUESTION

This study continued earlier explorations of the research core of the knowledge management literature begun by Wallace, Van Fleet, and Downs (2010, 2011) by conducting a Lotka study of author productivity for a sample drawn from the knowledge management literature. The study was designed to address one very straightforward research question:

Does the literature of knowledge management conform to the pattern of author productivity defined by Lotka's Law?

METHODOLOGY

Practitioners vs. Scholars in the Knowledge Management Literature

A characteristic of the knowledge management literature that must be considered in conducting a bibliometric study of that literature is the balance of contributions from practitioners and scholars. As Serenko et al (2009) pointed out, “The overall field of KM/IC research in the early 1990s was supported primarily by practitioners” (4). Serenko et al's results indicate a decline in the contributions of practitioners from the mid-1990s to the late 2000s, but still found that practitioners accounted for 17 percent of all knowledge management and intellectual capital authors publishing in scholarly journals from 1994 to 2008 (10). Wallace's (2007) categorization of the knowledge management literature was that “Much of what has been published is exhortatory in nature, encouraging the adoption of knowledge management as an overall organizational philosophy or the incorporation into organizational operations of certain aspects of knowledge management such as the transfer of tacit knowledge or the development of content management systems” (221). In a departure from previous studies that focused on the research nature of the scholarly knowledge management literature, this study deliberately worked within the framework of a broad-based database that includes both research and non-research publications. The results can therefore be expected to be reflective of the broader range of the field of knowledge management, which is driven by both practitioner and researcher interests.

Sample Selection

Data for the study were drawn from a search of the ABI/Inform database. ABI/Inform is distributed by ProQuest and covers the business literature, very broadly defined, from 1971 to the present. In a departure from previous studies that focused on the research nature of the knowledge management literature, this study deliberately worked within the framework of a broad-based database that includes both research and non-research publications. The results can therefore be expected to be reflective of the broader range of the field of knowledge management, which is driven by both practitioner and researcher interests.

The beginning point for the selection of the sample was a search of ABI/Inform using the subject headings “Knowledge Management” and “Organizational Learning.” Although a keyword search would have yielded more results, the identification of the two precoordinate subject headings was anticipated to add some precision to the search while still yielding an acceptable volume of results. The initial search yielded 19,826 references to articles indexed with one or more of the two subject headings.

Sample Refinement

Anonymous sources were eliminated from the initial list of references, producing a final database of 14,149 references suitable for an author productivity analysis. The 14,149 publications represent a total of 405 authors.

RESULTS

Author Productivity

The 14,149 articles were tabulated in ascending order of author productivity to facilitate a Lotka analysis. The results are presented in Table 1. The mean number of contributions per author was 1.4; both the median and the mode were 1, confirming the general pattern of a preponderance of authors having made only one contribution to the literature of the field.

Power Curve Analysis

Lotka analysis typically yields a negative power curve that describes the inverse power relationship between the numbers of authors making specified contributions and the sizes of the contributions made by authors of varying productivity. The relationship between the number of contributions and the number of authors for the sample of knowledge management literature, with an imposed power curve, is shown in Figure 1. The imposed power curve corresponds nearly perfectly to the distribution observed for the knowledge management literature based on the sample drawn from the search of ABI/Inform.

Table 1. Author Frequency Distribution
x (Number of Contributions)y (Number of Authors)
111,334
21,685
3578
4244
5101
672
731
830
922
1011
119
124
136
144
154
162
182
202
211
221
231
312
341
381
621
Figure 1.

Observed and Power Curves Observed and Expected Values

The two constants for Lotka's Law, n and c can be used to calculate expected values for the number of authors associated with a specified level of authorship x. Lotka's original work found that the values of n were 2.021for the Chemical Abstracts Data and 1.888 for the Geschichtstafeln der Physik data. Lotka's values for c were .6079 and .5669. Vlachy (1974) suggested that the value of n should normally between 1 and 3 and that the value of c can be expected to rarely vary significantly from the values reported by Lotka. The calculated values of n and c for the knowledge management literature are 2.7273 and .5669, respectively. Using these constants, the expected values of y are shown, with the observed values, in Table 2.

Table 2. Author Frequency Distribution, Observed And Expected Values
x (Number of Contributions)y (Number of Authors) Observedy (Number of Authors) Expected
111,33411,319
21,6851,709
3578566
4244258
5101140
67285
73156
83039
92228
101121
11916
12413
13610
1448
1547
1626
1824
2023
2113
2212
2312
3121
3411
3811
6210

Kolmogorov-Smirnov Analysis

Pao (1985) described and discussed the use of the Kolmogorov-Smirnov test as a means for assessing the goodness-of-fit of an observed Lotka distribution in comparison to the expected Lotka distribution. The Kolmogorov-Smirnov test is used to is used to test whether the maximum absolute difference in the overall distribution of two independent samples is significant. A Kolmogorov-Smirnov analysis conducted to compare the distributions of the observed and expected values of y for the knowledge management literature indicated no significant difference in the two distributions (p < .000). The observed distribution is graphed against the expected distribution in Figure 2. The two distributions are so closely correlated that there appears to be no distinguish between the graphed lines of the distributions. This visual representation provides further support for the results of the Kolmogorov-Smirnov test.

Academics vs. Practitioners as Authors

This study was deliberately designed to include both research and non-research publications, in keeping with the principle that knowledge management is of interest to both practitioners and academics and that both practitioners and academics contribute to the knowledge management literature. The balance of academics and practitioners as authors of the publications in the database was examined by determining the affiliations of all authors who had contributed ten or more publications to the database. The distribution of articles among these authors is shown in Table 3. The fifty-two authors who contributed ten or more publications each (14 percent of all authors) were responsible for a total of 821 publications, 17 percent of all publications in the database. The mean number of publications per author in this group was 15.79.

Table 3. Authors Contributing Ten or More Publications
Number of ContributionsNumber of Authors
621
381
341
312
231
221
211
202
182
162
154
144
136
124
119
1011

Of the fifty-two authors who each made ten or more contributions, thirty-four were academics (65.4 percent) and eighteen were practitioners (34.6 percent). Although academics substantially outweighed practitioners as contributing authors, the five most productive authors were all practitioners.

Figure 2.

Observed and Expected Distributions

DISCUSSION AND CONCLUSIONS

The results of this Lotka analysis suggest that the literature of knowledge management is in perfect conformity with the general expectations for author productivity revealed by past Lotka studies. The values for the two constants n and c are within anticipated parameters. The observed distribution based on results from a search of the ABI/Inform database related perfectly to the expected distribution generated from application of the constants.

Most Lotka analyses have been based on scholarly or research productivity. This study of a sample of publications known to include both scholarly and non-research publications may be suggestive of broader applicability of Lotka's Law outside the realm of scholarly productivity. It is noteworthy that, although a substantial majority of the authors who had contributed ten or more publications were academics, the top five authors were practitioners. More than a third of the publications by authors responsible for ten or more contributions were authored by practitioners. The role of the practitioner in the literature of knowledge management continues to be an important one, at least based on the sample used in this study.

Extension of the methodology employed for this study may be useful for exploring author productivity in an expanded realm of publications, particularly in domains in which practitioners can be expected to play a substantial role as contributors to the literature. This is true in general of the information and knowledge domains and may have substantial relevance in other areas as well.

Ancillary